ResNet-VGG-Inception-4Piece-PCA-Model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "fd1efdf1",
   "metadata": {},
   "source": [
    "### Importing Necessary Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "id": "0938de48",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pickle\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "from sklearn.decomposition import PCA\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.preprocessing.image import load_img, img_to_array\n",
    "\n",
    "from tensorflow.keras.layers import Input, Flatten, Dense, Dropout, GlobalAvgPool2D, BatchNormalization\n",
    "# from tensorflow.keras.applications import EfficientNetV2L\n",
    "from tensorflow.keras.regularizers import l2\n",
    "\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "\n",
    "from sklearn.metrics import precision_recall_fscore_support as score"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c954b162",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "68302673",
   "metadata": {},
   "source": [
    "A detailed discussion and visualization of the data can be seen in [here](Data-Overview.ipynb)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "id": "75b06a70",
   "metadata": {},
   "outputs": [],
   "source": [
    "labels = pd.read_csv(\"./Data/train_labels.csv\")\n",
    "images = ['./Data/train/train/'+str(name)+'.tif' for name in labels['id']]\n",
    "labels['images'] = images\n",
    "labels = labels[['id', 'images', 'label']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "id": "fd2fe082",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>images</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>./Data/train/train/0.tif</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>./Data/train/train/1.tif</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>./Data/train/train/2.tif</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>./Data/train/train/3.tif</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>./Data/train/train/4.tif</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15995</th>\n",
       "      <td>15995</td>\n",
       "      <td>./Data/train/train/15995.tif</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15996</th>\n",
       "      <td>15996</td>\n",
       "      <td>./Data/train/train/15996.tif</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15997</th>\n",
       "      <td>15997</td>\n",
       "      <td>./Data/train/train/15997.tif</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15998</th>\n",
       "      <td>15998</td>\n",
       "      <td>./Data/train/train/15998.tif</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15999</th>\n",
       "      <td>15999</td>\n",
       "      <td>./Data/train/train/15999.tif</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>16000 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          id                        images  label\n",
       "0          0      ./Data/train/train/0.tif      1\n",
       "1          1      ./Data/train/train/1.tif     13\n",
       "2          2      ./Data/train/train/2.tif     13\n",
       "3          3      ./Data/train/train/3.tif     14\n",
       "4          4      ./Data/train/train/4.tif      6\n",
       "...      ...                           ...    ...\n",
       "15995  15995  ./Data/train/train/15995.tif      2\n",
       "15996  15996  ./Data/train/train/15996.tif     15\n",
       "15997  15997  ./Data/train/train/15997.tif      3\n",
       "15998  15998  ./Data/train/train/15998.tif      9\n",
       "15999  15999  ./Data/train/train/15999.tif      9\n",
       "\n",
       "[16000 rows x 3 columns]"
      ]
     },
     "execution_count": 222,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "id": "276e2059",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_labels = pd.read_csv('./Data/train_labels_1.csv')\n",
    "test_labels = pd.read_csv('./Data/test_labels_1.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "id": "2aea9ea9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>images</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>./Data/train/train/0.tif</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>./Data/train/train/1.tif</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>./Data/train/train/2.tif</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>./Data/train/train/3.tif</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>./Data/train/train/4.tif</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15195</th>\n",
       "      <td>15995</td>\n",
       "      <td>./Data/train/train/15995.tif</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15196</th>\n",
       "      <td>15996</td>\n",
       "      <td>./Data/train/train/15996.tif</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15197</th>\n",
       "      <td>15997</td>\n",
       "      <td>./Data/train/train/15997.tif</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15198</th>\n",
       "      <td>15998</td>\n",
       "      <td>./Data/train/train/15998.tif</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15199</th>\n",
       "      <td>15999</td>\n",
       "      <td>./Data/train/train/15999.tif</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>15200 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          id                        images  label\n",
       "0          0      ./Data/train/train/0.tif      1\n",
       "1          1      ./Data/train/train/1.tif     13\n",
       "2          2      ./Data/train/train/2.tif     13\n",
       "3          3      ./Data/train/train/3.tif     14\n",
       "4          4      ./Data/train/train/4.tif      6\n",
       "...      ...                           ...    ...\n",
       "15195  15995  ./Data/train/train/15995.tif      2\n",
       "15196  15996  ./Data/train/train/15996.tif     15\n",
       "15197  15997  ./Data/train/train/15997.tif      3\n",
       "15198  15998  ./Data/train/train/15998.tif      9\n",
       "15199  15999  ./Data/train/train/15999.tif      9\n",
       "\n",
       "[15200 rows x 3 columns]"
      ]
     },
     "execution_count": 224,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "id": "b9ea19c9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>images</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18</td>\n",
       "      <td>./Data/train/train/18.tif</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>25</td>\n",
       "      <td>./Data/train/train/25.tif</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>33</td>\n",
       "      <td>./Data/train/train/33.tif</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>41</td>\n",
       "      <td>./Data/train/train/41.tif</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>65</td>\n",
       "      <td>./Data/train/train/65.tif</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>795</th>\n",
       "      <td>15816</td>\n",
       "      <td>./Data/train/train/15816.tif</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>796</th>\n",
       "      <td>15824</td>\n",
       "      <td>./Data/train/train/15824.tif</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>797</th>\n",
       "      <td>15832</td>\n",
       "      <td>./Data/train/train/15832.tif</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>798</th>\n",
       "      <td>15863</td>\n",
       "      <td>./Data/train/train/15863.tif</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>799</th>\n",
       "      <td>15994</td>\n",
       "      <td>./Data/train/train/15994.tif</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>800 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        id                        images  label\n",
       "0       18     ./Data/train/train/18.tif     11\n",
       "1       25     ./Data/train/train/25.tif      8\n",
       "2       33     ./Data/train/train/33.tif      6\n",
       "3       41     ./Data/train/train/41.tif      1\n",
       "4       65     ./Data/train/train/65.tif     14\n",
       "..     ...                           ...    ...\n",
       "795  15816  ./Data/train/train/15816.tif      2\n",
       "796  15824  ./Data/train/train/15824.tif      2\n",
       "797  15832  ./Data/train/train/15832.tif     11\n",
       "798  15863  ./Data/train/train/15863.tif     15\n",
       "799  15994  ./Data/train/train/15994.tif      6\n",
       "\n",
       "[800 rows x 3 columns]"
      ]
     },
     "execution_count": 225,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 226,
   "id": "29190fd4",
   "metadata": {},
   "outputs": [],
   "source": [
    "class_labels = list(set(labels['label']))\n",
    "class_names = [\n",
    "    'letter', 'form', 'email', 'handwritten', 'advertisement', 'scientific report', 'scientific publication',\n",
    "    'specification', 'file folder', 'news article', 'budget', 'invoice', 'presentation', 'questionnaire', 'resume',\n",
    "    'memo'\n",
    "]\n",
    "label_names = pd.DataFrame({\n",
    "    'labels': class_labels,\n",
    "    'names': class_names\n",
    "})"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c2e251c1",
   "metadata": {},
   "source": [
    "## Data Preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85b2ccd1",
   "metadata": {},
   "source": [
    "We have already generated the Extracted Feature Vectors from the ResNet Model now we train using that, so we load those feature vectors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "id": "37a36e2b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16000, 5, 2048)"
      ]
     },
     "execution_count": 227,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_resnet = np.load('./Data/ResNet-4P-train-precompute/whole_train_data_precomp.npy')\n",
    "fine_tune_data_resnet.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "id": "6da49e51",
   "metadata": {},
   "outputs": [],
   "source": [
    "N_COMPS_1 = 1300 # 1600\n",
    "N_COMPS_2 = 300 # 400\n",
    "N_COMPS_3 = 900 # 1000\n",
    "random_state=43"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "id": "72186ebc",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "pcas_resnet = [PCA(N_COMPS_1, random_state=random_state) for _ in range(5)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "id": "a3ed948b",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(5):\n",
    "    pcas_resnet[i].fit(fine_tune_data_resnet[:,i,:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "id": "714f49b5",
   "metadata": {},
   "outputs": [],
   "source": [
    "fine_tune_data_train_red_resnet = np.zeros((fine_tune_data_resnet.shape[0], 5, N_COMPS_1))\n",
    "for i in range(5):\n",
    "    fine_tune_data_train_red_resnet[:,i,:] = pcas_resnet[i].transform(fine_tune_data_resnet[:,i,:])\n",
    "fine_tune_data_resnet = fine_tune_data_train_red_resnet\n",
    "del fine_tune_data_train_red_resnet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "id": "f01feda5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16000, 5, 1300)"
      ]
     },
     "execution_count": 232,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_resnet.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "id": "4cc9fe34",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16000, 5, 512)"
      ]
     },
     "execution_count": 233,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_vgg = np.load('./Data/VGG16-4P-train-precompute/whole_train_data_precomp.npy')\n",
    "fine_tune_data_vgg.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "id": "5346129b",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "pcas_vgg = [PCA(N_COMPS_2, random_state=random_state) for _ in range(5)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "id": "9610e5c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(5):\n",
    "    pcas_vgg[i].fit(fine_tune_data_vgg[:,i,:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "id": "c230ea1e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16000, 5, 300)"
      ]
     },
     "execution_count": 236,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_train_red_vgg = np.zeros((fine_tune_data_vgg.shape[0], 5, N_COMPS_2))\n",
    "for i in range(5):\n",
    "    fine_tune_data_train_red_vgg[:,i,:] = pcas_vgg[i].transform(fine_tune_data_vgg[:,i,:])\n",
    "fine_tune_data_vgg = fine_tune_data_train_red_vgg\n",
    "del fine_tune_data_train_red_vgg\n",
    "fine_tune_data_vgg.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "id": "40e99544",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16000, 5, 1536)"
      ]
     },
     "execution_count": 237,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_inception_resnet = np.load('./Data/Inception-ResNet-4P-train-precompute/whole_train_data_precomp.npy')\n",
    "fine_tune_data_inception_resnet.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "id": "2140d86a",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "pcas_inception_resnet = [PCA(N_COMPS_3, random_state=random_state) for _ in range(5)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "id": "c0b483bd",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(5):\n",
    "    pcas_inception_resnet[i].fit(fine_tune_data_inception_resnet[:,i,:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "id": "eca6a59d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16000, 5, 900)"
      ]
     },
     "execution_count": 240,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_train_red_inception_resnet = np.zeros((fine_tune_data_inception_resnet.shape[0], 5, N_COMPS_3))\n",
    "for i in range(5):\n",
    "    fine_tune_data_train_red_inception_resnet[:,i,:] = pcas_inception_resnet[i].transform(fine_tune_data_inception_resnet[:,i,:])\n",
    "fine_tune_data_inception_resnet = fine_tune_data_train_red_inception_resnet\n",
    "del fine_tune_data_train_red_inception_resnet\n",
    "fine_tune_data_inception_resnet.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "id": "6e4e8b18",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16000, 5, 2500)"
      ]
     },
     "execution_count": 241,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data = np.concatenate([fine_tune_data_resnet, fine_tune_data_vgg, fine_tune_data_inception_resnet], axis=-1)\n",
    "fine_tune_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "id": "568ffcb1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16000,)"
      ]
     },
     "execution_count": 242,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_labels = labels['label'].to_numpy()\n",
    "fine_tune_labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "id": "937e1e86",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(15200, 5, 2500)"
      ]
     },
     "execution_count": 243,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_train = fine_tune_data[list(train_labels['id'])]\n",
    "fine_tune_data_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "id": "af08c830",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(15200,)"
      ]
     },
     "execution_count": 244,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_labels_train = train_labels['label'].to_numpy()\n",
    "fine_tune_labels_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "id": "c00f0b32",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(800, 5, 2500)"
      ]
     },
     "execution_count": 245,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_test = fine_tune_data[list(test_labels['id'])]\n",
    "fine_tune_data_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "id": "16f4bb93",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(800,)"
      ]
     },
     "execution_count": 246,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_labels_test = test_labels['label'].to_numpy()\n",
    "fine_tune_labels_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dff1f776",
   "metadata": {},
   "source": [
    "## Fine-Tuning Models"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7380947f",
   "metadata": {},
   "source": [
    "The fine-tuning part of our model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "id": "96a2bdfe",
   "metadata": {},
   "outputs": [],
   "source": [
    "def ftmodel():\n",
    "    inputs = Input(shape=(5, N_COMPS_1 + N_COMPS_2 + N_COMPS_3))\n",
    "    x = Flatten()(inputs)\n",
    "    x = BatchNormalization()(x)\n",
    "\n",
    "    top_dropout_rate = 0.5\n",
    "    # x = Dropout(top_dropout_rate, name='top_dropout_1')(x)\n",
    "    # x = Dense(1024, name='top_dense_1', kernel_initializer=\"he_normal\", kernel_regularizer=l2(5e-5), activation='relu')(x)\n",
    "    # x = BatchNormalization()(x)\n",
    "    x = Dropout(top_dropout_rate, name='top_dropout_2')(x)\n",
    "    x = Dense(512, name='top_dense_2', kernel_initializer=\"he_normal\", kernel_regularizer=l2(5e-5), activation='relu')(x)\n",
    "    x = BatchNormalization()(x)\n",
    "    x = Dropout(top_dropout_rate, name='top_dropout_3')(x)\n",
    "    # x = Dense(256, name='top_dense_3', kernel_initializer=\"he_normal\", kernel_regularizer=l2(5e-5), activation='relu')(x)\n",
    "    # x = BatchNormalization()(x)\n",
    "    # x = Dropout(top_dropout_rate, name='top_dropout_4')(x)\n",
    "    outputs = Dense(16, activation='softmax', name='pred', kernel_regularizer=l2(5e-5))(x)\n",
    "\n",
    "    model1 = tf.keras.Model(inputs, outputs, name='RestNet-VGG-Inception-ResNet-FineTune')\n",
    "    return model1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 248,
   "id": "c6574006",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"RestNet-VGG-Inception-ResNet-FineTune\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_5 (InputLayer)        [(None, 5, 2500)]         0         \n",
      "                                                                 \n",
      " flatten_4 (Flatten)         (None, 12500)             0         \n",
      "                                                                 \n",
      " batch_normalization_8 (Batc  (None, 12500)            50000     \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " top_dropout_2 (Dropout)     (None, 12500)             0         \n",
      "                                                                 \n",
      " top_dense_2 (Dense)         (None, 512)               6400512   \n",
      "                                                                 \n",
      " batch_normalization_9 (Batc  (None, 512)              2048      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " top_dropout_3 (Dropout)     (None, 512)               0         \n",
      "                                                                 \n",
      " pred (Dense)                (None, 16)                8208      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 6,460,768\n",
      "Trainable params: 6,434,744\n",
      "Non-trainable params: 26,024\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model1 = ftmodel()\n",
    "model1.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "id": "04f47ff8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 249,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf.keras.utils.plot_model(model1, show_shapes=True, to_file='./Images/ResNet-VGG-Inception-ResNet-4Piece-PCA-Model.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e56505e2",
   "metadata": {},
   "source": [
    "### Model Training"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "70b7e206",
   "metadata": {},
   "source": [
    "The fine-tuning will be done on the ResNet+VGG+Inception-ResNet's PCA reduced $1024D+256D+768D$ representation vectors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "id": "96524578",
   "metadata": {},
   "outputs": [],
   "source": [
    "model1.compile(loss = 'sparse_categorical_crossentropy', metrics = ['accuracy'], optimizer = Adam(learning_rate = 0.0003))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "id": "20df8897",
   "metadata": {},
   "outputs": [],
   "source": [
    "EPOCHS = 1000\n",
    "checkpoint_filepaths = ['./Models/ResNet-VGG-Inception-ResNet-4P/auto-ResNet-VGG-Inception-ResNet-FT-model-weight']\n",
    "model_checkpoint_callback = [tf.keras.callbacks.ModelCheckpoint(\n",
    "    filepath=checkpoint_filepath,\n",
    "    save_weights_only=True,\n",
    "    monitor = \"loss\",\n",
    "    mode='min',\n",
    "    save_best_only=True) for checkpoint_filepath in checkpoint_filepaths]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 252,
   "id": "03354278",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/1000\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-10-09 15:49:05.324599: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "18/19 [===========================>..] - ETA: 0s - loss: 3.2938 - accuracy: 0.2027"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-10-09 15:49:06.792927: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19/19 [==============================] - 2s 84ms/step - loss: 3.2484 - accuracy: 0.2113 - val_loss: 1.5581 - val_accuracy: 0.5875\n",
      "Epoch 2/1000\n",
      "19/19 [==============================] - 1s 36ms/step - loss: 1.6424 - accuracy: 0.5553 - val_loss: 1.3209 - val_accuracy: 0.6513\n",
      "Epoch 3/1000\n",
      "19/19 [==============================] - 1s 37ms/step - loss: 1.1199 - accuracy: 0.6949 - val_loss: 1.1850 - val_accuracy: 0.6912\n",
      "Epoch 4/1000\n",
      "19/19 [==============================] - 1s 30ms/step - loss: 0.8680 - accuracy: 0.7616 - val_loss: 1.0946 - val_accuracy: 0.7225\n",
      "Epoch 5/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.7042 - accuracy: 0.8087 - val_loss: 1.0307 - val_accuracy: 0.7362\n",
      "Epoch 6/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.5991 - accuracy: 0.8413 - val_loss: 0.9849 - val_accuracy: 0.7450\n",
      "Epoch 7/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.5176 - accuracy: 0.8659 - val_loss: 0.9468 - val_accuracy: 0.7512\n",
      "Epoch 8/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.4511 - accuracy: 0.8840 - val_loss: 0.9157 - val_accuracy: 0.7612\n",
      "Epoch 9/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.3989 - accuracy: 0.8984 - val_loss: 0.8905 - val_accuracy: 0.7738\n",
      "Epoch 10/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.3540 - accuracy: 0.9122 - val_loss: 0.8678 - val_accuracy: 0.7862\n",
      "Epoch 11/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.3216 - accuracy: 0.9248 - val_loss: 0.8496 - val_accuracy: 0.7825\n",
      "Epoch 12/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.2972 - accuracy: 0.9303 - val_loss: 0.8333 - val_accuracy: 0.7862\n",
      "Epoch 13/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.2722 - accuracy: 0.9388 - val_loss: 0.8223 - val_accuracy: 0.7925\n",
      "Epoch 14/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.2497 - accuracy: 0.9460 - val_loss: 0.8112 - val_accuracy: 0.7887\n",
      "Epoch 15/1000\n",
      "19/19 [==============================] - 1s 30ms/step - loss: 0.2369 - accuracy: 0.9505 - val_loss: 0.8046 - val_accuracy: 0.7912\n",
      "Epoch 16/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.2136 - accuracy: 0.9589 - val_loss: 0.7977 - val_accuracy: 0.7912\n",
      "Epoch 17/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.2015 - accuracy: 0.9624 - val_loss: 0.7882 - val_accuracy: 0.7900\n",
      "Epoch 18/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1947 - accuracy: 0.9616 - val_loss: 0.7843 - val_accuracy: 0.7950\n",
      "Epoch 19/1000\n",
      "19/19 [==============================] - 1s 29ms/step - loss: 0.1859 - accuracy: 0.9663 - val_loss: 0.7822 - val_accuracy: 0.7950\n",
      "Epoch 20/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1781 - accuracy: 0.9682 - val_loss: 0.7840 - val_accuracy: 0.8012\n",
      "Epoch 21/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1667 - accuracy: 0.9722 - val_loss: 0.7799 - val_accuracy: 0.8025\n",
      "Epoch 22/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1578 - accuracy: 0.9749 - val_loss: 0.7760 - val_accuracy: 0.8050\n",
      "Epoch 23/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1543 - accuracy: 0.9752 - val_loss: 0.7730 - val_accuracy: 0.8000\n",
      "Epoch 24/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1459 - accuracy: 0.9777 - val_loss: 0.7682 - val_accuracy: 0.8050\n",
      "Epoch 25/1000\n",
      "19/19 [==============================] - 1s 41ms/step - loss: 0.1446 - accuracy: 0.9778 - val_loss: 0.7718 - val_accuracy: 0.8125\n",
      "Epoch 26/1000\n",
      "19/19 [==============================] - 1s 27ms/step - loss: 0.1385 - accuracy: 0.9791 - val_loss: 0.7760 - val_accuracy: 0.8112\n",
      "Epoch 27/1000\n",
      "19/19 [==============================] - 1s 38ms/step - loss: 0.1304 - accuracy: 0.9820 - val_loss: 0.7769 - val_accuracy: 0.8187\n",
      "Epoch 28/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.1308 - accuracy: 0.9828 - val_loss: 0.7796 - val_accuracy: 0.8162\n",
      "Epoch 29/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1283 - accuracy: 0.9834 - val_loss: 0.7819 - val_accuracy: 0.8150\n",
      "Epoch 30/1000\n",
      "19/19 [==============================] - 1s 29ms/step - loss: 0.1257 - accuracy: 0.9826 - val_loss: 0.7873 - val_accuracy: 0.8112\n",
      "Epoch 31/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1226 - accuracy: 0.9846 - val_loss: 0.7932 - val_accuracy: 0.8112\n",
      "Epoch 32/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1192 - accuracy: 0.9849 - val_loss: 0.8008 - val_accuracy: 0.8137\n",
      "Epoch 33/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.1193 - accuracy: 0.9860 - val_loss: 0.8058 - val_accuracy: 0.8162\n",
      "Epoch 34/1000\n",
      "19/19 [==============================] - 1s 29ms/step - loss: 0.1145 - accuracy: 0.9850 - val_loss: 0.8074 - val_accuracy: 0.8112\n",
      "Epoch 35/1000\n",
      "19/19 [==============================] - 1s 30ms/step - loss: 0.1128 - accuracy: 0.9870 - val_loss: 0.8109 - val_accuracy: 0.8112\n",
      "Epoch 36/1000\n",
      "19/19 [==============================] - 1s 30ms/step - loss: 0.1125 - accuracy: 0.9870 - val_loss: 0.8219 - val_accuracy: 0.8112\n",
      "Epoch 37/1000\n",
      "19/19 [==============================] - 1s 29ms/step - loss: 0.1094 - accuracy: 0.9882 - val_loss: 0.8274 - val_accuracy: 0.8087\n",
      "Epoch 38/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1070 - accuracy: 0.9866 - val_loss: 0.8305 - val_accuracy: 0.8100\n",
      "Epoch 39/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.1075 - accuracy: 0.9882 - val_loss: 0.8371 - val_accuracy: 0.8112\n",
      "Epoch 40/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.1029 - accuracy: 0.9895 - val_loss: 0.8409 - val_accuracy: 0.8125\n",
      "Epoch 41/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.1033 - accuracy: 0.9882 - val_loss: 0.8430 - val_accuracy: 0.8125\n",
      "Epoch 42/1000\n",
      "19/19 [==============================] - 1s 29ms/step - loss: 0.1028 - accuracy: 0.9891 - val_loss: 0.8456 - val_accuracy: 0.8112\n",
      "Epoch 43/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.1036 - accuracy: 0.9891 - val_loss: 0.8479 - val_accuracy: 0.8112\n",
      "Epoch 44/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0978 - accuracy: 0.9920 - val_loss: 0.8533 - val_accuracy: 0.8137\n",
      "Epoch 45/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0980 - accuracy: 0.9903 - val_loss: 0.8543 - val_accuracy: 0.8137\n",
      "Epoch 46/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.1005 - accuracy: 0.9889 - val_loss: 0.8573 - val_accuracy: 0.8125\n",
      "Epoch 47/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0997 - accuracy: 0.9888 - val_loss: 0.8606 - val_accuracy: 0.8125\n",
      "Epoch 48/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0979 - accuracy: 0.9904 - val_loss: 0.8665 - val_accuracy: 0.8162\n",
      "Epoch 49/1000\n",
      "19/19 [==============================] - 1s 30ms/step - loss: 0.0960 - accuracy: 0.9905 - val_loss: 0.8691 - val_accuracy: 0.8150\n",
      "Epoch 50/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0933 - accuracy: 0.9921 - val_loss: 0.8659 - val_accuracy: 0.8187\n",
      "Epoch 51/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0958 - accuracy: 0.9905 - val_loss: 0.8649 - val_accuracy: 0.8162\n",
      "Epoch 52/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0946 - accuracy: 0.9910 - val_loss: 0.8670 - val_accuracy: 0.8112\n",
      "Epoch 53/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0940 - accuracy: 0.9911 - val_loss: 0.8698 - val_accuracy: 0.8125\n",
      "Epoch 54/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0933 - accuracy: 0.9916 - val_loss: 0.8703 - val_accuracy: 0.8100\n",
      "Epoch 55/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0926 - accuracy: 0.9915 - val_loss: 0.8684 - val_accuracy: 0.8112\n",
      "Epoch 56/1000\n",
      "19/19 [==============================] - 1s 31ms/step - loss: 0.0903 - accuracy: 0.9922 - val_loss: 0.8756 - val_accuracy: 0.8162\n",
      "Epoch 57/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0917 - accuracy: 0.9919 - val_loss: 0.8773 - val_accuracy: 0.8137\n",
      "Epoch 58/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9916 - val_loss: 0.8816 - val_accuracy: 0.8137\n",
      "Epoch 59/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0889 - accuracy: 0.9914 - val_loss: 0.8867 - val_accuracy: 0.8175\n",
      "Epoch 60/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0884 - accuracy: 0.9924 - val_loss: 0.8901 - val_accuracy: 0.8187\n",
      "Epoch 61/1000\n",
      "19/19 [==============================] - 1s 32ms/step - loss: 0.0874 - accuracy: 0.9930 - val_loss: 0.8941 - val_accuracy: 0.8175\n",
      "Epoch 62/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0901 - accuracy: 0.9920 - val_loss: 0.8918 - val_accuracy: 0.8150\n",
      "Epoch 63/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0874 - accuracy: 0.9930 - val_loss: 0.8947 - val_accuracy: 0.8137\n",
      "Epoch 64/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0880 - accuracy: 0.9927 - val_loss: 0.8897 - val_accuracy: 0.8137\n",
      "Epoch 65/1000\n",
      "19/19 [==============================] - 1s 29ms/step - loss: 0.0873 - accuracy: 0.9932 - val_loss: 0.8866 - val_accuracy: 0.8162\n",
      "Epoch 66/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0841 - accuracy: 0.9934 - val_loss: 0.8918 - val_accuracy: 0.8162\n",
      "Epoch 67/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0858 - accuracy: 0.9932 - val_loss: 0.8945 - val_accuracy: 0.8150\n",
      "Epoch 68/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0856 - accuracy: 0.9937 - val_loss: 0.8897 - val_accuracy: 0.8137\n",
      "Epoch 69/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0852 - accuracy: 0.9934 - val_loss: 0.8851 - val_accuracy: 0.8137\n",
      "Epoch 70/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0839 - accuracy: 0.9944 - val_loss: 0.8908 - val_accuracy: 0.8187\n",
      "Epoch 71/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0849 - accuracy: 0.9937 - val_loss: 0.8964 - val_accuracy: 0.8150\n",
      "Epoch 72/1000\n",
      "19/19 [==============================] - 1s 31ms/step - loss: 0.0818 - accuracy: 0.9947 - val_loss: 0.9025 - val_accuracy: 0.8150\n",
      "Epoch 73/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0851 - accuracy: 0.9934 - val_loss: 0.9012 - val_accuracy: 0.8187\n",
      "Epoch 74/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0822 - accuracy: 0.9947 - val_loss: 0.9053 - val_accuracy: 0.8187\n",
      "Epoch 75/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0836 - accuracy: 0.9939 - val_loss: 0.9119 - val_accuracy: 0.8125\n",
      "Epoch 76/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0849 - accuracy: 0.9935 - val_loss: 0.9071 - val_accuracy: 0.8112\n",
      "Epoch 77/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0820 - accuracy: 0.9950 - val_loss: 0.9173 - val_accuracy: 0.8137\n",
      "Epoch 78/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0807 - accuracy: 0.9951 - val_loss: 0.9180 - val_accuracy: 0.8175\n",
      "Epoch 79/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0825 - accuracy: 0.9941 - val_loss: 0.9113 - val_accuracy: 0.8187\n",
      "Epoch 80/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0813 - accuracy: 0.9948 - val_loss: 0.9140 - val_accuracy: 0.8162\n",
      "Epoch 81/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0812 - accuracy: 0.9949 - val_loss: 0.9128 - val_accuracy: 0.8125\n",
      "Epoch 82/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0812 - accuracy: 0.9947 - val_loss: 0.9168 - val_accuracy: 0.8137\n",
      "Epoch 83/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0818 - accuracy: 0.9939 - val_loss: 0.9222 - val_accuracy: 0.8150\n",
      "Epoch 84/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0808 - accuracy: 0.9950 - val_loss: 0.9235 - val_accuracy: 0.8112\n",
      "Epoch 85/1000\n",
      "19/19 [==============================] - 1s 32ms/step - loss: 0.0785 - accuracy: 0.9955 - val_loss: 0.9224 - val_accuracy: 0.8112\n",
      "Epoch 86/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0804 - accuracy: 0.9949 - val_loss: 0.9228 - val_accuracy: 0.8137\n",
      "Epoch 87/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0813 - accuracy: 0.9940 - val_loss: 0.9235 - val_accuracy: 0.8200\n",
      "Epoch 88/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0788 - accuracy: 0.9951 - val_loss: 0.9264 - val_accuracy: 0.8225\n",
      "Epoch 89/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0787 - accuracy: 0.9960 - val_loss: 0.9324 - val_accuracy: 0.8187\n",
      "Epoch 90/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0787 - accuracy: 0.9952 - val_loss: 0.9394 - val_accuracy: 0.8150\n",
      "Epoch 91/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0773 - accuracy: 0.9955 - val_loss: 0.9348 - val_accuracy: 0.8175\n",
      "Epoch 92/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0792 - accuracy: 0.9953 - val_loss: 0.9295 - val_accuracy: 0.8187\n",
      "Epoch 93/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0793 - accuracy: 0.9950 - val_loss: 0.9317 - val_accuracy: 0.8200\n",
      "Epoch 94/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0791 - accuracy: 0.9951 - val_loss: 0.9366 - val_accuracy: 0.8175\n",
      "Epoch 95/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0805 - accuracy: 0.9934 - val_loss: 0.9492 - val_accuracy: 0.8175\n",
      "Epoch 96/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0796 - accuracy: 0.9939 - val_loss: 0.9577 - val_accuracy: 0.8187\n",
      "Epoch 97/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0773 - accuracy: 0.9959 - val_loss: 0.9598 - val_accuracy: 0.8187\n",
      "Epoch 98/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0777 - accuracy: 0.9954 - val_loss: 0.9546 - val_accuracy: 0.8175\n",
      "Epoch 99/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0782 - accuracy: 0.9953 - val_loss: 0.9532 - val_accuracy: 0.8125\n",
      "Epoch 100/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0779 - accuracy: 0.9951 - val_loss: 0.9559 - val_accuracy: 0.8112\n",
      "Epoch 101/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0775 - accuracy: 0.9957 - val_loss: 0.9559 - val_accuracy: 0.8087\n",
      "Epoch 102/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0778 - accuracy: 0.9955 - val_loss: 0.9546 - val_accuracy: 0.8112\n",
      "Epoch 103/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0799 - accuracy: 0.9947 - val_loss: 0.9496 - val_accuracy: 0.8125\n",
      "Epoch 104/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0778 - accuracy: 0.9951 - val_loss: 0.9509 - val_accuracy: 0.8125\n",
      "Epoch 105/1000\n",
      "19/19 [==============================] - 1s 29ms/step - loss: 0.0768 - accuracy: 0.9955 - val_loss: 0.9491 - val_accuracy: 0.8112\n",
      "Epoch 106/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0777 - accuracy: 0.9953 - val_loss: 0.9428 - val_accuracy: 0.8100\n",
      "Epoch 107/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0788 - accuracy: 0.9952 - val_loss: 0.9437 - val_accuracy: 0.8075\n",
      "Epoch 108/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0770 - accuracy: 0.9955 - val_loss: 0.9474 - val_accuracy: 0.8100\n",
      "Epoch 109/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0787 - accuracy: 0.9953 - val_loss: 0.9434 - val_accuracy: 0.8137\n",
      "Epoch 110/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0767 - accuracy: 0.9966 - val_loss: 0.9488 - val_accuracy: 0.8112\n",
      "Epoch 111/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0776 - accuracy: 0.9953 - val_loss: 0.9518 - val_accuracy: 0.8087\n",
      "Epoch 112/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0767 - accuracy: 0.9957 - val_loss: 0.9497 - val_accuracy: 0.8137\n",
      "Epoch 113/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0770 - accuracy: 0.9957 - val_loss: 0.9550 - val_accuracy: 0.8100\n",
      "Epoch 114/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0780 - accuracy: 0.9953 - val_loss: 0.9588 - val_accuracy: 0.8062\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 115/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0760 - accuracy: 0.9965 - val_loss: 0.9572 - val_accuracy: 0.8100\n",
      "Epoch 116/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0761 - accuracy: 0.9961 - val_loss: 0.9558 - val_accuracy: 0.8100\n",
      "Epoch 117/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0779 - accuracy: 0.9955 - val_loss: 0.9566 - val_accuracy: 0.8175\n",
      "Epoch 118/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0766 - accuracy: 0.9966 - val_loss: 0.9512 - val_accuracy: 0.8162\n",
      "Epoch 119/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0801 - accuracy: 0.9947 - val_loss: 0.9519 - val_accuracy: 0.8175\n",
      "Epoch 120/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0771 - accuracy: 0.9953 - val_loss: 0.9583 - val_accuracy: 0.8162\n",
      "Epoch 121/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0771 - accuracy: 0.9952 - val_loss: 0.9635 - val_accuracy: 0.8162\n",
      "Epoch 122/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0765 - accuracy: 0.9958 - val_loss: 0.9640 - val_accuracy: 0.8212\n",
      "Epoch 123/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0756 - accuracy: 0.9963 - val_loss: 0.9565 - val_accuracy: 0.8200\n",
      "Epoch 124/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0763 - accuracy: 0.9959 - val_loss: 0.9530 - val_accuracy: 0.8162\n",
      "Epoch 125/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0760 - accuracy: 0.9968 - val_loss: 0.9493 - val_accuracy: 0.8162\n",
      "Epoch 126/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0774 - accuracy: 0.9956 - val_loss: 0.9536 - val_accuracy: 0.8162\n",
      "Epoch 127/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0785 - accuracy: 0.9953 - val_loss: 0.9633 - val_accuracy: 0.8162\n",
      "Epoch 128/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0804 - accuracy: 0.9941 - val_loss: 0.9622 - val_accuracy: 0.8175\n",
      "Epoch 129/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0768 - accuracy: 0.9961 - val_loss: 0.9566 - val_accuracy: 0.8175\n",
      "Epoch 130/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0763 - accuracy: 0.9966 - val_loss: 0.9532 - val_accuracy: 0.8250\n",
      "Epoch 131/1000\n",
      "19/19 [==============================] - 0s 20ms/step - loss: 0.0808 - accuracy: 0.9949 - val_loss: 0.9487 - val_accuracy: 0.8225\n",
      "Epoch 132/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0762 - accuracy: 0.9961 - val_loss: 0.9499 - val_accuracy: 0.8262\n",
      "Epoch 133/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0759 - accuracy: 0.9960 - val_loss: 0.9545 - val_accuracy: 0.8200\n",
      "Epoch 134/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0765 - accuracy: 0.9968 - val_loss: 0.9601 - val_accuracy: 0.8150\n",
      "Epoch 135/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0777 - accuracy: 0.9955 - val_loss: 0.9535 - val_accuracy: 0.8162\n",
      "Epoch 136/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0765 - accuracy: 0.9964 - val_loss: 0.9472 - val_accuracy: 0.8200\n",
      "Epoch 137/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0769 - accuracy: 0.9955 - val_loss: 0.9481 - val_accuracy: 0.8212\n",
      "Epoch 138/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0783 - accuracy: 0.9955 - val_loss: 0.9534 - val_accuracy: 0.8212\n",
      "Epoch 139/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0769 - accuracy: 0.9966 - val_loss: 0.9478 - val_accuracy: 0.8150\n",
      "Epoch 140/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0773 - accuracy: 0.9955 - val_loss: 0.9524 - val_accuracy: 0.8100\n",
      "Epoch 141/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0754 - accuracy: 0.9964 - val_loss: 0.9480 - val_accuracy: 0.8137\n",
      "Epoch 142/1000\n",
      "19/19 [==============================] - 1s 29ms/step - loss: 0.0746 - accuracy: 0.9969 - val_loss: 0.9520 - val_accuracy: 0.8112\n",
      "Epoch 143/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0756 - accuracy: 0.9967 - val_loss: 0.9512 - val_accuracy: 0.8125\n",
      "Epoch 144/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0754 - accuracy: 0.9964 - val_loss: 0.9537 - val_accuracy: 0.8150\n",
      "Epoch 145/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0769 - accuracy: 0.9958 - val_loss: 0.9475 - val_accuracy: 0.8137\n",
      "Epoch 146/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0761 - accuracy: 0.9966 - val_loss: 0.9490 - val_accuracy: 0.8162\n",
      "Epoch 147/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0748 - accuracy: 0.9970 - val_loss: 0.9485 - val_accuracy: 0.8137\n",
      "Epoch 148/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0755 - accuracy: 0.9959 - val_loss: 0.9502 - val_accuracy: 0.8150\n",
      "Epoch 149/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0775 - accuracy: 0.9960 - val_loss: 0.9473 - val_accuracy: 0.8125\n",
      "Epoch 150/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0778 - accuracy: 0.9957 - val_loss: 0.9477 - val_accuracy: 0.8125\n",
      "Epoch 151/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0747 - accuracy: 0.9969 - val_loss: 0.9450 - val_accuracy: 0.8087\n",
      "Epoch 152/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0785 - accuracy: 0.9951 - val_loss: 0.9491 - val_accuracy: 0.8112\n",
      "Epoch 153/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0748 - accuracy: 0.9964 - val_loss: 0.9513 - val_accuracy: 0.8050\n",
      "Epoch 154/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0767 - accuracy: 0.9958 - val_loss: 0.9553 - val_accuracy: 0.8037\n",
      "Epoch 155/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0770 - accuracy: 0.9964 - val_loss: 0.9594 - val_accuracy: 0.8075\n",
      "Epoch 156/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0765 - accuracy: 0.9963 - val_loss: 0.9549 - val_accuracy: 0.8100\n",
      "Epoch 157/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0770 - accuracy: 0.9964 - val_loss: 0.9548 - val_accuracy: 0.8162\n",
      "Epoch 158/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0762 - accuracy: 0.9962 - val_loss: 0.9605 - val_accuracy: 0.8200\n",
      "Epoch 159/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0744 - accuracy: 0.9966 - val_loss: 0.9595 - val_accuracy: 0.8137\n",
      "Epoch 160/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0771 - accuracy: 0.9957 - val_loss: 0.9587 - val_accuracy: 0.8112\n",
      "Epoch 161/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0748 - accuracy: 0.9967 - val_loss: 0.9570 - val_accuracy: 0.8075\n",
      "Epoch 162/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0749 - accuracy: 0.9963 - val_loss: 0.9588 - val_accuracy: 0.8100\n",
      "Epoch 163/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0753 - accuracy: 0.9968 - val_loss: 0.9524 - val_accuracy: 0.8100\n",
      "Epoch 164/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0775 - accuracy: 0.9961 - val_loss: 0.9616 - val_accuracy: 0.8125\n",
      "Epoch 165/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0782 - accuracy: 0.9960 - val_loss: 0.9733 - val_accuracy: 0.8062\n",
      "Epoch 166/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0764 - accuracy: 0.9968 - val_loss: 0.9743 - val_accuracy: 0.8112\n",
      "Epoch 167/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0774 - accuracy: 0.9956 - val_loss: 0.9626 - val_accuracy: 0.8125\n",
      "Epoch 168/1000\n",
      "19/19 [==============================] - 1s 28ms/step - loss: 0.0742 - accuracy: 0.9976 - val_loss: 0.9625 - val_accuracy: 0.8137\n",
      "Epoch 169/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0773 - accuracy: 0.9960 - val_loss: 0.9613 - val_accuracy: 0.8150\n",
      "Epoch 170/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0768 - accuracy: 0.9961 - val_loss: 0.9649 - val_accuracy: 0.8162\n",
      "Epoch 171/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0758 - accuracy: 0.9964 - val_loss: 0.9634 - val_accuracy: 0.8200\n",
      "Epoch 172/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0789 - accuracy: 0.9955 - val_loss: 0.9555 - val_accuracy: 0.8175\n",
      "Epoch 173/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0768 - accuracy: 0.9964 - val_loss: 0.9590 - val_accuracy: 0.8200\n",
      "Epoch 174/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0771 - accuracy: 0.9963 - val_loss: 0.9703 - val_accuracy: 0.8200\n",
      "Epoch 175/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0800 - accuracy: 0.9961 - val_loss: 0.9658 - val_accuracy: 0.8175\n",
      "Epoch 176/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0783 - accuracy: 0.9960 - val_loss: 0.9608 - val_accuracy: 0.8187\n",
      "Epoch 177/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0767 - accuracy: 0.9973 - val_loss: 0.9616 - val_accuracy: 0.8075\n",
      "Epoch 178/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0791 - accuracy: 0.9955 - val_loss: 0.9645 - val_accuracy: 0.8125\n",
      "Epoch 179/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0770 - accuracy: 0.9959 - val_loss: 0.9646 - val_accuracy: 0.8112\n",
      "Epoch 180/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0796 - accuracy: 0.9963 - val_loss: 0.9636 - val_accuracy: 0.8112\n",
      "Epoch 181/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0750 - accuracy: 0.9972 - val_loss: 0.9632 - val_accuracy: 0.8112\n",
      "Epoch 182/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0796 - accuracy: 0.9961 - val_loss: 0.9581 - val_accuracy: 0.8187\n",
      "Epoch 183/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0791 - accuracy: 0.9964 - val_loss: 0.9566 - val_accuracy: 0.8187\n",
      "Epoch 184/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0801 - accuracy: 0.9959 - val_loss: 0.9569 - val_accuracy: 0.8187\n",
      "Epoch 185/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0796 - accuracy: 0.9959 - val_loss: 0.9524 - val_accuracy: 0.8175\n",
      "Epoch 186/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0774 - accuracy: 0.9970 - val_loss: 0.9570 - val_accuracy: 0.8212\n",
      "Epoch 187/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0789 - accuracy: 0.9961 - val_loss: 0.9606 - val_accuracy: 0.8212\n",
      "Epoch 188/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0791 - accuracy: 0.9964 - val_loss: 0.9562 - val_accuracy: 0.8200\n",
      "Epoch 189/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0778 - accuracy: 0.9967 - val_loss: 0.9587 - val_accuracy: 0.8187\n",
      "Epoch 190/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0799 - accuracy: 0.9956 - val_loss: 0.9612 - val_accuracy: 0.8212\n",
      "Epoch 191/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0776 - accuracy: 0.9968 - val_loss: 0.9579 - val_accuracy: 0.8200\n",
      "Epoch 192/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0803 - accuracy: 0.9959 - val_loss: 0.9544 - val_accuracy: 0.8237\n",
      "Epoch 193/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0785 - accuracy: 0.9966 - val_loss: 0.9507 - val_accuracy: 0.8200\n",
      "Epoch 194/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0809 - accuracy: 0.9957 - val_loss: 0.9400 - val_accuracy: 0.8212\n",
      "Epoch 195/1000\n",
      "19/19 [==============================] - 0s 20ms/step - loss: 0.0803 - accuracy: 0.9955 - val_loss: 0.9364 - val_accuracy: 0.8275\n",
      "Epoch 196/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0774 - accuracy: 0.9966 - val_loss: 0.9371 - val_accuracy: 0.8262\n",
      "Epoch 197/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0786 - accuracy: 0.9968 - val_loss: 0.9357 - val_accuracy: 0.8262\n",
      "Epoch 198/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0787 - accuracy: 0.9964 - val_loss: 0.9341 - val_accuracy: 0.8262\n",
      "Epoch 199/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0803 - accuracy: 0.9954 - val_loss: 0.9349 - val_accuracy: 0.8237\n",
      "Epoch 200/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0801 - accuracy: 0.9961 - val_loss: 0.9338 - val_accuracy: 0.8212\n",
      "Epoch 201/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0789 - accuracy: 0.9968 - val_loss: 0.9338 - val_accuracy: 0.8225\n",
      "Epoch 202/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0786 - accuracy: 0.9973 - val_loss: 0.9332 - val_accuracy: 0.8262\n",
      "Epoch 203/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0784 - accuracy: 0.9967 - val_loss: 0.9301 - val_accuracy: 0.8237\n",
      "Epoch 204/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0790 - accuracy: 0.9962 - val_loss: 0.9415 - val_accuracy: 0.8212\n",
      "Epoch 205/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0791 - accuracy: 0.9968 - val_loss: 0.9412 - val_accuracy: 0.8162\n",
      "Epoch 206/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0784 - accuracy: 0.9964 - val_loss: 0.9455 - val_accuracy: 0.8137\n",
      "Epoch 207/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0795 - accuracy: 0.9968 - val_loss: 0.9449 - val_accuracy: 0.8162\n",
      "Epoch 208/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0792 - accuracy: 0.9965 - val_loss: 0.9363 - val_accuracy: 0.8200\n",
      "Epoch 209/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0806 - accuracy: 0.9966 - val_loss: 0.9282 - val_accuracy: 0.8175\n",
      "Epoch 210/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0796 - accuracy: 0.9972 - val_loss: 0.9335 - val_accuracy: 0.8212\n",
      "Epoch 211/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0793 - accuracy: 0.9977 - val_loss: 0.9401 - val_accuracy: 0.8225\n",
      "Epoch 212/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0786 - accuracy: 0.9967 - val_loss: 0.9454 - val_accuracy: 0.8200\n",
      "Epoch 213/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0811 - accuracy: 0.9962 - val_loss: 0.9413 - val_accuracy: 0.8225\n",
      "Epoch 214/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0798 - accuracy: 0.9970 - val_loss: 0.9466 - val_accuracy: 0.8225\n",
      "Epoch 215/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0807 - accuracy: 0.9966 - val_loss: 0.9468 - val_accuracy: 0.8237\n",
      "Epoch 216/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0800 - accuracy: 0.9971 - val_loss: 0.9407 - val_accuracy: 0.8300\n",
      "Epoch 217/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0783 - accuracy: 0.9972 - val_loss: 0.9303 - val_accuracy: 0.8237\n",
      "Epoch 218/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0790 - accuracy: 0.9966 - val_loss: 0.9162 - val_accuracy: 0.8187\n",
      "Epoch 219/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0806 - accuracy: 0.9956 - val_loss: 0.9203 - val_accuracy: 0.8225\n",
      "Epoch 220/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0801 - accuracy: 0.9969 - val_loss: 0.9283 - val_accuracy: 0.8150\n",
      "Epoch 221/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0787 - accuracy: 0.9973 - val_loss: 0.9373 - val_accuracy: 0.8187\n",
      "Epoch 222/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0820 - accuracy: 0.9963 - val_loss: 0.9444 - val_accuracy: 0.8150\n",
      "Epoch 223/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0799 - accuracy: 0.9964 - val_loss: 0.9401 - val_accuracy: 0.8162\n",
      "Epoch 224/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0807 - accuracy: 0.9969 - val_loss: 0.9390 - val_accuracy: 0.8187\n",
      "Epoch 225/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0802 - accuracy: 0.9966 - val_loss: 0.9363 - val_accuracy: 0.8187\n",
      "Epoch 226/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0806 - accuracy: 0.9965 - val_loss: 0.9363 - val_accuracy: 0.8162\n",
      "Epoch 227/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0799 - accuracy: 0.9970 - val_loss: 0.9391 - val_accuracy: 0.8150\n",
      "Epoch 228/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0789 - accuracy: 0.9974 - val_loss: 0.9413 - val_accuracy: 0.8150\n",
      "Epoch 229/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0796 - accuracy: 0.9968 - val_loss: 0.9356 - val_accuracy: 0.8137\n",
      "Epoch 230/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0810 - accuracy: 0.9964 - val_loss: 0.9281 - val_accuracy: 0.8200\n",
      "Epoch 231/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0804 - accuracy: 0.9963 - val_loss: 0.9348 - val_accuracy: 0.8137\n",
      "Epoch 232/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0810 - accuracy: 0.9972 - val_loss: 0.9396 - val_accuracy: 0.8112\n",
      "Epoch 233/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0803 - accuracy: 0.9970 - val_loss: 0.9389 - val_accuracy: 0.8175\n",
      "Epoch 234/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0804 - accuracy: 0.9969 - val_loss: 0.9286 - val_accuracy: 0.8162\n",
      "Epoch 235/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0799 - accuracy: 0.9972 - val_loss: 0.9249 - val_accuracy: 0.8200\n",
      "Epoch 236/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0804 - accuracy: 0.9968 - val_loss: 0.9245 - val_accuracy: 0.8162\n",
      "Epoch 237/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0813 - accuracy: 0.9968 - val_loss: 0.9224 - val_accuracy: 0.8175\n",
      "Epoch 238/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0809 - accuracy: 0.9964 - val_loss: 0.9237 - val_accuracy: 0.8225\n",
      "Epoch 239/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0808 - accuracy: 0.9968 - val_loss: 0.9206 - val_accuracy: 0.8187\n",
      "Epoch 240/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0798 - accuracy: 0.9969 - val_loss: 0.9194 - val_accuracy: 0.8200\n",
      "Epoch 241/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0816 - accuracy: 0.9960 - val_loss: 0.9176 - val_accuracy: 0.8200\n",
      "Epoch 242/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0818 - accuracy: 0.9968 - val_loss: 0.9107 - val_accuracy: 0.8225\n",
      "Epoch 243/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0816 - accuracy: 0.9964 - val_loss: 0.9126 - val_accuracy: 0.8237\n",
      "Epoch 244/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0810 - accuracy: 0.9970 - val_loss: 0.9233 - val_accuracy: 0.8262\n",
      "Epoch 245/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0810 - accuracy: 0.9970 - val_loss: 0.9337 - val_accuracy: 0.8212\n",
      "Epoch 246/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0803 - accuracy: 0.9968 - val_loss: 0.9425 - val_accuracy: 0.8225\n",
      "Epoch 247/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0820 - accuracy: 0.9968 - val_loss: 0.9411 - val_accuracy: 0.8250\n",
      "Epoch 248/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0830 - accuracy: 0.9970 - val_loss: 0.9415 - val_accuracy: 0.8237\n",
      "Epoch 249/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0809 - accuracy: 0.9976 - val_loss: 0.9443 - val_accuracy: 0.8262\n",
      "Epoch 250/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0828 - accuracy: 0.9966 - val_loss: 0.9368 - val_accuracy: 0.8262\n",
      "Epoch 251/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0827 - accuracy: 0.9962 - val_loss: 0.9307 - val_accuracy: 0.8262\n",
      "Epoch 252/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0830 - accuracy: 0.9966 - val_loss: 0.9275 - val_accuracy: 0.8250\n",
      "Epoch 253/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0817 - accuracy: 0.9974 - val_loss: 0.9304 - val_accuracy: 0.8250\n",
      "Epoch 254/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0830 - accuracy: 0.9963 - val_loss: 0.9297 - val_accuracy: 0.8237\n",
      "Epoch 255/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0809 - accuracy: 0.9972 - val_loss: 0.9297 - val_accuracy: 0.8237\n",
      "Epoch 256/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0831 - accuracy: 0.9970 - val_loss: 0.9332 - val_accuracy: 0.8237\n",
      "Epoch 257/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0816 - accuracy: 0.9970 - val_loss: 0.9295 - val_accuracy: 0.8225\n",
      "Epoch 258/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0834 - accuracy: 0.9964 - val_loss: 0.9390 - val_accuracy: 0.8212\n",
      "Epoch 259/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0831 - accuracy: 0.9966 - val_loss: 0.9533 - val_accuracy: 0.8175\n",
      "Epoch 260/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0845 - accuracy: 0.9959 - val_loss: 0.9582 - val_accuracy: 0.8212\n",
      "Epoch 261/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0831 - accuracy: 0.9968 - val_loss: 0.9533 - val_accuracy: 0.8262\n",
      "Epoch 262/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0843 - accuracy: 0.9962 - val_loss: 0.9557 - val_accuracy: 0.8200\n",
      "Epoch 263/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0809 - accuracy: 0.9974 - val_loss: 0.9471 - val_accuracy: 0.8175\n",
      "Epoch 264/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0831 - accuracy: 0.9969 - val_loss: 0.9513 - val_accuracy: 0.8187\n",
      "Epoch 265/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0837 - accuracy: 0.9964 - val_loss: 0.9481 - val_accuracy: 0.8187\n",
      "Epoch 266/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0841 - accuracy: 0.9963 - val_loss: 0.9417 - val_accuracy: 0.8200\n",
      "Epoch 267/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0838 - accuracy: 0.9974 - val_loss: 0.9387 - val_accuracy: 0.8212\n",
      "Epoch 268/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0827 - accuracy: 0.9966 - val_loss: 0.9378 - val_accuracy: 0.8200\n",
      "Epoch 269/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0839 - accuracy: 0.9966 - val_loss: 0.9396 - val_accuracy: 0.8175\n",
      "Epoch 270/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0840 - accuracy: 0.9971 - val_loss: 0.9423 - val_accuracy: 0.8162\n",
      "Epoch 271/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0831 - accuracy: 0.9973 - val_loss: 0.9420 - val_accuracy: 0.8175\n",
      "Epoch 272/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0842 - accuracy: 0.9970 - val_loss: 0.9436 - val_accuracy: 0.8200\n",
      "Epoch 273/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0857 - accuracy: 0.9963 - val_loss: 0.9420 - val_accuracy: 0.8175\n",
      "Epoch 274/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0837 - accuracy: 0.9970 - val_loss: 0.9507 - val_accuracy: 0.8175\n",
      "Epoch 275/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0832 - accuracy: 0.9972 - val_loss: 0.9490 - val_accuracy: 0.8137\n",
      "Epoch 276/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0846 - accuracy: 0.9970 - val_loss: 0.9463 - val_accuracy: 0.8150\n",
      "Epoch 277/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0843 - accuracy: 0.9966 - val_loss: 0.9453 - val_accuracy: 0.8187\n",
      "Epoch 278/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0836 - accuracy: 0.9971 - val_loss: 0.9557 - val_accuracy: 0.8150\n",
      "Epoch 279/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0865 - accuracy: 0.9961 - val_loss: 0.9500 - val_accuracy: 0.8175\n",
      "Epoch 280/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0840 - accuracy: 0.9972 - val_loss: 0.9574 - val_accuracy: 0.8150\n",
      "Epoch 281/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0826 - accuracy: 0.9976 - val_loss: 0.9593 - val_accuracy: 0.8187\n",
      "Epoch 282/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0839 - accuracy: 0.9967 - val_loss: 0.9554 - val_accuracy: 0.8187\n",
      "Epoch 283/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0836 - accuracy: 0.9970 - val_loss: 0.9617 - val_accuracy: 0.8112\n",
      "Epoch 284/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0839 - accuracy: 0.9972 - val_loss: 0.9680 - val_accuracy: 0.8087\n",
      "Epoch 285/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0852 - accuracy: 0.9968 - val_loss: 0.9689 - val_accuracy: 0.8087\n",
      "Epoch 286/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0853 - accuracy: 0.9970 - val_loss: 0.9665 - val_accuracy: 0.8100\n",
      "Epoch 287/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0852 - accuracy: 0.9963 - val_loss: 0.9674 - val_accuracy: 0.8112\n",
      "Epoch 288/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0839 - accuracy: 0.9973 - val_loss: 0.9826 - val_accuracy: 0.8100\n",
      "Epoch 289/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0840 - accuracy: 0.9970 - val_loss: 0.9882 - val_accuracy: 0.8062\n",
      "Epoch 290/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0856 - accuracy: 0.9961 - val_loss: 0.9931 - val_accuracy: 0.8112\n",
      "Epoch 291/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0861 - accuracy: 0.9971 - val_loss: 0.9893 - val_accuracy: 0.8125\n",
      "Epoch 292/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0851 - accuracy: 0.9969 - val_loss: 0.9830 - val_accuracy: 0.8137\n",
      "Epoch 293/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0839 - accuracy: 0.9970 - val_loss: 0.9827 - val_accuracy: 0.8137\n",
      "Epoch 294/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0834 - accuracy: 0.9968 - val_loss: 0.9806 - val_accuracy: 0.8112\n",
      "Epoch 295/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0852 - accuracy: 0.9966 - val_loss: 0.9814 - val_accuracy: 0.8150\n",
      "Epoch 296/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0849 - accuracy: 0.9961 - val_loss: 0.9748 - val_accuracy: 0.8100\n",
      "Epoch 297/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0852 - accuracy: 0.9971 - val_loss: 0.9717 - val_accuracy: 0.8087\n",
      "Epoch 298/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0865 - accuracy: 0.9966 - val_loss: 0.9751 - val_accuracy: 0.8112\n",
      "Epoch 299/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0849 - accuracy: 0.9974 - val_loss: 0.9828 - val_accuracy: 0.8112\n",
      "Epoch 300/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0846 - accuracy: 0.9973 - val_loss: 0.9807 - val_accuracy: 0.8125\n",
      "Epoch 301/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0851 - accuracy: 0.9970 - val_loss: 0.9759 - val_accuracy: 0.8150\n",
      "Epoch 302/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0866 - accuracy: 0.9968 - val_loss: 0.9769 - val_accuracy: 0.8225\n",
      "Epoch 303/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0871 - accuracy: 0.9970 - val_loss: 0.9791 - val_accuracy: 0.8187\n",
      "Epoch 304/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0850 - accuracy: 0.9972 - val_loss: 0.9743 - val_accuracy: 0.8175\n",
      "Epoch 305/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0850 - accuracy: 0.9974 - val_loss: 0.9743 - val_accuracy: 0.8162\n",
      "Epoch 306/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0837 - accuracy: 0.9976 - val_loss: 0.9733 - val_accuracy: 0.8175\n",
      "Epoch 307/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0859 - accuracy: 0.9971 - val_loss: 0.9710 - val_accuracy: 0.8137\n",
      "Epoch 308/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0847 - accuracy: 0.9974 - val_loss: 0.9619 - val_accuracy: 0.8200\n",
      "Epoch 309/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0838 - accuracy: 0.9972 - val_loss: 0.9558 - val_accuracy: 0.8262\n",
      "Epoch 310/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0860 - accuracy: 0.9968 - val_loss: 0.9540 - val_accuracy: 0.8225\n",
      "Epoch 311/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0844 - accuracy: 0.9975 - val_loss: 0.9545 - val_accuracy: 0.8237\n",
      "Epoch 312/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0862 - accuracy: 0.9968 - val_loss: 0.9586 - val_accuracy: 0.8200\n",
      "Epoch 313/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0832 - accuracy: 0.9979 - val_loss: 0.9551 - val_accuracy: 0.8200\n",
      "Epoch 314/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0835 - accuracy: 0.9974 - val_loss: 0.9503 - val_accuracy: 0.8162\n",
      "Epoch 315/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0833 - accuracy: 0.9980 - val_loss: 0.9527 - val_accuracy: 0.8200\n",
      "Epoch 316/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0835 - accuracy: 0.9980 - val_loss: 0.9557 - val_accuracy: 0.8187\n",
      "Epoch 317/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0842 - accuracy: 0.9976 - val_loss: 0.9500 - val_accuracy: 0.8187\n",
      "Epoch 318/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0837 - accuracy: 0.9970 - val_loss: 0.9431 - val_accuracy: 0.8200\n",
      "Epoch 319/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0834 - accuracy: 0.9978 - val_loss: 0.9413 - val_accuracy: 0.8200\n",
      "Epoch 320/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0840 - accuracy: 0.9974 - val_loss: 0.9422 - val_accuracy: 0.8162\n",
      "Epoch 321/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0851 - accuracy: 0.9972 - val_loss: 0.9346 - val_accuracy: 0.8200\n",
      "Epoch 322/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0851 - accuracy: 0.9970 - val_loss: 0.9296 - val_accuracy: 0.8237\n",
      "Epoch 323/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0839 - accuracy: 0.9976 - val_loss: 0.9323 - val_accuracy: 0.8237\n",
      "Epoch 324/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0849 - accuracy: 0.9972 - val_loss: 0.9328 - val_accuracy: 0.8262\n",
      "Epoch 325/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0841 - accuracy: 0.9974 - val_loss: 0.9354 - val_accuracy: 0.8250\n",
      "Epoch 326/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0828 - accuracy: 0.9980 - val_loss: 0.9380 - val_accuracy: 0.8250\n",
      "Epoch 327/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0837 - accuracy: 0.9977 - val_loss: 0.9342 - val_accuracy: 0.8237\n",
      "Epoch 328/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0830 - accuracy: 0.9973 - val_loss: 0.9330 - val_accuracy: 0.8250\n",
      "Epoch 329/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0826 - accuracy: 0.9981 - val_loss: 0.9333 - val_accuracy: 0.8262\n",
      "Epoch 330/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0850 - accuracy: 0.9968 - val_loss: 0.9330 - val_accuracy: 0.8275\n",
      "Epoch 331/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0823 - accuracy: 0.9980 - val_loss: 0.9355 - val_accuracy: 0.8225\n",
      "Epoch 332/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0830 - accuracy: 0.9973 - val_loss: 0.9443 - val_accuracy: 0.8250\n",
      "Epoch 333/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0828 - accuracy: 0.9978 - val_loss: 0.9416 - val_accuracy: 0.8250\n",
      "Epoch 334/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0850 - accuracy: 0.9968 - val_loss: 0.9485 - val_accuracy: 0.8262\n",
      "Epoch 335/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0851 - accuracy: 0.9973 - val_loss: 0.9467 - val_accuracy: 0.8225\n",
      "Epoch 336/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0852 - accuracy: 0.9970 - val_loss: 0.9449 - val_accuracy: 0.8200\n",
      "Epoch 337/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0852 - accuracy: 0.9969 - val_loss: 0.9413 - val_accuracy: 0.8212\n",
      "Epoch 338/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0855 - accuracy: 0.9969 - val_loss: 0.9416 - val_accuracy: 0.8262\n",
      "Epoch 339/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0850 - accuracy: 0.9970 - val_loss: 0.9462 - val_accuracy: 0.8200\n",
      "Epoch 340/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0843 - accuracy: 0.9978 - val_loss: 0.9503 - val_accuracy: 0.8225\n",
      "Epoch 341/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0862 - accuracy: 0.9968 - val_loss: 0.9510 - val_accuracy: 0.8162\n",
      "Epoch 342/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0853 - accuracy: 0.9970 - val_loss: 0.9399 - val_accuracy: 0.8187\n",
      "Epoch 343/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0851 - accuracy: 0.9976 - val_loss: 0.9371 - val_accuracy: 0.8175\n",
      "Epoch 344/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0861 - accuracy: 0.9968 - val_loss: 0.9338 - val_accuracy: 0.8212\n",
      "Epoch 345/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0866 - accuracy: 0.9961 - val_loss: 0.9319 - val_accuracy: 0.8275\n",
      "Epoch 346/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0859 - accuracy: 0.9972 - val_loss: 0.9292 - val_accuracy: 0.8275\n",
      "Epoch 347/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0854 - accuracy: 0.9973 - val_loss: 0.9287 - val_accuracy: 0.8250\n",
      "Epoch 348/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0866 - accuracy: 0.9964 - val_loss: 0.9238 - val_accuracy: 0.8275\n",
      "Epoch 349/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0856 - accuracy: 0.9972 - val_loss: 0.9179 - val_accuracy: 0.8300\n",
      "Epoch 350/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0856 - accuracy: 0.9967 - val_loss: 0.9241 - val_accuracy: 0.8300\n",
      "Epoch 351/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0856 - accuracy: 0.9971 - val_loss: 0.9310 - val_accuracy: 0.8300\n",
      "Epoch 352/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0848 - accuracy: 0.9974 - val_loss: 0.9304 - val_accuracy: 0.8287\n",
      "Epoch 353/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0875 - accuracy: 0.9970 - val_loss: 0.9307 - val_accuracy: 0.8275\n",
      "Epoch 354/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0858 - accuracy: 0.9976 - val_loss: 0.9260 - val_accuracy: 0.8237\n",
      "Epoch 355/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0885 - accuracy: 0.9961 - val_loss: 0.9282 - val_accuracy: 0.8250\n",
      "Epoch 356/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0867 - accuracy: 0.9976 - val_loss: 0.9377 - val_accuracy: 0.8237\n",
      "Epoch 357/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0873 - accuracy: 0.9972 - val_loss: 0.9453 - val_accuracy: 0.8200\n",
      "Epoch 358/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0865 - accuracy: 0.9970 - val_loss: 0.9452 - val_accuracy: 0.8200\n",
      "Epoch 359/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0857 - accuracy: 0.9978 - val_loss: 0.9371 - val_accuracy: 0.8212\n",
      "Epoch 360/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0865 - accuracy: 0.9976 - val_loss: 0.9373 - val_accuracy: 0.8162\n",
      "Epoch 361/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0851 - accuracy: 0.9978 - val_loss: 0.9377 - val_accuracy: 0.8187\n",
      "Epoch 362/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0858 - accuracy: 0.9978 - val_loss: 0.9408 - val_accuracy: 0.8187\n",
      "Epoch 363/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0853 - accuracy: 0.9973 - val_loss: 0.9428 - val_accuracy: 0.8162\n",
      "Epoch 364/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0845 - accuracy: 0.9978 - val_loss: 0.9427 - val_accuracy: 0.8162\n",
      "Epoch 365/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0864 - accuracy: 0.9972 - val_loss: 0.9396 - val_accuracy: 0.8175\n",
      "Epoch 366/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0861 - accuracy: 0.9976 - val_loss: 0.9463 - val_accuracy: 0.8175\n",
      "Epoch 367/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0869 - accuracy: 0.9973 - val_loss: 0.9511 - val_accuracy: 0.8200\n",
      "Epoch 368/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0863 - accuracy: 0.9972 - val_loss: 0.9556 - val_accuracy: 0.8225\n",
      "Epoch 369/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0853 - accuracy: 0.9974 - val_loss: 0.9590 - val_accuracy: 0.8187\n",
      "Epoch 370/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0856 - accuracy: 0.9974 - val_loss: 0.9544 - val_accuracy: 0.8175\n",
      "Epoch 371/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0862 - accuracy: 0.9975 - val_loss: 0.9501 - val_accuracy: 0.8162\n",
      "Epoch 372/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0860 - accuracy: 0.9973 - val_loss: 0.9492 - val_accuracy: 0.8150\n",
      "Epoch 373/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0857 - accuracy: 0.9975 - val_loss: 0.9495 - val_accuracy: 0.8187\n",
      "Epoch 374/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0864 - accuracy: 0.9976 - val_loss: 0.9506 - val_accuracy: 0.8212\n",
      "Epoch 375/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0869 - accuracy: 0.9970 - val_loss: 0.9434 - val_accuracy: 0.8162\n",
      "Epoch 376/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0872 - accuracy: 0.9972 - val_loss: 0.9503 - val_accuracy: 0.8187\n",
      "Epoch 377/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0868 - accuracy: 0.9973 - val_loss: 0.9555 - val_accuracy: 0.8200\n",
      "Epoch 378/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0881 - accuracy: 0.9973 - val_loss: 0.9600 - val_accuracy: 0.8187\n",
      "Epoch 379/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0859 - accuracy: 0.9979 - val_loss: 0.9533 - val_accuracy: 0.8225\n",
      "Epoch 380/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0882 - accuracy: 0.9970 - val_loss: 0.9518 - val_accuracy: 0.8175\n",
      "Epoch 381/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0866 - accuracy: 0.9974 - val_loss: 0.9561 - val_accuracy: 0.8200\n",
      "Epoch 382/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0862 - accuracy: 0.9974 - val_loss: 0.9606 - val_accuracy: 0.8200\n",
      "Epoch 383/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0883 - accuracy: 0.9970 - val_loss: 0.9558 - val_accuracy: 0.8175\n",
      "Epoch 384/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0865 - accuracy: 0.9978 - val_loss: 0.9604 - val_accuracy: 0.8175\n",
      "Epoch 385/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0880 - accuracy: 0.9968 - val_loss: 0.9632 - val_accuracy: 0.8250\n",
      "Epoch 386/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0876 - accuracy: 0.9970 - val_loss: 0.9727 - val_accuracy: 0.8187\n",
      "Epoch 387/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0860 - accuracy: 0.9980 - val_loss: 0.9685 - val_accuracy: 0.8175\n",
      "Epoch 388/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0873 - accuracy: 0.9973 - val_loss: 0.9619 - val_accuracy: 0.8162\n",
      "Epoch 389/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0892 - accuracy: 0.9968 - val_loss: 0.9663 - val_accuracy: 0.8162\n",
      "Epoch 390/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0872 - accuracy: 0.9976 - val_loss: 0.9655 - val_accuracy: 0.8200\n",
      "Epoch 391/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0886 - accuracy: 0.9968 - val_loss: 0.9729 - val_accuracy: 0.8212\n",
      "Epoch 392/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0884 - accuracy: 0.9970 - val_loss: 0.9791 - val_accuracy: 0.8237\n",
      "Epoch 393/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0874 - accuracy: 0.9974 - val_loss: 0.9741 - val_accuracy: 0.8200\n",
      "Epoch 394/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0871 - accuracy: 0.9979 - val_loss: 0.9768 - val_accuracy: 0.8187\n",
      "Epoch 395/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0884 - accuracy: 0.9970 - val_loss: 0.9793 - val_accuracy: 0.8200\n",
      "Epoch 396/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0875 - accuracy: 0.9972 - val_loss: 0.9778 - val_accuracy: 0.8187\n",
      "Epoch 397/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0882 - accuracy: 0.9970 - val_loss: 0.9825 - val_accuracy: 0.8200\n",
      "Epoch 398/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0874 - accuracy: 0.9973 - val_loss: 0.9799 - val_accuracy: 0.8212\n",
      "Epoch 399/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0878 - accuracy: 0.9974 - val_loss: 0.9761 - val_accuracy: 0.8212\n",
      "Epoch 400/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0895 - accuracy: 0.9971 - val_loss: 0.9699 - val_accuracy: 0.8200\n",
      "Epoch 401/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0876 - accuracy: 0.9970 - val_loss: 0.9674 - val_accuracy: 0.8250\n",
      "Epoch 402/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0862 - accuracy: 0.9980 - val_loss: 0.9696 - val_accuracy: 0.8262\n",
      "Epoch 403/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0865 - accuracy: 0.9978 - val_loss: 0.9705 - val_accuracy: 0.8237\n",
      "Epoch 404/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0883 - accuracy: 0.9976 - val_loss: 0.9670 - val_accuracy: 0.8250\n",
      "Epoch 405/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0872 - accuracy: 0.9981 - val_loss: 0.9628 - val_accuracy: 0.8250\n",
      "Epoch 406/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0877 - accuracy: 0.9976 - val_loss: 0.9484 - val_accuracy: 0.8212\n",
      "Epoch 407/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0876 - accuracy: 0.9972 - val_loss: 0.9443 - val_accuracy: 0.8225\n",
      "Epoch 408/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0856 - accuracy: 0.9980 - val_loss: 0.9468 - val_accuracy: 0.8212\n",
      "Epoch 409/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0873 - accuracy: 0.9975 - val_loss: 0.9431 - val_accuracy: 0.8237\n",
      "Epoch 410/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0879 - accuracy: 0.9970 - val_loss: 0.9388 - val_accuracy: 0.8212\n",
      "Epoch 411/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0871 - accuracy: 0.9978 - val_loss: 0.9356 - val_accuracy: 0.8212\n",
      "Epoch 412/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0877 - accuracy: 0.9976 - val_loss: 0.9409 - val_accuracy: 0.8212\n",
      "Epoch 413/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0883 - accuracy: 0.9970 - val_loss: 0.9409 - val_accuracy: 0.8212\n",
      "Epoch 414/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0885 - accuracy: 0.9965 - val_loss: 0.9368 - val_accuracy: 0.8175\n",
      "Epoch 415/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0877 - accuracy: 0.9973 - val_loss: 0.9308 - val_accuracy: 0.8187\n",
      "Epoch 416/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0875 - accuracy: 0.9978 - val_loss: 0.9398 - val_accuracy: 0.8237\n",
      "Epoch 417/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0887 - accuracy: 0.9972 - val_loss: 0.9472 - val_accuracy: 0.8187\n",
      "Epoch 418/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0868 - accuracy: 0.9977 - val_loss: 0.9511 - val_accuracy: 0.8212\n",
      "Epoch 419/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0888 - accuracy: 0.9972 - val_loss: 0.9542 - val_accuracy: 0.8187\n",
      "Epoch 420/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0890 - accuracy: 0.9974 - val_loss: 0.9512 - val_accuracy: 0.8187\n",
      "Epoch 421/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0881 - accuracy: 0.9976 - val_loss: 0.9543 - val_accuracy: 0.8187\n",
      "Epoch 422/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0895 - accuracy: 0.9971 - val_loss: 0.9584 - val_accuracy: 0.8225\n",
      "Epoch 423/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0874 - accuracy: 0.9971 - val_loss: 0.9576 - val_accuracy: 0.8237\n",
      "Epoch 424/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0887 - accuracy: 0.9974 - val_loss: 0.9595 - val_accuracy: 0.8225\n",
      "Epoch 425/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0884 - accuracy: 0.9972 - val_loss: 0.9540 - val_accuracy: 0.8237\n",
      "Epoch 426/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0887 - accuracy: 0.9971 - val_loss: 0.9526 - val_accuracy: 0.8262\n",
      "Epoch 427/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0883 - accuracy: 0.9974 - val_loss: 0.9589 - val_accuracy: 0.8262\n",
      "Epoch 428/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0901 - accuracy: 0.9966 - val_loss: 0.9604 - val_accuracy: 0.8237\n",
      "Epoch 429/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0886 - accuracy: 0.9978 - val_loss: 0.9694 - val_accuracy: 0.8225\n",
      "Epoch 430/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0897 - accuracy: 0.9970 - val_loss: 0.9713 - val_accuracy: 0.8212\n",
      "Epoch 431/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0872 - accuracy: 0.9979 - val_loss: 0.9650 - val_accuracy: 0.8250\n",
      "Epoch 432/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0890 - accuracy: 0.9978 - val_loss: 0.9645 - val_accuracy: 0.8237\n",
      "Epoch 433/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0887 - accuracy: 0.9974 - val_loss: 0.9736 - val_accuracy: 0.8312\n",
      "Epoch 434/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0878 - accuracy: 0.9980 - val_loss: 0.9698 - val_accuracy: 0.8300\n",
      "Epoch 435/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0901 - accuracy: 0.9970 - val_loss: 0.9646 - val_accuracy: 0.8275\n",
      "Epoch 436/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0893 - accuracy: 0.9973 - val_loss: 0.9655 - val_accuracy: 0.8300\n",
      "Epoch 437/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0883 - accuracy: 0.9976 - val_loss: 0.9667 - val_accuracy: 0.8275\n",
      "Epoch 438/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0877 - accuracy: 0.9979 - val_loss: 0.9695 - val_accuracy: 0.8250\n",
      "Epoch 439/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0873 - accuracy: 0.9980 - val_loss: 0.9765 - val_accuracy: 0.8237\n",
      "Epoch 440/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0894 - accuracy: 0.9969 - val_loss: 0.9701 - val_accuracy: 0.8300\n",
      "Epoch 441/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0882 - accuracy: 0.9976 - val_loss: 0.9582 - val_accuracy: 0.8312\n",
      "Epoch 442/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0889 - accuracy: 0.9972 - val_loss: 0.9514 - val_accuracy: 0.8300\n",
      "Epoch 443/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0885 - accuracy: 0.9976 - val_loss: 0.9514 - val_accuracy: 0.8287\n",
      "Epoch 444/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0885 - accuracy: 0.9973 - val_loss: 0.9506 - val_accuracy: 0.8200\n",
      "Epoch 445/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0874 - accuracy: 0.9983 - val_loss: 0.9515 - val_accuracy: 0.8275\n",
      "Epoch 446/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0892 - accuracy: 0.9973 - val_loss: 0.9607 - val_accuracy: 0.8287\n",
      "Epoch 447/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0887 - accuracy: 0.9974 - val_loss: 0.9663 - val_accuracy: 0.8287\n",
      "Epoch 448/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0887 - accuracy: 0.9974 - val_loss: 0.9677 - val_accuracy: 0.8225\n",
      "Epoch 449/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0887 - accuracy: 0.9972 - val_loss: 0.9724 - val_accuracy: 0.8250\n",
      "Epoch 450/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0882 - accuracy: 0.9976 - val_loss: 0.9750 - val_accuracy: 0.8237\n",
      "Epoch 451/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0888 - accuracy: 0.9973 - val_loss: 0.9789 - val_accuracy: 0.8212\n",
      "Epoch 452/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0881 - accuracy: 0.9980 - val_loss: 0.9790 - val_accuracy: 0.8212\n",
      "Epoch 453/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0876 - accuracy: 0.9976 - val_loss: 0.9753 - val_accuracy: 0.8275\n",
      "Epoch 454/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0888 - accuracy: 0.9974 - val_loss: 0.9731 - val_accuracy: 0.8237\n",
      "Epoch 455/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0877 - accuracy: 0.9981 - val_loss: 0.9751 - val_accuracy: 0.8237\n",
      "Epoch 456/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0909 - accuracy: 0.9968 - val_loss: 0.9785 - val_accuracy: 0.8225\n",
      "Epoch 457/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0875 - accuracy: 0.9980 - val_loss: 0.9767 - val_accuracy: 0.8187\n",
      "Epoch 458/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0880 - accuracy: 0.9978 - val_loss: 0.9807 - val_accuracy: 0.8212\n",
      "Epoch 459/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0887 - accuracy: 0.9972 - val_loss: 0.9804 - val_accuracy: 0.8187\n",
      "Epoch 460/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0899 - accuracy: 0.9974 - val_loss: 0.9777 - val_accuracy: 0.8225\n",
      "Epoch 461/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0890 - accuracy: 0.9972 - val_loss: 0.9740 - val_accuracy: 0.8262\n",
      "Epoch 462/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0888 - accuracy: 0.9974 - val_loss: 0.9783 - val_accuracy: 0.8237\n",
      "Epoch 463/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0883 - accuracy: 0.9978 - val_loss: 0.9790 - val_accuracy: 0.8200\n",
      "Epoch 464/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0886 - accuracy: 0.9971 - val_loss: 0.9918 - val_accuracy: 0.8175\n",
      "Epoch 465/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0888 - accuracy: 0.9976 - val_loss: 1.0069 - val_accuracy: 0.8200\n",
      "Epoch 466/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0891 - accuracy: 0.9974 - val_loss: 1.0044 - val_accuracy: 0.8212\n",
      "Epoch 467/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0917 - accuracy: 0.9973 - val_loss: 1.0004 - val_accuracy: 0.8225\n",
      "Epoch 468/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0888 - accuracy: 0.9980 - val_loss: 1.0012 - val_accuracy: 0.8187\n",
      "Epoch 469/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0893 - accuracy: 0.9970 - val_loss: 1.0045 - val_accuracy: 0.8187\n",
      "Epoch 470/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0895 - accuracy: 0.9973 - val_loss: 0.9983 - val_accuracy: 0.8175\n",
      "Epoch 471/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0879 - accuracy: 0.9979 - val_loss: 0.9991 - val_accuracy: 0.8187\n",
      "Epoch 472/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0884 - accuracy: 0.9977 - val_loss: 1.0009 - val_accuracy: 0.8150\n",
      "Epoch 473/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0881 - accuracy: 0.9976 - val_loss: 0.9989 - val_accuracy: 0.8150\n",
      "Epoch 474/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0897 - accuracy: 0.9970 - val_loss: 1.0012 - val_accuracy: 0.8137\n",
      "Epoch 475/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0888 - accuracy: 0.9979 - val_loss: 1.0039 - val_accuracy: 0.8150\n",
      "Epoch 476/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0909 - accuracy: 0.9974 - val_loss: 0.9972 - val_accuracy: 0.8187\n",
      "Epoch 477/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0891 - accuracy: 0.9977 - val_loss: 0.9949 - val_accuracy: 0.8175\n",
      "Epoch 478/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0896 - accuracy: 0.9972 - val_loss: 0.9996 - val_accuracy: 0.8187\n",
      "Epoch 479/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0873 - accuracy: 0.9979 - val_loss: 1.0027 - val_accuracy: 0.8175\n",
      "Epoch 480/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0888 - accuracy: 0.9973 - val_loss: 1.0040 - val_accuracy: 0.8137\n",
      "Epoch 481/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0899 - accuracy: 0.9970 - val_loss: 1.0052 - val_accuracy: 0.8175\n",
      "Epoch 482/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0893 - accuracy: 0.9976 - val_loss: 1.0074 - val_accuracy: 0.8137\n",
      "Epoch 483/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0911 - accuracy: 0.9968 - val_loss: 1.0116 - val_accuracy: 0.8137\n",
      "Epoch 484/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0887 - accuracy: 0.9982 - val_loss: 1.0042 - val_accuracy: 0.8175\n",
      "Epoch 485/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9967 - val_loss: 1.0023 - val_accuracy: 0.8200\n",
      "Epoch 486/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0882 - accuracy: 0.9982 - val_loss: 1.0016 - val_accuracy: 0.8162\n",
      "Epoch 487/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0897 - accuracy: 0.9974 - val_loss: 1.0127 - val_accuracy: 0.8187\n",
      "Epoch 488/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9976 - val_loss: 1.0149 - val_accuracy: 0.8137\n",
      "Epoch 489/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0903 - accuracy: 0.9980 - val_loss: 1.0120 - val_accuracy: 0.8125\n",
      "Epoch 490/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0905 - accuracy: 0.9970 - val_loss: 1.0101 - val_accuracy: 0.8112\n",
      "Epoch 491/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0908 - accuracy: 0.9974 - val_loss: 1.0133 - val_accuracy: 0.8112\n",
      "Epoch 492/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0901 - accuracy: 0.9970 - val_loss: 1.0176 - val_accuracy: 0.8062\n",
      "Epoch 493/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0893 - accuracy: 0.9974 - val_loss: 1.0157 - val_accuracy: 0.8112\n",
      "Epoch 494/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0906 - accuracy: 0.9968 - val_loss: 1.0019 - val_accuracy: 0.8137\n",
      "Epoch 495/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0899 - accuracy: 0.9975 - val_loss: 1.0030 - val_accuracy: 0.8150\n",
      "Epoch 496/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0887 - accuracy: 0.9980 - val_loss: 0.9985 - val_accuracy: 0.8137\n",
      "Epoch 497/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0901 - accuracy: 0.9972 - val_loss: 1.0001 - val_accuracy: 0.8137\n",
      "Epoch 498/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0905 - accuracy: 0.9969 - val_loss: 0.9972 - val_accuracy: 0.8137\n",
      "Epoch 499/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9969 - val_loss: 0.9948 - val_accuracy: 0.8137\n",
      "Epoch 500/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0901 - accuracy: 0.9975 - val_loss: 0.9935 - val_accuracy: 0.8137\n",
      "Epoch 501/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0902 - accuracy: 0.9976 - val_loss: 0.9912 - val_accuracy: 0.8137\n",
      "Epoch 502/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0908 - accuracy: 0.9978 - val_loss: 0.9922 - val_accuracy: 0.8162\n",
      "Epoch 503/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0901 - accuracy: 0.9976 - val_loss: 0.9994 - val_accuracy: 0.8137\n",
      "Epoch 504/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0909 - accuracy: 0.9971 - val_loss: 0.9971 - val_accuracy: 0.8125\n",
      "Epoch 505/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0901 - accuracy: 0.9978 - val_loss: 1.0012 - val_accuracy: 0.8175\n",
      "Epoch 506/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0896 - accuracy: 0.9975 - val_loss: 1.0074 - val_accuracy: 0.8137\n",
      "Epoch 507/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0913 - accuracy: 0.9976 - val_loss: 1.0202 - val_accuracy: 0.8125\n",
      "Epoch 508/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0895 - accuracy: 0.9978 - val_loss: 1.0221 - val_accuracy: 0.8125\n",
      "Epoch 509/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0903 - accuracy: 0.9974 - val_loss: 1.0175 - val_accuracy: 0.8187\n",
      "Epoch 510/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0905 - accuracy: 0.9972 - val_loss: 0.9972 - val_accuracy: 0.8162\n",
      "Epoch 511/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0906 - accuracy: 0.9978 - val_loss: 0.9941 - val_accuracy: 0.8125\n",
      "Epoch 512/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0889 - accuracy: 0.9981 - val_loss: 0.9845 - val_accuracy: 0.8175\n",
      "Epoch 513/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0896 - accuracy: 0.9976 - val_loss: 0.9819 - val_accuracy: 0.8162\n",
      "Epoch 514/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0890 - accuracy: 0.9980 - val_loss: 0.9873 - val_accuracy: 0.8150\n",
      "Epoch 515/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0919 - accuracy: 0.9963 - val_loss: 0.9871 - val_accuracy: 0.8150\n",
      "Epoch 516/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0894 - accuracy: 0.9981 - val_loss: 0.9870 - val_accuracy: 0.8137\n",
      "Epoch 517/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0907 - accuracy: 0.9977 - val_loss: 0.9951 - val_accuracy: 0.8137\n",
      "Epoch 518/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0909 - accuracy: 0.9974 - val_loss: 0.9939 - val_accuracy: 0.8150\n",
      "Epoch 519/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0913 - accuracy: 0.9974 - val_loss: 0.9921 - val_accuracy: 0.8125\n",
      "Epoch 520/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0909 - accuracy: 0.9978 - val_loss: 0.9906 - val_accuracy: 0.8112\n",
      "Epoch 521/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0913 - accuracy: 0.9974 - val_loss: 0.9884 - val_accuracy: 0.8137\n",
      "Epoch 522/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0918 - accuracy: 0.9975 - val_loss: 0.9859 - val_accuracy: 0.8125\n",
      "Epoch 523/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0897 - accuracy: 0.9978 - val_loss: 0.9797 - val_accuracy: 0.8100\n",
      "Epoch 524/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0912 - accuracy: 0.9971 - val_loss: 0.9784 - val_accuracy: 0.8150\n",
      "Epoch 525/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0900 - accuracy: 0.9979 - val_loss: 0.9823 - val_accuracy: 0.8100\n",
      "Epoch 526/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9971 - val_loss: 0.9850 - val_accuracy: 0.8062\n",
      "Epoch 527/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0907 - accuracy: 0.9976 - val_loss: 0.9876 - val_accuracy: 0.8062\n",
      "Epoch 528/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0901 - accuracy: 0.9980 - val_loss: 0.9838 - val_accuracy: 0.8087\n",
      "Epoch 529/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0902 - accuracy: 0.9974 - val_loss: 0.9717 - val_accuracy: 0.8087\n",
      "Epoch 530/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0926 - accuracy: 0.9964 - val_loss: 0.9808 - val_accuracy: 0.8062\n",
      "Epoch 531/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0893 - accuracy: 0.9982 - val_loss: 0.9897 - val_accuracy: 0.8075\n",
      "Epoch 532/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0907 - accuracy: 0.9976 - val_loss: 0.9908 - val_accuracy: 0.8087\n",
      "Epoch 533/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0904 - accuracy: 0.9978 - val_loss: 0.9895 - val_accuracy: 0.8075\n",
      "Epoch 534/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0929 - accuracy: 0.9972 - val_loss: 0.9836 - val_accuracy: 0.8075\n",
      "Epoch 535/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0921 - accuracy: 0.9970 - val_loss: 0.9785 - val_accuracy: 0.8075\n",
      "Epoch 536/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0913 - accuracy: 0.9973 - val_loss: 0.9807 - val_accuracy: 0.8075\n",
      "Epoch 537/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0887 - accuracy: 0.9981 - val_loss: 0.9843 - val_accuracy: 0.8062\n",
      "Epoch 538/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0911 - accuracy: 0.9982 - val_loss: 0.9832 - val_accuracy: 0.8037\n",
      "Epoch 539/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0902 - accuracy: 0.9977 - val_loss: 0.9830 - val_accuracy: 0.8087\n",
      "Epoch 540/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0930 - accuracy: 0.9974 - val_loss: 0.9859 - val_accuracy: 0.8112\n",
      "Epoch 541/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0909 - accuracy: 0.9974 - val_loss: 0.9861 - val_accuracy: 0.8125\n",
      "Epoch 542/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0902 - accuracy: 0.9980 - val_loss: 0.9955 - val_accuracy: 0.8112\n",
      "Epoch 543/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0895 - accuracy: 0.9978 - val_loss: 1.0011 - val_accuracy: 0.8087\n",
      "Epoch 544/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0907 - accuracy: 0.9975 - val_loss: 1.0008 - val_accuracy: 0.8137\n",
      "Epoch 545/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0900 - accuracy: 0.9974 - val_loss: 1.0030 - val_accuracy: 0.8125\n",
      "Epoch 546/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0909 - accuracy: 0.9973 - val_loss: 0.9941 - val_accuracy: 0.8162\n",
      "Epoch 547/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0928 - accuracy: 0.9971 - val_loss: 0.9905 - val_accuracy: 0.8150\n",
      "Epoch 548/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0902 - accuracy: 0.9978 - val_loss: 0.9981 - val_accuracy: 0.8125\n",
      "Epoch 549/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0930 - accuracy: 0.9966 - val_loss: 0.9946 - val_accuracy: 0.8137\n",
      "Epoch 550/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0912 - accuracy: 0.9976 - val_loss: 0.9883 - val_accuracy: 0.8137\n",
      "Epoch 551/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0909 - accuracy: 0.9978 - val_loss: 0.9861 - val_accuracy: 0.8112\n",
      "Epoch 552/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0927 - accuracy: 0.9970 - val_loss: 0.9881 - val_accuracy: 0.8125\n",
      "Epoch 553/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0920 - accuracy: 0.9970 - val_loss: 0.9827 - val_accuracy: 0.8075\n",
      "Epoch 554/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0909 - accuracy: 0.9975 - val_loss: 0.9797 - val_accuracy: 0.8050\n",
      "Epoch 555/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0921 - accuracy: 0.9973 - val_loss: 0.9796 - val_accuracy: 0.8050\n",
      "Epoch 556/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0916 - accuracy: 0.9974 - val_loss: 0.9871 - val_accuracy: 0.8037\n",
      "Epoch 557/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0919 - accuracy: 0.9971 - val_loss: 0.9906 - val_accuracy: 0.8100\n",
      "Epoch 558/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0915 - accuracy: 0.9976 - val_loss: 0.9925 - val_accuracy: 0.8100\n",
      "Epoch 559/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0921 - accuracy: 0.9979 - val_loss: 0.9868 - val_accuracy: 0.8050\n",
      "Epoch 560/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0916 - accuracy: 0.9979 - val_loss: 0.9831 - val_accuracy: 0.8050\n",
      "Epoch 561/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9972 - val_loss: 0.9812 - val_accuracy: 0.8062\n",
      "Epoch 562/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0926 - accuracy: 0.9969 - val_loss: 0.9832 - val_accuracy: 0.8112\n",
      "Epoch 563/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0904 - accuracy: 0.9980 - val_loss: 0.9932 - val_accuracy: 0.8125\n",
      "Epoch 564/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0925 - accuracy: 0.9973 - val_loss: 0.9847 - val_accuracy: 0.8100\n",
      "Epoch 565/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0918 - accuracy: 0.9978 - val_loss: 0.9777 - val_accuracy: 0.8075\n",
      "Epoch 566/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0919 - accuracy: 0.9974 - val_loss: 0.9910 - val_accuracy: 0.8075\n",
      "Epoch 567/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0922 - accuracy: 0.9976 - val_loss: 0.9915 - val_accuracy: 0.8075\n",
      "Epoch 568/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0904 - accuracy: 0.9980 - val_loss: 0.9914 - val_accuracy: 0.8025\n",
      "Epoch 569/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0924 - accuracy: 0.9974 - val_loss: 0.9911 - val_accuracy: 0.8050\n",
      "Epoch 570/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0916 - accuracy: 0.9976 - val_loss: 0.9925 - val_accuracy: 0.8087\n",
      "Epoch 571/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0917 - accuracy: 0.9973 - val_loss: 1.0004 - val_accuracy: 0.8062\n",
      "Epoch 572/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0915 - accuracy: 0.9979 - val_loss: 0.9988 - val_accuracy: 0.8037\n",
      "Epoch 573/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0922 - accuracy: 0.9972 - val_loss: 0.9923 - val_accuracy: 0.8062\n",
      "Epoch 574/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0910 - accuracy: 0.9978 - val_loss: 0.9891 - val_accuracy: 0.8087\n",
      "Epoch 575/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0918 - accuracy: 0.9973 - val_loss: 0.9939 - val_accuracy: 0.8050\n",
      "Epoch 576/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0921 - accuracy: 0.9972 - val_loss: 0.9902 - val_accuracy: 0.8037\n",
      "Epoch 577/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0923 - accuracy: 0.9972 - val_loss: 0.9878 - val_accuracy: 0.8050\n",
      "Epoch 578/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0921 - accuracy: 0.9974 - val_loss: 0.9919 - val_accuracy: 0.8075\n",
      "Epoch 579/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0916 - accuracy: 0.9969 - val_loss: 1.0023 - val_accuracy: 0.8000\n",
      "Epoch 580/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0908 - accuracy: 0.9976 - val_loss: 0.9981 - val_accuracy: 0.8025\n",
      "Epoch 581/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0914 - accuracy: 0.9979 - val_loss: 0.9865 - val_accuracy: 0.8050\n",
      "Epoch 582/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0906 - accuracy: 0.9976 - val_loss: 0.9765 - val_accuracy: 0.8050\n",
      "Epoch 583/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0908 - accuracy: 0.9977 - val_loss: 0.9713 - val_accuracy: 0.8075\n",
      "Epoch 584/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0912 - accuracy: 0.9975 - val_loss: 0.9690 - val_accuracy: 0.8100\n",
      "Epoch 585/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0925 - accuracy: 0.9968 - val_loss: 0.9709 - val_accuracy: 0.8112\n",
      "Epoch 586/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0917 - accuracy: 0.9978 - val_loss: 0.9662 - val_accuracy: 0.8125\n",
      "Epoch 587/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0895 - accuracy: 0.9984 - val_loss: 0.9640 - val_accuracy: 0.8075\n",
      "Epoch 588/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0902 - accuracy: 0.9980 - val_loss: 0.9677 - val_accuracy: 0.8150\n",
      "Epoch 589/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0911 - accuracy: 0.9979 - val_loss: 0.9645 - val_accuracy: 0.8100\n",
      "Epoch 590/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0907 - accuracy: 0.9978 - val_loss: 0.9592 - val_accuracy: 0.8125\n",
      "Epoch 591/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0894 - accuracy: 0.9984 - val_loss: 0.9568 - val_accuracy: 0.8150\n",
      "Epoch 592/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0900 - accuracy: 0.9983 - val_loss: 0.9553 - val_accuracy: 0.8162\n",
      "Epoch 593/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9976 - val_loss: 0.9556 - val_accuracy: 0.8175\n",
      "Epoch 594/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0901 - accuracy: 0.9976 - val_loss: 0.9525 - val_accuracy: 0.8150\n",
      "Epoch 595/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0907 - accuracy: 0.9983 - val_loss: 0.9524 - val_accuracy: 0.8187\n",
      "Epoch 596/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0891 - accuracy: 0.9982 - val_loss: 0.9568 - val_accuracy: 0.8187\n",
      "Epoch 597/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0900 - accuracy: 0.9980 - val_loss: 0.9581 - val_accuracy: 0.8200\n",
      "Epoch 598/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0907 - accuracy: 0.9976 - val_loss: 0.9560 - val_accuracy: 0.8175\n",
      "Epoch 599/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0910 - accuracy: 0.9970 - val_loss: 0.9638 - val_accuracy: 0.8150\n",
      "Epoch 600/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0894 - accuracy: 0.9980 - val_loss: 0.9754 - val_accuracy: 0.8150\n",
      "Epoch 601/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0891 - accuracy: 0.9983 - val_loss: 0.9790 - val_accuracy: 0.8137\n",
      "Epoch 602/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0900 - accuracy: 0.9975 - val_loss: 0.9699 - val_accuracy: 0.8137\n",
      "Epoch 603/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0904 - accuracy: 0.9979 - val_loss: 0.9682 - val_accuracy: 0.8137\n",
      "Epoch 604/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0910 - accuracy: 0.9977 - val_loss: 0.9686 - val_accuracy: 0.8150\n",
      "Epoch 605/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0907 - accuracy: 0.9974 - val_loss: 0.9689 - val_accuracy: 0.8187\n",
      "Epoch 606/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0908 - accuracy: 0.9978 - val_loss: 0.9670 - val_accuracy: 0.8175\n",
      "Epoch 607/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0909 - accuracy: 0.9977 - val_loss: 0.9601 - val_accuracy: 0.8125\n",
      "Epoch 608/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9977 - val_loss: 0.9595 - val_accuracy: 0.8125\n",
      "Epoch 609/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0906 - accuracy: 0.9980 - val_loss: 0.9585 - val_accuracy: 0.8125\n",
      "Epoch 610/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0923 - accuracy: 0.9973 - val_loss: 0.9704 - val_accuracy: 0.8137\n",
      "Epoch 611/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0930 - accuracy: 0.9966 - val_loss: 0.9779 - val_accuracy: 0.8112\n",
      "Epoch 612/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0900 - accuracy: 0.9976 - val_loss: 0.9756 - val_accuracy: 0.8087\n",
      "Epoch 613/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0910 - accuracy: 0.9977 - val_loss: 0.9795 - val_accuracy: 0.8100\n",
      "Epoch 614/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0910 - accuracy: 0.9978 - val_loss: 0.9790 - val_accuracy: 0.8087\n",
      "Epoch 615/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0902 - accuracy: 0.9980 - val_loss: 0.9779 - val_accuracy: 0.8125\n",
      "Epoch 616/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0915 - accuracy: 0.9973 - val_loss: 0.9732 - val_accuracy: 0.8087\n",
      "Epoch 617/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0904 - accuracy: 0.9976 - val_loss: 0.9714 - val_accuracy: 0.8075\n",
      "Epoch 618/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0921 - accuracy: 0.9970 - val_loss: 0.9647 - val_accuracy: 0.8087\n",
      "Epoch 619/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9973 - val_loss: 0.9747 - val_accuracy: 0.8112\n",
      "Epoch 620/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0901 - accuracy: 0.9980 - val_loss: 0.9915 - val_accuracy: 0.8112\n",
      "Epoch 621/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0905 - accuracy: 0.9978 - val_loss: 0.9966 - val_accuracy: 0.8125\n",
      "Epoch 622/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0911 - accuracy: 0.9974 - val_loss: 0.9958 - val_accuracy: 0.8150\n",
      "Epoch 623/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0895 - accuracy: 0.9982 - val_loss: 0.9895 - val_accuracy: 0.8087\n",
      "Epoch 624/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0911 - accuracy: 0.9974 - val_loss: 0.9784 - val_accuracy: 0.8150\n",
      "Epoch 625/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0896 - accuracy: 0.9979 - val_loss: 0.9745 - val_accuracy: 0.8150\n",
      "Epoch 626/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0900 - accuracy: 0.9984 - val_loss: 0.9748 - val_accuracy: 0.8112\n",
      "Epoch 627/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0912 - accuracy: 0.9979 - val_loss: 0.9788 - val_accuracy: 0.8150\n",
      "Epoch 628/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0899 - accuracy: 0.9982 - val_loss: 0.9852 - val_accuracy: 0.8137\n",
      "Epoch 629/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9976 - val_loss: 0.9772 - val_accuracy: 0.8200\n",
      "Epoch 630/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0916 - accuracy: 0.9974 - val_loss: 0.9777 - val_accuracy: 0.8162\n",
      "Epoch 631/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0909 - accuracy: 0.9974 - val_loss: 0.9820 - val_accuracy: 0.8137\n",
      "Epoch 632/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0895 - accuracy: 0.9978 - val_loss: 0.9899 - val_accuracy: 0.8112\n",
      "Epoch 633/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0908 - accuracy: 0.9977 - val_loss: 0.9917 - val_accuracy: 0.8125\n",
      "Epoch 634/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0896 - accuracy: 0.9982 - val_loss: 0.9946 - val_accuracy: 0.8150\n",
      "Epoch 635/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9974 - val_loss: 0.9944 - val_accuracy: 0.8150\n",
      "Epoch 636/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0927 - accuracy: 0.9970 - val_loss: 1.0038 - val_accuracy: 0.8150\n",
      "Epoch 637/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9974 - val_loss: 1.0082 - val_accuracy: 0.8112\n",
      "Epoch 638/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0899 - accuracy: 0.9980 - val_loss: 0.9997 - val_accuracy: 0.8137\n",
      "Epoch 639/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0918 - accuracy: 0.9970 - val_loss: 1.0032 - val_accuracy: 0.8162\n",
      "Epoch 640/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0912 - accuracy: 0.9974 - val_loss: 1.0078 - val_accuracy: 0.8187\n",
      "Epoch 641/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0904 - accuracy: 0.9981 - val_loss: 1.0077 - val_accuracy: 0.8187\n",
      "Epoch 642/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9972 - val_loss: 1.0061 - val_accuracy: 0.8250\n",
      "Epoch 643/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0910 - accuracy: 0.9976 - val_loss: 1.0042 - val_accuracy: 0.8262\n",
      "Epoch 644/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0927 - accuracy: 0.9971 - val_loss: 1.0055 - val_accuracy: 0.8212\n",
      "Epoch 645/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0913 - accuracy: 0.9975 - val_loss: 1.0078 - val_accuracy: 0.8200\n",
      "Epoch 646/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0923 - accuracy: 0.9976 - val_loss: 1.0039 - val_accuracy: 0.8237\n",
      "Epoch 647/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0903 - accuracy: 0.9980 - val_loss: 1.0079 - val_accuracy: 0.8225\n",
      "Epoch 648/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0940 - accuracy: 0.9970 - val_loss: 1.0095 - val_accuracy: 0.8200\n",
      "Epoch 649/1000\n",
      "19/19 [==============================] - 0s 20ms/step - loss: 0.0925 - accuracy: 0.9974 - val_loss: 0.9930 - val_accuracy: 0.8200\n",
      "Epoch 650/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0940 - accuracy: 0.9964 - val_loss: 0.9925 - val_accuracy: 0.8150\n",
      "Epoch 651/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0915 - accuracy: 0.9977 - val_loss: 0.9887 - val_accuracy: 0.8187\n",
      "Epoch 652/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0911 - accuracy: 0.9980 - val_loss: 0.9827 - val_accuracy: 0.8137\n",
      "Epoch 653/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0918 - accuracy: 0.9977 - val_loss: 0.9852 - val_accuracy: 0.8162\n",
      "Epoch 654/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0916 - accuracy: 0.9979 - val_loss: 0.9851 - val_accuracy: 0.8175\n",
      "Epoch 655/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0933 - accuracy: 0.9971 - val_loss: 0.9787 - val_accuracy: 0.8125\n",
      "Epoch 656/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0924 - accuracy: 0.9974 - val_loss: 0.9803 - val_accuracy: 0.8137\n",
      "Epoch 657/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0913 - accuracy: 0.9979 - val_loss: 0.9798 - val_accuracy: 0.8100\n",
      "Epoch 658/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0926 - accuracy: 0.9977 - val_loss: 0.9787 - val_accuracy: 0.8137\n",
      "Epoch 659/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9971 - val_loss: 0.9811 - val_accuracy: 0.8162\n",
      "Epoch 660/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0924 - accuracy: 0.9974 - val_loss: 0.9833 - val_accuracy: 0.8125\n",
      "Epoch 661/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0912 - accuracy: 0.9980 - val_loss: 0.9843 - val_accuracy: 0.8100\n",
      "Epoch 662/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0918 - accuracy: 0.9976 - val_loss: 0.9746 - val_accuracy: 0.8112\n",
      "Epoch 663/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0920 - accuracy: 0.9978 - val_loss: 0.9664 - val_accuracy: 0.8112\n",
      "Epoch 664/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0935 - accuracy: 0.9969 - val_loss: 0.9613 - val_accuracy: 0.8150\n",
      "Epoch 665/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0918 - accuracy: 0.9975 - val_loss: 0.9637 - val_accuracy: 0.8150\n",
      "Epoch 666/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0922 - accuracy: 0.9978 - val_loss: 0.9712 - val_accuracy: 0.8162\n",
      "Epoch 667/1000\n",
      "19/19 [==============================] - 0s 21ms/step - loss: 0.0916 - accuracy: 0.9977 - val_loss: 0.9725 - val_accuracy: 0.8112\n",
      "Epoch 668/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0935 - accuracy: 0.9970 - val_loss: 0.9789 - val_accuracy: 0.8150\n",
      "Epoch 669/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9974 - val_loss: 0.9805 - val_accuracy: 0.8150\n",
      "Epoch 670/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9977 - val_loss: 0.9739 - val_accuracy: 0.8125\n",
      "Epoch 671/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9979 - val_loss: 0.9660 - val_accuracy: 0.8150\n",
      "Epoch 672/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0916 - accuracy: 0.9978 - val_loss: 0.9620 - val_accuracy: 0.8162\n",
      "Epoch 673/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0928 - accuracy: 0.9974 - val_loss: 0.9647 - val_accuracy: 0.8162\n",
      "Epoch 674/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0913 - accuracy: 0.9982 - val_loss: 0.9589 - val_accuracy: 0.8225\n",
      "Epoch 675/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0906 - accuracy: 0.9984 - val_loss: 0.9516 - val_accuracy: 0.8200\n",
      "Epoch 676/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9976 - val_loss: 0.9502 - val_accuracy: 0.8162\n",
      "Epoch 677/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9976 - val_loss: 0.9500 - val_accuracy: 0.8137\n",
      "Epoch 678/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0897 - accuracy: 0.9983 - val_loss: 0.9534 - val_accuracy: 0.8150\n",
      "Epoch 679/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0916 - accuracy: 0.9974 - val_loss: 0.9530 - val_accuracy: 0.8087\n",
      "Epoch 680/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0917 - accuracy: 0.9983 - val_loss: 0.9543 - val_accuracy: 0.8112\n",
      "Epoch 681/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0923 - accuracy: 0.9974 - val_loss: 0.9529 - val_accuracy: 0.8162\n",
      "Epoch 682/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0906 - accuracy: 0.9977 - val_loss: 0.9555 - val_accuracy: 0.8175\n",
      "Epoch 683/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0904 - accuracy: 0.9982 - val_loss: 0.9476 - val_accuracy: 0.8187\n",
      "Epoch 684/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9974 - val_loss: 0.9384 - val_accuracy: 0.8212\n",
      "Epoch 685/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0908 - accuracy: 0.9980 - val_loss: 0.9426 - val_accuracy: 0.8237\n",
      "Epoch 686/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9972 - val_loss: 0.9455 - val_accuracy: 0.8262\n",
      "Epoch 687/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9979 - val_loss: 0.9433 - val_accuracy: 0.8237\n",
      "Epoch 688/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0904 - accuracy: 0.9980 - val_loss: 0.9481 - val_accuracy: 0.8262\n",
      "Epoch 689/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9979 - val_loss: 0.9549 - val_accuracy: 0.8250\n",
      "Epoch 690/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0907 - accuracy: 0.9974 - val_loss: 0.9548 - val_accuracy: 0.8237\n",
      "Epoch 691/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0910 - accuracy: 0.9974 - val_loss: 0.9511 - val_accuracy: 0.8287\n",
      "Epoch 692/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0925 - accuracy: 0.9970 - val_loss: 0.9479 - val_accuracy: 0.8287\n",
      "Epoch 693/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0906 - accuracy: 0.9980 - val_loss: 0.9443 - val_accuracy: 0.8275\n",
      "Epoch 694/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0922 - accuracy: 0.9970 - val_loss: 0.9475 - val_accuracy: 0.8262\n",
      "Epoch 695/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0930 - accuracy: 0.9969 - val_loss: 0.9465 - val_accuracy: 0.8287\n",
      "Epoch 696/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9972 - val_loss: 0.9442 - val_accuracy: 0.8312\n",
      "Epoch 697/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0899 - accuracy: 0.9986 - val_loss: 0.9435 - val_accuracy: 0.8325\n",
      "Epoch 698/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0932 - accuracy: 0.9975 - val_loss: 0.9442 - val_accuracy: 0.8250\n",
      "Epoch 699/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0910 - accuracy: 0.9982 - val_loss: 0.9428 - val_accuracy: 0.8237\n",
      "Epoch 700/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9979 - val_loss: 0.9480 - val_accuracy: 0.8262\n",
      "Epoch 701/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9978 - val_loss: 0.9543 - val_accuracy: 0.8200\n",
      "Epoch 702/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0921 - accuracy: 0.9974 - val_loss: 0.9546 - val_accuracy: 0.8225\n",
      "Epoch 703/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0908 - accuracy: 0.9982 - val_loss: 0.9545 - val_accuracy: 0.8150\n",
      "Epoch 704/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0917 - accuracy: 0.9978 - val_loss: 0.9498 - val_accuracy: 0.8175\n",
      "Epoch 705/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0902 - accuracy: 0.9982 - val_loss: 0.9397 - val_accuracy: 0.8200\n",
      "Epoch 706/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0917 - accuracy: 0.9976 - val_loss: 0.9397 - val_accuracy: 0.8175\n",
      "Epoch 707/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0918 - accuracy: 0.9972 - val_loss: 0.9362 - val_accuracy: 0.8150\n",
      "Epoch 708/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0913 - accuracy: 0.9978 - val_loss: 0.9425 - val_accuracy: 0.8200\n",
      "Epoch 709/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0909 - accuracy: 0.9981 - val_loss: 0.9382 - val_accuracy: 0.8175\n",
      "Epoch 710/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0930 - accuracy: 0.9974 - val_loss: 0.9360 - val_accuracy: 0.8200\n",
      "Epoch 711/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0909 - accuracy: 0.9982 - val_loss: 0.9436 - val_accuracy: 0.8150\n",
      "Epoch 712/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0940 - accuracy: 0.9972 - val_loss: 0.9360 - val_accuracy: 0.8087\n",
      "Epoch 713/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0907 - accuracy: 0.9980 - val_loss: 0.9394 - val_accuracy: 0.8100\n",
      "Epoch 714/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0924 - accuracy: 0.9972 - val_loss: 0.9417 - val_accuracy: 0.8100\n",
      "Epoch 715/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0909 - accuracy: 0.9978 - val_loss: 0.9412 - val_accuracy: 0.8112\n",
      "Epoch 716/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0926 - accuracy: 0.9975 - val_loss: 0.9427 - val_accuracy: 0.8075\n",
      "Epoch 717/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0915 - accuracy: 0.9976 - val_loss: 0.9407 - val_accuracy: 0.8112\n",
      "Epoch 718/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0918 - accuracy: 0.9978 - val_loss: 0.9346 - val_accuracy: 0.8137\n",
      "Epoch 719/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0933 - accuracy: 0.9973 - val_loss: 0.9281 - val_accuracy: 0.8175\n",
      "Epoch 720/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0917 - accuracy: 0.9974 - val_loss: 0.9219 - val_accuracy: 0.8162\n",
      "Epoch 721/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0912 - accuracy: 0.9977 - val_loss: 0.9112 - val_accuracy: 0.8212\n",
      "Epoch 722/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0922 - accuracy: 0.9980 - val_loss: 0.9185 - val_accuracy: 0.8237\n",
      "Epoch 723/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0918 - accuracy: 0.9976 - val_loss: 0.9204 - val_accuracy: 0.8212\n",
      "Epoch 724/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0912 - accuracy: 0.9976 - val_loss: 0.9207 - val_accuracy: 0.8162\n",
      "Epoch 725/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9969 - val_loss: 0.9178 - val_accuracy: 0.8187\n",
      "Epoch 726/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0906 - accuracy: 0.9982 - val_loss: 0.9191 - val_accuracy: 0.8200\n",
      "Epoch 727/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0919 - accuracy: 0.9976 - val_loss: 0.9212 - val_accuracy: 0.8275\n",
      "Epoch 728/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0915 - accuracy: 0.9976 - val_loss: 0.9193 - val_accuracy: 0.8287\n",
      "Epoch 729/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0907 - accuracy: 0.9978 - val_loss: 0.9143 - val_accuracy: 0.8212\n",
      "Epoch 730/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0918 - accuracy: 0.9978 - val_loss: 0.9107 - val_accuracy: 0.8237\n",
      "Epoch 731/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9976 - val_loss: 0.9101 - val_accuracy: 0.8187\n",
      "Epoch 732/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0920 - accuracy: 0.9977 - val_loss: 0.9105 - val_accuracy: 0.8225\n",
      "Epoch 733/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0908 - accuracy: 0.9980 - val_loss: 0.9029 - val_accuracy: 0.8212\n",
      "Epoch 734/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0913 - accuracy: 0.9979 - val_loss: 0.8992 - val_accuracy: 0.8237\n",
      "Epoch 735/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0917 - accuracy: 0.9978 - val_loss: 0.9050 - val_accuracy: 0.8200\n",
      "Epoch 736/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0925 - accuracy: 0.9974 - val_loss: 0.9128 - val_accuracy: 0.8150\n",
      "Epoch 737/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0937 - accuracy: 0.9974 - val_loss: 0.9167 - val_accuracy: 0.8125\n",
      "Epoch 738/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0940 - accuracy: 0.9970 - val_loss: 0.9202 - val_accuracy: 0.8137\n",
      "Epoch 739/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0925 - accuracy: 0.9978 - val_loss: 0.9179 - val_accuracy: 0.8150\n",
      "Epoch 740/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0908 - accuracy: 0.9981 - val_loss: 0.9252 - val_accuracy: 0.8150\n",
      "Epoch 741/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0941 - accuracy: 0.9967 - val_loss: 0.9309 - val_accuracy: 0.8137\n",
      "Epoch 742/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0931 - accuracy: 0.9972 - val_loss: 0.9314 - val_accuracy: 0.8187\n",
      "Epoch 743/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0920 - accuracy: 0.9981 - val_loss: 0.9261 - val_accuracy: 0.8162\n",
      "Epoch 744/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0925 - accuracy: 0.9978 - val_loss: 0.9208 - val_accuracy: 0.8212\n",
      "Epoch 745/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9975 - val_loss: 0.9342 - val_accuracy: 0.8225\n",
      "Epoch 746/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9971 - val_loss: 0.9440 - val_accuracy: 0.8237\n",
      "Epoch 747/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0920 - accuracy: 0.9980 - val_loss: 0.9519 - val_accuracy: 0.8162\n",
      "Epoch 748/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0930 - accuracy: 0.9970 - val_loss: 0.9570 - val_accuracy: 0.8175\n",
      "Epoch 749/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0935 - accuracy: 0.9971 - val_loss: 0.9542 - val_accuracy: 0.8162\n",
      "Epoch 750/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0911 - accuracy: 0.9975 - val_loss: 0.9505 - val_accuracy: 0.8150\n",
      "Epoch 751/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0925 - accuracy: 0.9977 - val_loss: 0.9421 - val_accuracy: 0.8225\n",
      "Epoch 752/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0924 - accuracy: 0.9978 - val_loss: 0.9414 - val_accuracy: 0.8237\n",
      "Epoch 753/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0926 - accuracy: 0.9975 - val_loss: 0.9547 - val_accuracy: 0.8237\n",
      "Epoch 754/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0950 - accuracy: 0.9974 - val_loss: 0.9558 - val_accuracy: 0.8187\n",
      "Epoch 755/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0922 - accuracy: 0.9975 - val_loss: 0.9432 - val_accuracy: 0.8212\n",
      "Epoch 756/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0919 - accuracy: 0.9976 - val_loss: 0.9363 - val_accuracy: 0.8200\n",
      "Epoch 757/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0923 - accuracy: 0.9978 - val_loss: 0.9318 - val_accuracy: 0.8212\n",
      "Epoch 758/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0936 - accuracy: 0.9975 - val_loss: 0.9350 - val_accuracy: 0.8225\n",
      "Epoch 759/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0941 - accuracy: 0.9976 - val_loss: 0.9278 - val_accuracy: 0.8212\n",
      "Epoch 760/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0930 - accuracy: 0.9977 - val_loss: 0.9268 - val_accuracy: 0.8200\n",
      "Epoch 761/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0928 - accuracy: 0.9976 - val_loss: 0.9293 - val_accuracy: 0.8150\n",
      "Epoch 762/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0932 - accuracy: 0.9974 - val_loss: 0.9258 - val_accuracy: 0.8175\n",
      "Epoch 763/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0910 - accuracy: 0.9982 - val_loss: 0.9202 - val_accuracy: 0.8150\n",
      "Epoch 764/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0942 - accuracy: 0.9974 - val_loss: 0.9242 - val_accuracy: 0.8187\n",
      "Epoch 765/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0925 - accuracy: 0.9981 - val_loss: 0.9302 - val_accuracy: 0.8175\n",
      "Epoch 766/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9977 - val_loss: 0.9319 - val_accuracy: 0.8175\n",
      "Epoch 767/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0946 - accuracy: 0.9972 - val_loss: 0.9261 - val_accuracy: 0.8200\n",
      "Epoch 768/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0928 - accuracy: 0.9978 - val_loss: 0.9289 - val_accuracy: 0.8212\n",
      "Epoch 769/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0931 - accuracy: 0.9976 - val_loss: 0.9315 - val_accuracy: 0.8175\n",
      "Epoch 770/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0939 - accuracy: 0.9974 - val_loss: 0.9421 - val_accuracy: 0.8200\n",
      "Epoch 771/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0956 - accuracy: 0.9972 - val_loss: 0.9425 - val_accuracy: 0.8150\n",
      "Epoch 772/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0920 - accuracy: 0.9979 - val_loss: 0.9422 - val_accuracy: 0.8125\n",
      "Epoch 773/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9984 - val_loss: 0.9339 - val_accuracy: 0.8162\n",
      "Epoch 774/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0929 - accuracy: 0.9976 - val_loss: 0.9349 - val_accuracy: 0.8125\n",
      "Epoch 775/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0921 - accuracy: 0.9978 - val_loss: 0.9390 - val_accuracy: 0.8137\n",
      "Epoch 776/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0922 - accuracy: 0.9980 - val_loss: 0.9316 - val_accuracy: 0.8162\n",
      "Epoch 777/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0932 - accuracy: 0.9976 - val_loss: 0.9294 - val_accuracy: 0.8162\n",
      "Epoch 778/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0922 - accuracy: 0.9980 - val_loss: 0.9252 - val_accuracy: 0.8162\n",
      "Epoch 779/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0941 - accuracy: 0.9976 - val_loss: 0.9154 - val_accuracy: 0.8137\n",
      "Epoch 780/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0937 - accuracy: 0.9975 - val_loss: 0.9090 - val_accuracy: 0.8162\n",
      "Epoch 781/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0943 - accuracy: 0.9974 - val_loss: 0.9032 - val_accuracy: 0.8150\n",
      "Epoch 782/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0929 - accuracy: 0.9982 - val_loss: 0.8969 - val_accuracy: 0.8187\n",
      "Epoch 783/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0927 - accuracy: 0.9975 - val_loss: 0.8921 - val_accuracy: 0.8187\n",
      "Epoch 784/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0932 - accuracy: 0.9978 - val_loss: 0.8896 - val_accuracy: 0.8187\n",
      "Epoch 785/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0926 - accuracy: 0.9978 - val_loss: 0.8926 - val_accuracy: 0.8212\n",
      "Epoch 786/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0924 - accuracy: 0.9975 - val_loss: 0.8962 - val_accuracy: 0.8200\n",
      "Epoch 787/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0915 - accuracy: 0.9980 - val_loss: 0.9028 - val_accuracy: 0.8187\n",
      "Epoch 788/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0939 - accuracy: 0.9974 - val_loss: 0.9033 - val_accuracy: 0.8237\n",
      "Epoch 789/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0938 - accuracy: 0.9973 - val_loss: 0.9001 - val_accuracy: 0.8250\n",
      "Epoch 790/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0948 - accuracy: 0.9967 - val_loss: 0.9051 - val_accuracy: 0.8225\n",
      "Epoch 791/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0942 - accuracy: 0.9972 - val_loss: 0.9135 - val_accuracy: 0.8162\n",
      "Epoch 792/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0924 - accuracy: 0.9984 - val_loss: 0.9214 - val_accuracy: 0.8162\n",
      "Epoch 793/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0924 - accuracy: 0.9980 - val_loss: 0.9182 - val_accuracy: 0.8150\n",
      "Epoch 794/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0934 - accuracy: 0.9977 - val_loss: 0.9209 - val_accuracy: 0.8150\n",
      "Epoch 795/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0918 - accuracy: 0.9980 - val_loss: 0.9202 - val_accuracy: 0.8175\n",
      "Epoch 796/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0921 - accuracy: 0.9982 - val_loss: 0.9176 - val_accuracy: 0.8212\n",
      "Epoch 797/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0934 - accuracy: 0.9978 - val_loss: 0.9204 - val_accuracy: 0.8200\n",
      "Epoch 798/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0928 - accuracy: 0.9978 - val_loss: 0.9210 - val_accuracy: 0.8187\n",
      "Epoch 799/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0932 - accuracy: 0.9975 - val_loss: 0.9212 - val_accuracy: 0.8187\n",
      "Epoch 800/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0918 - accuracy: 0.9978 - val_loss: 0.9278 - val_accuracy: 0.8200\n",
      "Epoch 801/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0919 - accuracy: 0.9982 - val_loss: 0.9265 - val_accuracy: 0.8212\n",
      "Epoch 802/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0914 - accuracy: 0.9982 - val_loss: 0.9271 - val_accuracy: 0.8250\n",
      "Epoch 803/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0928 - accuracy: 0.9978 - val_loss: 0.9262 - val_accuracy: 0.8250\n",
      "Epoch 804/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0934 - accuracy: 0.9972 - val_loss: 0.9281 - val_accuracy: 0.8237\n",
      "Epoch 805/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0923 - accuracy: 0.9975 - val_loss: 0.9284 - val_accuracy: 0.8237\n",
      "Epoch 806/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0914 - accuracy: 0.9979 - val_loss: 0.9334 - val_accuracy: 0.8262\n",
      "Epoch 807/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0915 - accuracy: 0.9980 - val_loss: 0.9414 - val_accuracy: 0.8187\n",
      "Epoch 808/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0922 - accuracy: 0.9977 - val_loss: 0.9400 - val_accuracy: 0.8250\n",
      "Epoch 809/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0918 - accuracy: 0.9979 - val_loss: 0.9277 - val_accuracy: 0.8187\n",
      "Epoch 810/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0916 - accuracy: 0.9976 - val_loss: 0.9243 - val_accuracy: 0.8250\n",
      "Epoch 811/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0907 - accuracy: 0.9983 - val_loss: 0.9196 - val_accuracy: 0.8212\n",
      "Epoch 812/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0937 - accuracy: 0.9976 - val_loss: 0.9147 - val_accuracy: 0.8212\n",
      "Epoch 813/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0911 - accuracy: 0.9977 - val_loss: 0.9126 - val_accuracy: 0.8250\n",
      "Epoch 814/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0918 - accuracy: 0.9984 - val_loss: 0.9151 - val_accuracy: 0.8237\n",
      "Epoch 815/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0906 - accuracy: 0.9982 - val_loss: 0.9238 - val_accuracy: 0.8200\n",
      "Epoch 816/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0920 - accuracy: 0.9972 - val_loss: 0.9216 - val_accuracy: 0.8225\n",
      "Epoch 817/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9982 - val_loss: 0.9119 - val_accuracy: 0.8137\n",
      "Epoch 818/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0913 - accuracy: 0.9980 - val_loss: 0.9110 - val_accuracy: 0.8237\n",
      "Epoch 819/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0909 - accuracy: 0.9978 - val_loss: 0.9130 - val_accuracy: 0.8237\n",
      "Epoch 820/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0909 - accuracy: 0.9977 - val_loss: 0.9130 - val_accuracy: 0.8250\n",
      "Epoch 821/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0921 - accuracy: 0.9977 - val_loss: 0.9005 - val_accuracy: 0.8287\n",
      "Epoch 822/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0919 - accuracy: 0.9976 - val_loss: 0.9008 - val_accuracy: 0.8262\n",
      "Epoch 823/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0916 - accuracy: 0.9982 - val_loss: 0.9045 - val_accuracy: 0.8275\n",
      "Epoch 824/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0928 - accuracy: 0.9974 - val_loss: 0.9102 - val_accuracy: 0.8212\n",
      "Epoch 825/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0915 - accuracy: 0.9978 - val_loss: 0.9150 - val_accuracy: 0.8300\n",
      "Epoch 826/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0920 - accuracy: 0.9973 - val_loss: 0.9178 - val_accuracy: 0.8275\n",
      "Epoch 827/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0919 - accuracy: 0.9978 - val_loss: 0.9222 - val_accuracy: 0.8250\n",
      "Epoch 828/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0914 - accuracy: 0.9976 - val_loss: 0.9199 - val_accuracy: 0.8175\n",
      "Epoch 829/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0931 - accuracy: 0.9972 - val_loss: 0.9275 - val_accuracy: 0.8200\n",
      "Epoch 830/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0924 - accuracy: 0.9980 - val_loss: 0.9248 - val_accuracy: 0.8212\n",
      "Epoch 831/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0911 - accuracy: 0.9977 - val_loss: 0.9238 - val_accuracy: 0.8212\n",
      "Epoch 832/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0928 - accuracy: 0.9978 - val_loss: 0.9284 - val_accuracy: 0.8187\n",
      "Epoch 833/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0912 - accuracy: 0.9983 - val_loss: 0.9342 - val_accuracy: 0.8175\n",
      "Epoch 834/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0921 - accuracy: 0.9978 - val_loss: 0.9262 - val_accuracy: 0.8162\n",
      "Epoch 835/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0926 - accuracy: 0.9974 - val_loss: 0.9291 - val_accuracy: 0.8175\n",
      "Epoch 836/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0916 - accuracy: 0.9979 - val_loss: 0.9304 - val_accuracy: 0.8187\n",
      "Epoch 837/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0918 - accuracy: 0.9973 - val_loss: 0.9325 - val_accuracy: 0.8225\n",
      "Epoch 838/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0922 - accuracy: 0.9979 - val_loss: 0.9302 - val_accuracy: 0.8212\n",
      "Epoch 839/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0934 - accuracy: 0.9976 - val_loss: 0.9264 - val_accuracy: 0.8150\n",
      "Epoch 840/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0915 - accuracy: 0.9982 - val_loss: 0.9261 - val_accuracy: 0.8150\n",
      "Epoch 841/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0917 - accuracy: 0.9979 - val_loss: 0.9245 - val_accuracy: 0.8112\n",
      "Epoch 842/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0936 - accuracy: 0.9972 - val_loss: 0.9279 - val_accuracy: 0.8162\n",
      "Epoch 843/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0928 - accuracy: 0.9978 - val_loss: 0.9278 - val_accuracy: 0.8162\n",
      "Epoch 844/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0930 - accuracy: 0.9974 - val_loss: 0.9327 - val_accuracy: 0.8175\n",
      "Epoch 845/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0939 - accuracy: 0.9974 - val_loss: 0.9401 - val_accuracy: 0.8112\n",
      "Epoch 846/1000\n",
      "19/19 [==============================] - 0s 20ms/step - loss: 0.0926 - accuracy: 0.9979 - val_loss: 0.9504 - val_accuracy: 0.8137\n",
      "Epoch 847/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0935 - accuracy: 0.9973 - val_loss: 0.9480 - val_accuracy: 0.8137\n",
      "Epoch 848/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0925 - accuracy: 0.9976 - val_loss: 0.9427 - val_accuracy: 0.8187\n",
      "Epoch 849/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0952 - accuracy: 0.9972 - val_loss: 0.9390 - val_accuracy: 0.8137\n",
      "Epoch 850/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0954 - accuracy: 0.9972 - val_loss: 0.9406 - val_accuracy: 0.8125\n",
      "Epoch 851/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0945 - accuracy: 0.9980 - val_loss: 0.9395 - val_accuracy: 0.8112\n",
      "Epoch 852/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0939 - accuracy: 0.9978 - val_loss: 0.9390 - val_accuracy: 0.8162\n",
      "Epoch 853/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0927 - accuracy: 0.9978 - val_loss: 0.9413 - val_accuracy: 0.8187\n",
      "Epoch 854/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0959 - accuracy: 0.9977 - val_loss: 0.9435 - val_accuracy: 0.8162\n",
      "Epoch 855/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0932 - accuracy: 0.9977 - val_loss: 0.9401 - val_accuracy: 0.8187\n",
      "Epoch 856/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0940 - accuracy: 0.9975 - val_loss: 0.9249 - val_accuracy: 0.8187\n",
      "Epoch 857/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0931 - accuracy: 0.9978 - val_loss: 0.9202 - val_accuracy: 0.8212\n",
      "Epoch 858/1000\n",
      "19/19 [==============================] - 0s 21ms/step - loss: 0.0930 - accuracy: 0.9980 - val_loss: 0.9255 - val_accuracy: 0.8200\n",
      "Epoch 859/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0934 - accuracy: 0.9980 - val_loss: 0.9215 - val_accuracy: 0.8187\n",
      "Epoch 860/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0937 - accuracy: 0.9976 - val_loss: 0.9265 - val_accuracy: 0.8175\n",
      "Epoch 861/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0935 - accuracy: 0.9981 - val_loss: 0.9286 - val_accuracy: 0.8175\n",
      "Epoch 862/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0931 - accuracy: 0.9979 - val_loss: 0.9351 - val_accuracy: 0.8137\n",
      "Epoch 863/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0930 - accuracy: 0.9980 - val_loss: 0.9506 - val_accuracy: 0.8137\n",
      "Epoch 864/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0938 - accuracy: 0.9978 - val_loss: 0.9500 - val_accuracy: 0.8150\n",
      "Epoch 865/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0934 - accuracy: 0.9978 - val_loss: 0.9472 - val_accuracy: 0.8187\n",
      "Epoch 866/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0941 - accuracy: 0.9975 - val_loss: 0.9572 - val_accuracy: 0.8187\n",
      "Epoch 867/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0924 - accuracy: 0.9983 - val_loss: 0.9592 - val_accuracy: 0.8150\n",
      "Epoch 868/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0935 - accuracy: 0.9977 - val_loss: 0.9545 - val_accuracy: 0.8137\n",
      "Epoch 869/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0930 - accuracy: 0.9974 - val_loss: 0.9579 - val_accuracy: 0.8212\n",
      "Epoch 870/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0934 - accuracy: 0.9976 - val_loss: 0.9660 - val_accuracy: 0.8112\n",
      "Epoch 871/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0939 - accuracy: 0.9974 - val_loss: 0.9717 - val_accuracy: 0.8162\n",
      "Epoch 872/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0926 - accuracy: 0.9982 - val_loss: 0.9619 - val_accuracy: 0.8112\n",
      "Epoch 873/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0912 - accuracy: 0.9984 - val_loss: 0.9617 - val_accuracy: 0.8175\n",
      "Epoch 874/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0926 - accuracy: 0.9977 - val_loss: 0.9536 - val_accuracy: 0.8225\n",
      "Epoch 875/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0906 - accuracy: 0.9986 - val_loss: 0.9561 - val_accuracy: 0.8200\n",
      "Epoch 876/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0926 - accuracy: 0.9978 - val_loss: 0.9550 - val_accuracy: 0.8187\n",
      "Epoch 877/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0922 - accuracy: 0.9980 - val_loss: 0.9531 - val_accuracy: 0.8175\n",
      "Epoch 878/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0916 - accuracy: 0.9981 - val_loss: 0.9502 - val_accuracy: 0.8175\n",
      "Epoch 879/1000\n",
      "19/19 [==============================] - 0s 20ms/step - loss: 0.0930 - accuracy: 0.9974 - val_loss: 0.9409 - val_accuracy: 0.8162\n",
      "Epoch 880/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0927 - accuracy: 0.9973 - val_loss: 0.9378 - val_accuracy: 0.8200\n",
      "Epoch 881/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0935 - accuracy: 0.9978 - val_loss: 0.9403 - val_accuracy: 0.8212\n",
      "Epoch 882/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9976 - val_loss: 0.9408 - val_accuracy: 0.8200\n",
      "Epoch 883/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0923 - accuracy: 0.9977 - val_loss: 0.9463 - val_accuracy: 0.8237\n",
      "Epoch 884/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0917 - accuracy: 0.9980 - val_loss: 0.9546 - val_accuracy: 0.8262\n",
      "Epoch 885/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0915 - accuracy: 0.9976 - val_loss: 0.9516 - val_accuracy: 0.8237\n",
      "Epoch 886/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0917 - accuracy: 0.9977 - val_loss: 0.9527 - val_accuracy: 0.8212\n",
      "Epoch 887/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0934 - accuracy: 0.9975 - val_loss: 0.9521 - val_accuracy: 0.8262\n",
      "Epoch 888/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0916 - accuracy: 0.9983 - val_loss: 0.9438 - val_accuracy: 0.8262\n",
      "Epoch 889/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9974 - val_loss: 0.9398 - val_accuracy: 0.8250\n",
      "Epoch 890/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0924 - accuracy: 0.9976 - val_loss: 0.9443 - val_accuracy: 0.8237\n",
      "Epoch 891/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0916 - accuracy: 0.9982 - val_loss: 0.9423 - val_accuracy: 0.8212\n",
      "Epoch 892/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0928 - accuracy: 0.9978 - val_loss: 0.9514 - val_accuracy: 0.8237\n",
      "Epoch 893/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0928 - accuracy: 0.9970 - val_loss: 0.9486 - val_accuracy: 0.8237\n",
      "Epoch 894/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0930 - accuracy: 0.9980 - val_loss: 0.9586 - val_accuracy: 0.8237\n",
      "Epoch 895/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0941 - accuracy: 0.9978 - val_loss: 0.9663 - val_accuracy: 0.8237\n",
      "Epoch 896/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0943 - accuracy: 0.9969 - val_loss: 0.9567 - val_accuracy: 0.8287\n",
      "Epoch 897/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0927 - accuracy: 0.9979 - val_loss: 0.9670 - val_accuracy: 0.8225\n",
      "Epoch 898/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0924 - accuracy: 0.9980 - val_loss: 0.9736 - val_accuracy: 0.8212\n",
      "Epoch 899/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0916 - accuracy: 0.9980 - val_loss: 0.9670 - val_accuracy: 0.8212\n",
      "Epoch 900/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0928 - accuracy: 0.9978 - val_loss: 0.9662 - val_accuracy: 0.8200\n",
      "Epoch 901/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0933 - accuracy: 0.9973 - val_loss: 0.9596 - val_accuracy: 0.8200\n",
      "Epoch 902/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9974 - val_loss: 0.9613 - val_accuracy: 0.8137\n",
      "Epoch 903/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0936 - accuracy: 0.9972 - val_loss: 0.9516 - val_accuracy: 0.8175\n",
      "Epoch 904/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0920 - accuracy: 0.9979 - val_loss: 0.9458 - val_accuracy: 0.8162\n",
      "Epoch 905/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0932 - accuracy: 0.9976 - val_loss: 0.9429 - val_accuracy: 0.8237\n",
      "Epoch 906/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0929 - accuracy: 0.9977 - val_loss: 0.9484 - val_accuracy: 0.8200\n",
      "Epoch 907/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0920 - accuracy: 0.9978 - val_loss: 0.9552 - val_accuracy: 0.8175\n",
      "Epoch 908/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0929 - accuracy: 0.9977 - val_loss: 0.9695 - val_accuracy: 0.8150\n",
      "Epoch 909/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9978 - val_loss: 0.9662 - val_accuracy: 0.8112\n",
      "Epoch 910/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9976 - val_loss: 0.9576 - val_accuracy: 0.8137\n",
      "Epoch 911/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0920 - accuracy: 0.9982 - val_loss: 0.9511 - val_accuracy: 0.8112\n",
      "Epoch 912/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9977 - val_loss: 0.9502 - val_accuracy: 0.8175\n",
      "Epoch 913/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0908 - accuracy: 0.9980 - val_loss: 0.9548 - val_accuracy: 0.8112\n",
      "Epoch 914/1000\n",
      "19/19 [==============================] - 0s 18ms/step - loss: 0.0928 - accuracy: 0.9981 - val_loss: 0.9608 - val_accuracy: 0.8075\n",
      "Epoch 915/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9981 - val_loss: 0.9685 - val_accuracy: 0.8062\n",
      "Epoch 916/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0922 - accuracy: 0.9976 - val_loss: 0.9643 - val_accuracy: 0.8087\n",
      "Epoch 917/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0934 - accuracy: 0.9973 - val_loss: 0.9614 - val_accuracy: 0.8137\n",
      "Epoch 918/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0952 - accuracy: 0.9968 - val_loss: 0.9578 - val_accuracy: 0.8162\n",
      "Epoch 919/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0932 - accuracy: 0.9978 - val_loss: 0.9534 - val_accuracy: 0.8150\n",
      "Epoch 920/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0935 - accuracy: 0.9977 - val_loss: 0.9530 - val_accuracy: 0.8187\n",
      "Epoch 921/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0922 - accuracy: 0.9982 - val_loss: 0.9543 - val_accuracy: 0.8150\n",
      "Epoch 922/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0931 - accuracy: 0.9976 - val_loss: 0.9544 - val_accuracy: 0.8137\n",
      "Epoch 923/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0930 - accuracy: 0.9976 - val_loss: 0.9557 - val_accuracy: 0.8087\n",
      "Epoch 924/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0938 - accuracy: 0.9976 - val_loss: 0.9527 - val_accuracy: 0.8062\n",
      "Epoch 925/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0947 - accuracy: 0.9976 - val_loss: 0.9497 - val_accuracy: 0.8137\n",
      "Epoch 926/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0916 - accuracy: 0.9985 - val_loss: 0.9435 - val_accuracy: 0.8150\n",
      "Epoch 927/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0937 - accuracy: 0.9977 - val_loss: 0.9406 - val_accuracy: 0.8150\n",
      "Epoch 928/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0914 - accuracy: 0.9982 - val_loss: 0.9417 - val_accuracy: 0.8175\n",
      "Epoch 929/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0924 - accuracy: 0.9976 - val_loss: 0.9500 - val_accuracy: 0.8150\n",
      "Epoch 930/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0931 - accuracy: 0.9979 - val_loss: 0.9562 - val_accuracy: 0.8075\n",
      "Epoch 931/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0920 - accuracy: 0.9983 - val_loss: 0.9543 - val_accuracy: 0.8100\n",
      "Epoch 932/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0932 - accuracy: 0.9977 - val_loss: 0.9546 - val_accuracy: 0.8087\n",
      "Epoch 933/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0929 - accuracy: 0.9979 - val_loss: 0.9570 - val_accuracy: 0.8100\n",
      "Epoch 934/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0927 - accuracy: 0.9973 - val_loss: 0.9563 - val_accuracy: 0.8075\n",
      "Epoch 935/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0938 - accuracy: 0.9970 - val_loss: 0.9618 - val_accuracy: 0.8062\n",
      "Epoch 936/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0956 - accuracy: 0.9972 - val_loss: 0.9688 - val_accuracy: 0.8050\n",
      "Epoch 937/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0941 - accuracy: 0.9979 - val_loss: 0.9735 - val_accuracy: 0.8112\n",
      "Epoch 938/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0928 - accuracy: 0.9976 - val_loss: 0.9716 - val_accuracy: 0.8100\n",
      "Epoch 939/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0937 - accuracy: 0.9972 - val_loss: 0.9687 - val_accuracy: 0.8125\n",
      "Epoch 940/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0953 - accuracy: 0.9969 - val_loss: 0.9690 - val_accuracy: 0.8100\n",
      "Epoch 941/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0937 - accuracy: 0.9975 - val_loss: 0.9700 - val_accuracy: 0.8125\n",
      "Epoch 942/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0941 - accuracy: 0.9975 - val_loss: 0.9632 - val_accuracy: 0.8125\n",
      "Epoch 943/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0939 - accuracy: 0.9980 - val_loss: 0.9647 - val_accuracy: 0.8100\n",
      "Epoch 944/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0936 - accuracy: 0.9980 - val_loss: 0.9657 - val_accuracy: 0.8087\n",
      "Epoch 945/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0957 - accuracy: 0.9968 - val_loss: 0.9646 - val_accuracy: 0.8100\n",
      "Epoch 946/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0950 - accuracy: 0.9976 - val_loss: 0.9623 - val_accuracy: 0.8125\n",
      "Epoch 947/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0941 - accuracy: 0.9983 - val_loss: 0.9636 - val_accuracy: 0.8100\n",
      "Epoch 948/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9985 - val_loss: 0.9628 - val_accuracy: 0.8075\n",
      "Epoch 949/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0938 - accuracy: 0.9976 - val_loss: 0.9652 - val_accuracy: 0.8125\n",
      "Epoch 950/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0942 - accuracy: 0.9974 - val_loss: 0.9597 - val_accuracy: 0.8087\n",
      "Epoch 951/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0932 - accuracy: 0.9977 - val_loss: 0.9595 - val_accuracy: 0.8062\n",
      "Epoch 952/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0925 - accuracy: 0.9978 - val_loss: 0.9659 - val_accuracy: 0.8037\n",
      "Epoch 953/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0936 - accuracy: 0.9980 - val_loss: 0.9647 - val_accuracy: 0.7987\n",
      "Epoch 954/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0917 - accuracy: 0.9983 - val_loss: 0.9691 - val_accuracy: 0.7987\n",
      "Epoch 955/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0946 - accuracy: 0.9974 - val_loss: 0.9672 - val_accuracy: 0.7962\n",
      "Epoch 956/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0929 - accuracy: 0.9976 - val_loss: 0.9572 - val_accuracy: 0.7987\n",
      "Epoch 957/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0932 - accuracy: 0.9977 - val_loss: 0.9538 - val_accuracy: 0.8062\n",
      "Epoch 958/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0938 - accuracy: 0.9974 - val_loss: 0.9649 - val_accuracy: 0.8037\n",
      "Epoch 959/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0939 - accuracy: 0.9976 - val_loss: 0.9733 - val_accuracy: 0.8062\n",
      "Epoch 960/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0948 - accuracy: 0.9972 - val_loss: 0.9742 - val_accuracy: 0.8087\n",
      "Epoch 961/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0924 - accuracy: 0.9978 - val_loss: 0.9791 - val_accuracy: 0.8100\n",
      "Epoch 962/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0934 - accuracy: 0.9976 - val_loss: 0.9778 - val_accuracy: 0.8012\n",
      "Epoch 963/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0935 - accuracy: 0.9980 - val_loss: 0.9838 - val_accuracy: 0.8075\n",
      "Epoch 964/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0935 - accuracy: 0.9977 - val_loss: 0.9961 - val_accuracy: 0.8050\n",
      "Epoch 965/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0933 - accuracy: 0.9982 - val_loss: 0.9928 - val_accuracy: 0.8087\n",
      "Epoch 966/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0939 - accuracy: 0.9978 - val_loss: 1.0010 - val_accuracy: 0.8062\n",
      "Epoch 967/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0936 - accuracy: 0.9975 - val_loss: 0.9942 - val_accuracy: 0.8075\n",
      "Epoch 968/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0935 - accuracy: 0.9975 - val_loss: 0.9859 - val_accuracy: 0.8100\n",
      "Epoch 969/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0945 - accuracy: 0.9975 - val_loss: 0.9781 - val_accuracy: 0.8050\n",
      "Epoch 970/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0971 - accuracy: 0.9973 - val_loss: 0.9725 - val_accuracy: 0.7975\n",
      "Epoch 971/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0940 - accuracy: 0.9979 - val_loss: 0.9733 - val_accuracy: 0.8000\n",
      "Epoch 972/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0935 - accuracy: 0.9980 - val_loss: 0.9731 - val_accuracy: 0.8037\n",
      "Epoch 973/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0939 - accuracy: 0.9974 - val_loss: 0.9753 - val_accuracy: 0.8050\n",
      "Epoch 974/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0932 - accuracy: 0.9980 - val_loss: 0.9685 - val_accuracy: 0.8075\n",
      "Epoch 975/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0942 - accuracy: 0.9978 - val_loss: 0.9714 - val_accuracy: 0.8100\n",
      "Epoch 976/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0930 - accuracy: 0.9979 - val_loss: 0.9669 - val_accuracy: 0.8025\n",
      "Epoch 977/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0931 - accuracy: 0.9979 - val_loss: 0.9618 - val_accuracy: 0.8087\n",
      "Epoch 978/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0944 - accuracy: 0.9977 - val_loss: 0.9651 - val_accuracy: 0.8112\n",
      "Epoch 979/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9978 - val_loss: 0.9651 - val_accuracy: 0.8087\n",
      "Epoch 980/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0923 - accuracy: 0.9981 - val_loss: 0.9596 - val_accuracy: 0.8137\n",
      "Epoch 981/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9980 - val_loss: 0.9672 - val_accuracy: 0.8075\n",
      "Epoch 982/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0940 - accuracy: 0.9970 - val_loss: 0.9695 - val_accuracy: 0.8075\n",
      "Epoch 983/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0928 - accuracy: 0.9982 - val_loss: 0.9774 - val_accuracy: 0.8037\n",
      "Epoch 984/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9978 - val_loss: 0.9808 - val_accuracy: 0.8000\n",
      "Epoch 985/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0924 - accuracy: 0.9984 - val_loss: 0.9810 - val_accuracy: 0.8000\n",
      "Epoch 986/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0915 - accuracy: 0.9982 - val_loss: 0.9804 - val_accuracy: 0.8012\n",
      "Epoch 987/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0921 - accuracy: 0.9980 - val_loss: 0.9846 - val_accuracy: 0.8062\n",
      "Epoch 988/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0918 - accuracy: 0.9978 - val_loss: 0.9871 - val_accuracy: 0.8075\n",
      "Epoch 989/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0919 - accuracy: 0.9981 - val_loss: 0.9843 - val_accuracy: 0.8137\n",
      "Epoch 990/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0943 - accuracy: 0.9976 - val_loss: 0.9865 - val_accuracy: 0.8087\n",
      "Epoch 991/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0930 - accuracy: 0.9977 - val_loss: 0.9972 - val_accuracy: 0.8087\n",
      "Epoch 992/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0934 - accuracy: 0.9973 - val_loss: 1.0036 - val_accuracy: 0.8100\n",
      "Epoch 993/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9979 - val_loss: 0.9985 - val_accuracy: 0.8112\n",
      "Epoch 994/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0926 - accuracy: 0.9978 - val_loss: 0.9896 - val_accuracy: 0.8062\n",
      "Epoch 995/1000\n",
      "19/19 [==============================] - 0s 19ms/step - loss: 0.0962 - accuracy: 0.9965 - val_loss: 0.9812 - val_accuracy: 0.8112\n",
      "Epoch 996/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0910 - accuracy: 0.9987 - val_loss: 0.9833 - val_accuracy: 0.8125\n",
      "Epoch 997/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0940 - accuracy: 0.9970 - val_loss: 0.9902 - val_accuracy: 0.8087\n",
      "Epoch 998/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0945 - accuracy: 0.9969 - val_loss: 0.9874 - val_accuracy: 0.8062\n",
      "Epoch 999/1000\n",
      "19/19 [==============================] - 0s 17ms/step - loss: 0.0928 - accuracy: 0.9980 - val_loss: 0.9862 - val_accuracy: 0.8025\n",
      "Epoch 1000/1000\n",
      "19/19 [==============================] - 0s 16ms/step - loss: 0.0936 - accuracy: 0.9976 - val_loss: 0.9899 - val_accuracy: 0.8062\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x2a756b610>"
      ]
     },
     "execution_count": 252,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history = model1.fit(fine_tune_data_train, fine_tune_labels_train, epochs=EPOCHS, batch_size=800, validation_data=(fine_tune_data_test, fine_tune_labels_test), validation_batch_size=800, callbacks=[model_checkpoint_callback])\n",
    "model1.save_weights(\"./Models/ResNet-VGG-Inception-ResNet-4P/resnet-vgg-inception-resnet-4p-model-weights\")\n",
    "model1.load_weights('./Models/ResNet-VGG-Inception-ResNet-4P/resnet-vgg-inception-resnet-4p-model-weights')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "id": "b1f883d7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x3080a26a0>"
      ]
     },
     "execution_count": 253,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model1.load_weights('./Models/ResNet-VGG-Inception-ResNet-4P/auto-ResNet-VGG-Inception-ResNet-FT-model-weight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 254,
   "id": "9116a6a6",
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_train_loss(history, model_name):  \n",
    "    plt.figure(figsize=(8,6))\n",
    "    plt.plot(history.history['loss'])\n",
    "    plt.plot(history.history['accuracy'])\n",
    "    plt.plot(history.history['val_loss'])\n",
    "    plt.plot(history.history['val_accuracy'])\n",
    "    plt.title(model_name+' loss')\n",
    "    plt.ylabel('loss')\n",
    "    plt.xlabel('epochs')\n",
    "    plt.legend(['train loss', 'train accuracy', 'val loss', 'val accuracy'], loc='upper left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 255,
   "id": "c55b5870",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_train_loss(history, 'Fine-Tuned ResNet-VGG-Inception-ResNet-4Piece Model')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "03390fa6",
   "metadata": {},
   "source": [
    "### Model Evaluation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "48701107",
   "metadata": {},
   "source": [
    "#### On Train Data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b0594069",
   "metadata": {},
   "source": [
    "Now, after the model is being trained, we will look at the performance of the model on the training set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "id": "cf7ded38",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 57/475 [==>...........................] - ETA: 1s"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-10-09 15:56:49.262692: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "475/475 [==============================] - 2s 4ms/step\n"
     ]
    }
   ],
   "source": [
    "train_preds = model1.predict(fine_tune_data_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "id": "f2a28269",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_pred_labels = np.argmax(train_preds, axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "id": "768775b2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 1, 13, 13, ...,  3,  9,  9])"
      ]
     },
     "execution_count": 258,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_pred_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "id": "eff52a3d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "15197"
      ]
     },
     "execution_count": 259,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.sum(train_pred_labels == train_labels['label'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3fd403fa",
   "metadata": {},
   "source": [
    "Looking at the misclassified images to get a better idea about what features the model is unable to capture."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "id": "8f594a33",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>images</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3597</th>\n",
       "      <td>3781</td>\n",
       "      <td>./Data/train/train/3781.tif</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9198</th>\n",
       "      <td>9682</td>\n",
       "      <td>./Data/train/train/9682.tif</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12255</th>\n",
       "      <td>12896</td>\n",
       "      <td>./Data/train/train/12896.tif</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          id                        images  label\n",
       "3597    3781   ./Data/train/train/3781.tif      0\n",
       "9198    9682   ./Data/train/train/9682.tif     12\n",
       "12255  12896  ./Data/train/train/12896.tif      9"
      ]
     },
     "execution_count": 260,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mis_train_data = train_labels[train_pred_labels != train_labels['label']]\n",
    "mis_train_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "id": "2ea8f949",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>fscore</th>\n",
       "      <th>support</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.998978</td>\n",
       "      <td>0.999488</td>\n",
       "      <td>978</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>947</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.998926</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.999463</td>\n",
       "      <td>930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>962</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.997814</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.998906</td>\n",
       "      <td>913</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.998951</td>\n",
       "      <td>0.999475</td>\n",
       "      <td>953</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>916</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>955</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.998976</td>\n",
       "      <td>0.999488</td>\n",
       "      <td>977</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>943</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>928</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>945</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    precision    recall    fscore  support\n",
       "0    1.000000  0.998978  0.999488      978\n",
       "1    1.000000  1.000000  1.000000      925\n",
       "2    1.000000  1.000000  1.000000      925\n",
       "3    1.000000  1.000000  1.000000      947\n",
       "4    0.998926  1.000000  0.999463      930\n",
       "5    1.000000  1.000000  1.000000      962\n",
       "6    1.000000  1.000000  1.000000      998\n",
       "7    0.997814  1.000000  0.998906      913\n",
       "8    1.000000  1.000000  1.000000     1005\n",
       "9    1.000000  0.998951  0.999475      953\n",
       "10   1.000000  1.000000  1.000000      916\n",
       "11   1.000000  1.000000  1.000000      955\n",
       "12   1.000000  0.998976  0.999488      977\n",
       "13   1.000000  1.000000  1.000000      943\n",
       "14   1.000000  1.000000  1.000000      928\n",
       "15   1.000000  1.000000  1.000000      945"
      ]
     },
     "execution_count": 261,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precision, recall, fscore, support = score(train_labels['label'], train_pred_labels)\n",
    "pd.DataFrame({\n",
    "    'precision': precision,\n",
    "    'recall': recall,\n",
    "    'fscore': fscore,\n",
    "    'support': support\n",
    "})"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "659ae313",
   "metadata": {},
   "source": [
    "#### On Train-Test Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "id": "0d1e8f85",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25/25 [==============================] - 0s 4ms/step\n"
     ]
    }
   ],
   "source": [
    "test_preds = model1.predict(fine_tune_data_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "id": "49233509",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_pred_labels = np.argmax(test_preds, axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "id": "8f5cba39",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([11,  8,  0,  1, 10,  9, 13, 12, 14,  1,  6,  0,  0, 14,  1, 11,  0,\n",
       "        9,  6, 13,  8, 14,  3, 12,  6,  0,  3,  5,  3,  3, 15,  1,  9,  2,\n",
       "        5,  8,  4,  7, 11,  9,  6, 12, 14,  2,  5, 14,  5,  0, 13,  4,  0,\n",
       "       10,  5,  0, 11, 14, 15,  9,  9,  2, 15, 13,  2, 11,  6, 10,  8,  0,\n",
       "        6,  3,  4,  4, 12,  0,  8,  6,  7,  0, 12, 11, 11,  0, 12, 12,  9,\n",
       "        8,  1, 13, 13, 13, 10, 11,  4, 10,  1,  7, 11,  8, 13, 10,  3, 11,\n",
       "       12,  9,  7, 14,  8,  2,  5, 15,  7,  5,  9,  4, 12, 12, 12, 13, 11,\n",
       "        6, 13,  1,  0, 12,  1,  5, 13, 12,  0, 14,  3, 11,  6,  1,  6,  9,\n",
       "       14, 10,  7, 14, 10,  8, 11, 12,  8,  5,  8,  9,  2, 13, 15, 14,  4,\n",
       "       14,  4,  9,  4, 14,  7,  7,  8,  4,  4, 12, 13,  8,  2,  8,  8,  8,\n",
       "       13,  6,  3, 11,  7, 12, 10, 14, 14,  8,  6, 10, 13, 11,  6, 13,  4,\n",
       "       14, 11,  7, 15,  0,  3,  0, 10, 13, 13, 10, 11, 11, 10, 13,  9,  1,\n",
       "        9,  5, 13, 14, 12,  1,  4,  8,  7,  6, 13,  5,  4, 12,  8,  0,  9,\n",
       "       13, 15,  1,  7,  5,  2,  2,  9, 11,  7,  0,  3, 12,  3,  5,  2,  5,\n",
       "        4,  0, 12, 15, 12,  8, 13, 10,  0,  0, 11,  0, 11, 13, 15, 15,  8,\n",
       "        1, 10, 13,  5,  7,  9, 14,  6, 13, 12, 13,  1, 14, 11, 11, 14,  3,\n",
       "        0, 11,  4,  3,  1,  4,  6,  2,  7,  8, 10, 10,  8,  5,  7,  8, 13,\n",
       "        6,  8, 15,  0,  2,  3,  4, 10,  0,  8, 13,  0,  7, 15, 10, 10,  5,\n",
       "        3, 13,  1, 10,  9,  2, 13,  3, 15,  6,  5,  4, 14, 11, 13,  2, 15,\n",
       "        9, 15, 15, 11,  0, 15,  3, 11,  0, 15, 15,  1, 11,  9, 13, 15,  4,\n",
       "       13,  0,  3,  3,  2,  8,  2, 15,  1,  3,  3,  2,  5, 14,  1,  0,  3,\n",
       "        8, 12,  1,  3,  0, 12, 10,  3,  7, 15, 15,  3,  2, 12, 15,  3,  6,\n",
       "       15, 14, 14,  0,  5,  8,  2,  6, 12, 13,  6,  9,  8,  7, 10,  1,  2,\n",
       "       11, 14,  2, 14, 15, 15,  0,  1,  2, 10,  1,  0, 14,  6,  9,  0,  2,\n",
       "        7,  3, 14,  9, 13, 11,  4,  8, 13,  1,  4,  7, 13, 14,  5, 11,  8,\n",
       "        0,  5, 11,  0, 11,  1,  7,  2,  5,  6, 15,  9, 11, 13,  3,  0,  8,\n",
       "       12,  0,  3, 13,  5,  9,  6, 14,  7,  8,  3,  2, 13, 12,  1, 13, 11,\n",
       "        9,  8, 10, 14, 12,  0, 15,  1,  9,  9, 15, 12,  7, 14,  9,  5,  1,\n",
       "       12, 10,  4,  8,  9,  3,  7,  8,  6, 10,  1, 10,  7, 15,  1, 13, 10,\n",
       "        2,  3,  1, 10, 14,  7,  8,  5, 12,  1,  6,  7, 15,  0, 13,  7, 12,\n",
       "       12,  1,  0, 12, 13,  1, 10,  7,  9, 10, 14,  1,  5, 14,  6, 14,  7,\n",
       "        1, 14,  6, 14,  9,  5, 12, 14,  1,  5,  2, 13,  7,  2,  0,  8,  8,\n",
       "       15,  8,  6,  5, 14,  3,  6, 11,  8, 14,  6,  9, 11,  9, 11,  2, 13,\n",
       "        3, 10, 11, 10, 15,  2,  7,  5, 14,  9,  1,  1,  9, 13,  1, 10, 12,\n",
       "        9, 11,  6, 10,  5,  8, 10, 14, 14,  9,  0,  0,  8,  5,  8,  6, 12,\n",
       "        9,  5,  0, 15, 14,  1,  3,  2, 12,  7, 15, 11, 10, 12, 13,  1,  1,\n",
       "        1, 13, 12,  4,  4,  9, 11, 15, 14, 13,  4,  8,  5, 10, 13,  4, 15,\n",
       "       13,  0,  1, 13, 13,  7,  6, 12,  8,  7,  5, 15,  5,  9,  4,  1, 14,\n",
       "        1, 13,  6,  6,  3,  7,  7,  5,  5, 12,  6,  8,  0,  4,  7,  7,  0,\n",
       "       12, 14, 12,  0,  3, 11,  2,  3,  8,  4,  4,  1, 15,  5, 14, 15,  7,\n",
       "       11, 12, 10,  7,  2, 13, 15,  8, 11,  6,  9,  2, 10, 13,  7,  6,  3,\n",
       "       13,  3,  2, 14,  2,  8, 12, 12,  8, 13, 13, 14,  4, 12,  2,  5, 14,\n",
       "       12, 13, 13,  4, 12,  0, 10,  0,  8, 13, 13,  8, 12,  9,  2,  6,  0,\n",
       "        1,  2, 14, 15,  0,  3,  9, 11,  0,  8,  9,  7,  4, 11, 13, 11,  3,\n",
       "        7,  7, 13,  8, 12,  2,  5, 15,  7, 10, 13,  5, 12,  6, 14,  0, 13,\n",
       "        5,  3, 12, 13, 14,  7,  2,  8, 10, 13,  2,  9,  8, 14, 12, 14, 12,\n",
       "        3,  5, 13, 15,  4, 11, 15,  0,  9,  9,  5, 13, 10,  2,  2, 11, 15,\n",
       "        6])"
      ]
     },
     "execution_count": 264,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_pred_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "id": "ee078822",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "651"
      ]
     },
     "execution_count": 265,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.sum(test_pred_labels == test_labels['label'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "id": "c3db9b5c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>images</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>33</td>\n",
       "      <td>./Data/train/train/33.tif</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>65</td>\n",
       "      <td>./Data/train/train/65.tif</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>100</td>\n",
       "      <td>./Data/train/train/100.tif</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>164</td>\n",
       "      <td>./Data/train/train/164.tif</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>179</td>\n",
       "      <td>./Data/train/train/179.tif</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>764</th>\n",
       "      <td>15341</td>\n",
       "      <td>./Data/train/train/15341.tif</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>776</th>\n",
       "      <td>15528</td>\n",
       "      <td>./Data/train/train/15528.tif</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>779</th>\n",
       "      <td>15568</td>\n",
       "      <td>./Data/train/train/15568.tif</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>784</th>\n",
       "      <td>15596</td>\n",
       "      <td>./Data/train/train/15596.tif</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>788</th>\n",
       "      <td>15687</td>\n",
       "      <td>./Data/train/train/15687.tif</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>149 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        id                        images  label\n",
       "2       33     ./Data/train/train/33.tif      6\n",
       "4       65     ./Data/train/train/65.tif     14\n",
       "7      100    ./Data/train/train/100.tif     10\n",
       "11     164    ./Data/train/train/164.tif     15\n",
       "12     179    ./Data/train/train/179.tif      7\n",
       "..     ...                           ...    ...\n",
       "764  15341  ./Data/train/train/15341.tif      7\n",
       "776  15528  ./Data/train/train/15528.tif      5\n",
       "779  15568  ./Data/train/train/15568.tif      9\n",
       "784  15596  ./Data/train/train/15596.tif      8\n",
       "788  15687  ./Data/train/train/15687.tif     12\n",
       "\n",
       "[149 rows x 3 columns]"
      ]
     },
     "execution_count": 266,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mis_train_data = test_labels[test_pred_labels != test_labels['label']]\n",
    "mis_train_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 267,
   "id": "b538b721",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>fscore</th>\n",
       "      <th>support</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.771930</td>\n",
       "      <td>0.814815</td>\n",
       "      <td>0.792793</td>\n",
       "      <td>54</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.653061</td>\n",
       "      <td>0.727273</td>\n",
       "      <td>0.688172</td>\n",
       "      <td>44</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.886364</td>\n",
       "      <td>0.906977</td>\n",
       "      <td>0.896552</td>\n",
       "      <td>43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.954545</td>\n",
       "      <td>0.913043</td>\n",
       "      <td>0.933333</td>\n",
       "      <td>46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.885714</td>\n",
       "      <td>0.861111</td>\n",
       "      <td>0.873239</td>\n",
       "      <td>36</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.717391</td>\n",
       "      <td>0.660000</td>\n",
       "      <td>0.687500</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.928571</td>\n",
       "      <td>0.847826</td>\n",
       "      <td>0.886364</td>\n",
       "      <td>46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.854167</td>\n",
       "      <td>0.788462</td>\n",
       "      <td>0.820000</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0.844828</td>\n",
       "      <td>0.907407</td>\n",
       "      <td>0.875000</td>\n",
       "      <td>54</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0.812500</td>\n",
       "      <td>0.829787</td>\n",
       "      <td>0.821053</td>\n",
       "      <td>47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>0.844444</td>\n",
       "      <td>0.655172</td>\n",
       "      <td>0.737864</td>\n",
       "      <td>58</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>0.840000</td>\n",
       "      <td>0.750000</td>\n",
       "      <td>0.792453</td>\n",
       "      <td>56</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>0.649123</td>\n",
       "      <td>0.860465</td>\n",
       "      <td>0.740000</td>\n",
       "      <td>43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>0.767123</td>\n",
       "      <td>0.875000</td>\n",
       "      <td>0.817518</td>\n",
       "      <td>64</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>0.964912</td>\n",
       "      <td>0.859375</td>\n",
       "      <td>0.909091</td>\n",
       "      <td>64</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>0.723404</td>\n",
       "      <td>0.790698</td>\n",
       "      <td>0.755556</td>\n",
       "      <td>43</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    precision    recall    fscore  support\n",
       "0    0.771930  0.814815  0.792793       54\n",
       "1    0.653061  0.727273  0.688172       44\n",
       "2    0.886364  0.906977  0.896552       43\n",
       "3    0.954545  0.913043  0.933333       46\n",
       "4    0.885714  0.861111  0.873239       36\n",
       "5    0.717391  0.660000  0.687500       50\n",
       "6    0.928571  0.847826  0.886364       46\n",
       "7    0.854167  0.788462  0.820000       52\n",
       "8    0.844828  0.907407  0.875000       54\n",
       "9    0.812500  0.829787  0.821053       47\n",
       "10   0.844444  0.655172  0.737864       58\n",
       "11   0.840000  0.750000  0.792453       56\n",
       "12   0.649123  0.860465  0.740000       43\n",
       "13   0.767123  0.875000  0.817518       64\n",
       "14   0.964912  0.859375  0.909091       64\n",
       "15   0.723404  0.790698  0.755556       43"
      ]
     },
     "execution_count": 267,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precision, recall, fscore, support = score(test_labels['label'], test_pred_labels)\n",
    "pd.DataFrame({\n",
    "    'precision': precision,\n",
    "    'recall': recall,\n",
    "    'fscore': fscore,\n",
    "    'support': support\n",
    "})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "id": "ed4db76c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8141554511573338"
      ]
     },
     "execution_count": 268,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(fscore)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "067c5096",
   "metadata": {},
   "source": [
    "#### On Validation Data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3e665591",
   "metadata": {},
   "source": [
    "Generating Predictions for the Validation Set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "id": "4b085c0c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>images</th>\n",
       "      <th>id</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>850</th>\n",
       "      <td>./Data/validation/validation/17801.tif</td>\n",
       "      <td>17801</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>898</th>\n",
       "      <td>./Data/validation/validation/17802.tif</td>\n",
       "      <td>17802</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>852</th>\n",
       "      <td>./Data/validation/validation/17803.tif</td>\n",
       "      <td>17803</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>798</th>\n",
       "      <td>./Data/validation/validation/17804.tif</td>\n",
       "      <td>17804</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>750</th>\n",
       "      <td>./Data/validation/validation/17805.tif</td>\n",
       "      <td>17805</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>246</th>\n",
       "      <td>./Data/validation/validation/18696.tif</td>\n",
       "      <td>18696</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>211</th>\n",
       "      <td>./Data/validation/validation/18697.tif</td>\n",
       "      <td>18697</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760</th>\n",
       "      <td>./Data/validation/validation/18698.tif</td>\n",
       "      <td>18698</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>787</th>\n",
       "      <td>./Data/validation/validation/18699.tif</td>\n",
       "      <td>18699</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>891</th>\n",
       "      <td>./Data/validation/validation/18700.tif</td>\n",
       "      <td>18700</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>900 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     images     id  label\n",
       "850  ./Data/validation/validation/17801.tif  17801     -1\n",
       "898  ./Data/validation/validation/17802.tif  17802     -1\n",
       "852  ./Data/validation/validation/17803.tif  17803     -1\n",
       "798  ./Data/validation/validation/17804.tif  17804     -1\n",
       "750  ./Data/validation/validation/17805.tif  17805     -1\n",
       "..                                      ...    ...    ...\n",
       "246  ./Data/validation/validation/18696.tif  18696     -1\n",
       "211  ./Data/validation/validation/18697.tif  18697     -1\n",
       "760  ./Data/validation/validation/18698.tif  18698     -1\n",
       "787  ./Data/validation/validation/18699.tif  18699     -1\n",
       "891  ./Data/validation/validation/18700.tif  18700     -1\n",
       "\n",
       "[900 rows x 3 columns]"
      ]
     },
     "execution_count": 269,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "validation_data = pd.DataFrame({'images':['./Data/validation/validation/'+name for name in os.listdir('./Data/validation/validation/')]})\n",
    "validation_data['id'] = [name.split('.')[0] for name in os.listdir('./Data/validation/validation/')]\n",
    "validation_data['label'] = -1 # Simply added to prevent re-writing code\n",
    "validation_data.sort_values(by=['id'], inplace=True)\n",
    "validation_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "id": "53635253",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(900, 5, 2048)"
      ]
     },
     "execution_count": 270,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_validation_resnet = np.load('./Data/ResNet-4P-validation-precompute/whole_validation_data_precomp.npy')\n",
    "fine_tune_data_validation_resnet.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "id": "8e78b6f1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(900, 5, 512)"
      ]
     },
     "execution_count": 271,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_validation_vgg = np.load('./Data/VGG16-4P-validation-precompute/whole_validation_data_precomp.npy')\n",
    "fine_tune_data_validation_vgg.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "id": "88e39e91",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(900, 5, 1536)"
      ]
     },
     "execution_count": 272,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_validation_inception_resnet = np.load('./Data/Inception-ResNet-4P-validation-precompute/whole_validation_data_precomp.npy')\n",
    "fine_tune_data_validation_inception_resnet.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "id": "13b10856",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(900, 5, 1300)"
      ]
     },
     "execution_count": 273,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_validation_red_resnet = np.zeros((fine_tune_data_validation_resnet.shape[0], 5, N_COMPS_1))\n",
    "for i in range(5):\n",
    "    fine_tune_data_validation_red_resnet[:,i,:] = pcas_resnet[i].transform(fine_tune_data_validation_resnet[:,i,:])\n",
    "fine_tune_data_validation_resnet = fine_tune_data_validation_red_resnet\n",
    "del fine_tune_data_validation_red_resnet\n",
    "fine_tune_data_validation_resnet.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "id": "b0a31a0b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(900, 5, 300)"
      ]
     },
     "execution_count": 274,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_validation_red_vgg = np.zeros((fine_tune_data_validation_vgg.shape[0], 5, N_COMPS_2))\n",
    "for i in range(5):\n",
    "    fine_tune_data_validation_red_vgg[:,i,:] = pcas_vgg[i].transform(fine_tune_data_validation_vgg[:,i,:])\n",
    "fine_tune_data_validation_vgg = fine_tune_data_validation_red_vgg\n",
    "del fine_tune_data_validation_red_vgg\n",
    "fine_tune_data_validation_vgg.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "id": "5e12894e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(900, 5, 900)"
      ]
     },
     "execution_count": 275,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_validation_red_inception_resnet = np.zeros((fine_tune_data_validation_inception_resnet.shape[0], 5, N_COMPS_3))\n",
    "for i in range(5):\n",
    "    fine_tune_data_validation_red_inception_resnet[:,i,:] = pcas_inception_resnet[i].transform(fine_tune_data_validation_inception_resnet[:,i,:])\n",
    "fine_tune_data_validation_inception_resnet = fine_tune_data_validation_red_inception_resnet\n",
    "del fine_tune_data_validation_red_inception_resnet\n",
    "fine_tune_data_validation_inception_resnet.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "id": "586ac8df",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(900, 5, 2500)"
      ]
     },
     "execution_count": 276,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fine_tune_data_validation = np.concatenate([fine_tune_data_validation_resnet, fine_tune_data_validation_vgg, fine_tune_data_validation_inception_resnet], axis=-1)\n",
    "fine_tune_data_validation.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "id": "db7d1256",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "29/29 [==============================] - 0s 3ms/step\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-10-09 15:57:30.272902: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    }
   ],
   "source": [
    "valid_preds = model1.predict(fine_tune_data_validation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "id": "18b17d2e",
   "metadata": {},
   "outputs": [],
   "source": [
    "valid_pred_labels = np.argmax(valid_preds, axis=-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2bece594",
   "metadata": {},
   "source": [
    "### Submission"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "304c47de",
   "metadata": {},
   "source": [
    "For Kaggle Submission"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 279,
   "id": "e7ca9d19",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>17801</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>17802</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>17803</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>17804</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>17805</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>895</th>\n",
       "      <td>18696</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>896</th>\n",
       "      <td>18697</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>897</th>\n",
       "      <td>18698</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>898</th>\n",
       "      <td>18699</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>899</th>\n",
       "      <td>18700</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>900 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        id  label\n",
       "0    17801      4\n",
       "1    17802      6\n",
       "2    17803      4\n",
       "3    17804      8\n",
       "4    17805      3\n",
       "..     ...    ...\n",
       "895  18696      7\n",
       "896  18697     12\n",
       "897  18698      8\n",
       "898  18699     10\n",
       "899  18700      4\n",
       "\n",
       "[900 rows x 2 columns]"
      ]
     },
     "execution_count": 279,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "submission_data = pd.DataFrame({'id':list(range(17801, 18701))})\n",
    "submission_data['label'] = valid_pred_labels\n",
    "submission_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 280,
   "id": "6eea7d9c",
   "metadata": {},
   "outputs": [],
   "source": [
    "submission_data.to_csv('./Submissions/submission-resnet-vgg-inception-4p.csv', index=False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}