material/deep_learning/template_numeric_sequence_data_classification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "template_numeric_sequence_data_classification.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mZAFnuGfwqy-"
      },
      "source": [
        "# 단일 숫자 순차열 데이터 분류 Template\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Hb0EOx6OyNIw"
      },
      "source": [
        "# 데이터 준비"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VBfCpbwh0h9R"
      },
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "B5AtaEi6yPhX"
      },
      "source": [
        "## 기반 숫자열 생성"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V6wKtOwFwqXM"
      },
      "source": [
        "numbers = np.zeros((2000,1))\n",
        "numbers[:,0] = np.sin(np.arange(0,1000,0.5))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "2ixw1zJVS69z",
        "outputId": "939e2682-2d37-4124-9e2e-40056f811319"
      },
      "source": [
        "plt.plot(numbers[:100,0])\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6RAkHiJbxNfE"
      },
      "source": [
        "SEQUENCE_LENGTH = 5\n",
        "OFFSET = 1"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f-z-YNfNDBn5"
      },
      "source": [
        "## 입출력 데이터 생성\n",
        "\n",
        "12345678901234567890의 숫자열로 다음의 입출력 데이터를 생성한다.\n",
        "```\n",
        "12345 -> T\n",
        "23456 -> F\n",
        "34567 -> T\n",
        "...\n",
        "56789 -> F\n",
        "```"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-IAAZqcIhfe0"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yYxpmUQtwqTY"
      },
      "source": [
        "raw_x = []\n",
        "raw_y = []\n",
        "\n",
        "MAX_SAMPLE_COUNT = len(numbers)-(SEQUENCE_LENGTH+OFFSET)+1\n",
        "SAMPLING_COUNT = MAX_SAMPLE_COUNT\n",
        "for j in range(SAMPLING_COUNT):\n",
        "  i = np.random.randint(MAX_SAMPLE_COUNT)\n",
        "  raw_x.append(numbers[i:i+SEQUENCE_LENGTH,:])\n",
        "  next_value = numbers[i+SEQUENCE_LENGTH+OFFSET-1]\n",
        "  raw_y.append([ int((next_value)>0) ])\n",
        "#              0         1              \n",
        "#              01234567890123456789\n",
        "# numbers[0] = 12345678901234567890\n",
        "#                 i = 3\n",
        "#                 <-x->   = [3:8] = 45678\n",
        "#                      y  = [8]   = 9>0 -> True -> 1"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "15RBiNi9wqQU",
        "outputId": "6c619947-7814-4e9d-d89c-da89249258e0"
      },
      "source": [
        "print(len(raw_x))\n",
        "# len(raw_x) = 1994 = len(numbers) - SEQUENCE_LENGTH - OFFSET = 2000 - 5 - 1"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1994\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S7eFX6jIyrnx"
      },
      "source": [
        "## 데이터를 2차원 벡터화 하기\n",
        "\n",
        "RNN은 2차원 입력을 전제로 한다. 다중 숫자열은 그 자체로 2차원 데이터이다. 따로 해줄 것은 없다.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BagxNXc8MTuU",
        "outputId": "ae5884de-73d2-434f-a30e-c1ed368ce0a5"
      },
      "source": [
        "print(raw_x[0])\n",
        "print(raw_y[0])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[0.        ]\n",
            " [0.47942554]\n",
            " [0.84147098]\n",
            " [0.99749499]\n",
            " [0.90929743]]\n",
            "[1]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4mg54AkZ3boa",
        "outputId": "93708089-fdd3-4b5f-c829-c1e75d594b04"
      },
      "source": [
        "x = np.array(raw_x)\n",
        "y = np.array(raw_y)\n",
        "print(x.shape)\n",
        "print(y.shape)\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "train_x, test_x, train_y, test_y = train_test_split(x, y, test_size=0.1, shuffle=True)\n",
        "\n",
        "print(\"Training Data:\")\n",
        "print(train_x.shape)\n",
        "print(train_y.shape)\n",
        "\n",
        "print(\"Test Data:\")\n",
        "print(test_x.shape)\n",
        "print(test_y.shape)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(1994, 5, 1)\n",
            "(1994, 1)\n",
            "Training Data:\n",
            "(1794, 5, 1)\n",
            "(1794, 1)\n",
            "Test Data:\n",
            "(200, 5, 1)\n",
            "(200, 1)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lMq2Om3w8qNy"
      },
      "source": [
        "준비된 데이터는 다음과 같습니다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "41UcKyn-q5Ez",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 422
        },
        "outputId": "11a76a23-6c1c-4f29-9d63-3a98069038a9"
      },
      "source": [
        "print(\"train_x[0]\\n\", train_x[0])\n",
        "print()\n",
        "print(\"train_y[0]\\n\", train_y[0])\n",
        "\n",
        "plt.plot(train_x[0], '.')\n",
        "plt.plot([len(train_x[0])], train_y[0], 'x')\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "train_x[0]\n",
            " [[ 0.94250438]\n",
            " [ 0.66690437]\n",
            " [ 0.22802291]\n",
            " [-0.26668651]\n",
            " [-0.69610177]]\n",
            "\n",
            "train_y[0]\n",
            " [0]\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GeG-X2N3WmZz"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# DNN 사용"
      ],
      "metadata": {
        "id": "HHwjYpcHvqG9"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import Dense, RepeatVector, Flatten, Input\n",
        "from tensorflow.keras.layers import Bidirectional, LSTM, GRU\n",
        "\n",
        "input_shape = train_x.shape[1:]\n",
        "# input_shape = (5,1)\n",
        "\n",
        "model = keras.Sequential()\n",
        "model.add(Input(input_shape))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(10, activation=\"relu\"))\n",
        "model.add(Dense(2, activation=\"softmax\"))\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=\"adam\", metrics=[\"acc\"])\n",
        "model.summary()\n",
        "\n",
        "history = model.fit(train_x, train_y, epochs=10, verbose=1, validation_split=0.1, batch_size=128)\n",
        "\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.show()\n",
        "\n",
        "\n",
        "\n",
        "loss, acc = model.evaluate(test_x, test_y)\n",
        "print(\"loss=\", loss)\n",
        "print(\"acc=\", acc)\n",
        "\n",
        "\n",
        "y_ = model.predict(test_x)\n",
        "predicted = np.argmax(y_, axis=-1)\n",
        "plt.plot(test_y[:30], 'o')\n",
        "plt.plot(predicted[:30], '.')\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "EaLHz08DvrfK",
        "outputId": "b7f64575-d98f-425a-b235-1b09bdac8596",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 825
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " flatten (Flatten)           (None, 5)                 0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 10)                60        \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 2)                 22        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 82\n",
            "Trainable params: 82\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "7/7 [==============================] - 0s 2ms/step - loss: 0.3147 - acc: 0.9300\n",
            "loss= 0.3147069215774536\n",
            "acc= 0.9300000071525574\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0-mHtIAk_qU9"
      },
      "source": [
        "# Template"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k3CG02lDAI73"
      },
      "source": [
        "## LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 860
        },
        "id": "GEffW4RR3bgJ",
        "outputId": "74d74a9c-38ac-407b-9b9e-a8b8dc5ad91b"
      },
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import Dense, RepeatVector, Flatten, Input\n",
        "from tensorflow.keras.layers import Bidirectional, LSTM, GRU\n",
        "\n",
        "input_shape = train_x.shape[1:]\n",
        "# input_shape = (5,1)\n",
        "\n",
        "model = keras.Sequential()\n",
        "model.add(Input(input_shape))\n",
        "model.add(LSTM(128))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(10, activation=\"relu\"))\n",
        "model.add(Dense(2, activation=\"softmax\"))\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=\"adam\", metrics=[\"acc\"])\n",
        "model.summary()\n",
        "\n",
        "history = model.fit(train_x, train_y, epochs=10, verbose=1, validation_split=0.1, batch_size=128)\n",
        "\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.show()\n",
        "\n",
        "\n",
        "\n",
        "loss, acc = model.evaluate(test_x, test_y)\n",
        "print(\"loss=\", loss)\n",
        "print(\"acc=\", acc)\n",
        "\n",
        "\n",
        "y_ = model.predict(test_x)\n",
        "predicted = np.argmax(y_, axis=-1)\n",
        "plt.plot(test_y[:30], 'o')\n",
        "plt.plot(predicted[:30], '.')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " lstm (LSTM)                 (None, 128)               66560     \n",
            "                                                                 \n",
            " flatten_1 (Flatten)         (None, 128)               0         \n",
            "                                                                 \n",
            " dense_2 (Dense)             (None, 10)                1290      \n",
            "                                                                 \n",
            " dense_3 (Dense)             (None, 2)                 22        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 67,872\n",
            "Trainable params: 67,872\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "7/7 [==============================] - 0s 4ms/step - loss: 0.0428 - acc: 0.9950\n",
            "loss= 0.042769335210323334\n",
            "acc= 0.9950000047683716\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "M1sV7tBtAWEz"
      },
      "source": [
        "## GRU"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 860
        },
        "id": "usltj8hI3bdL",
        "outputId": "fbddfc7d-bc9c-43cb-b3b7-4d6ab83a93d5"
      },
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import Dense, RepeatVector, Flatten, Input\n",
        "from tensorflow.keras.layers import Bidirectional, LSTM, GRU\n",
        "\n",
        "input_shape = train_x.shape[1:]\n",
        "# input_shape = (5,1)\n",
        "\n",
        "model = keras.Sequential()\n",
        "model.add(Input(input_shape))\n",
        "model.add(GRU(128))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(10, activation=\"relu\"))\n",
        "model.add(Dense(2, activation=\"softmax\"))\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=\"adam\", metrics=[\"acc\"])\n",
        "model.summary()\n",
        "\n",
        "history = model.fit(train_x, train_y, epochs=10, verbose=1, validation_split=0.1, batch_size=128)\n",
        "\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.show()\n",
        "\n",
        "\n",
        "\n",
        "loss, acc = model.evaluate(test_x, test_y)\n",
        "print(\"loss=\", loss)\n",
        "print(\"acc=\", acc)\n",
        "\n",
        "\n",
        "y_ = model.predict(test_x)\n",
        "predicted = np.argmax(y_, axis=-1)\n",
        "plt.plot(test_y[:30], 'o')\n",
        "plt.plot(predicted[:30], '.')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " gru (GRU)                   (None, 128)               50304     \n",
            "                                                                 \n",
            " flatten_2 (Flatten)         (None, 128)               0         \n",
            "                                                                 \n",
            " dense_4 (Dense)             (None, 10)                1290      \n",
            "                                                                 \n",
            " dense_5 (Dense)             (None, 2)                 22        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 51,616\n",
            "Trainable params: 51,616\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "7/7 [==============================] - 0s 3ms/step - loss: 0.0413 - acc: 0.9950\n",
            "loss= 0.04133167862892151\n",
            "acc= 0.9950000047683716\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jCBiKCg9AeHT"
      },
      "source": [
        "## Bidirectional + LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "azGH_FDbwqHT",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 878
        },
        "outputId": "85d09c27-3250-406f-a1f9-a30148452cf3"
      },
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import Dense, RepeatVector, Flatten, Input\n",
        "from tensorflow.keras.layers import Bidirectional, LSTM, GRU\n",
        "\n",
        "input_shape = train_x.shape[1:]\n",
        "# input_shape = (5,1)\n",
        "\n",
        "model = keras.Sequential()\n",
        "model.add(Input(input_shape))\n",
        "model.add(Bidirectional(LSTM(128)))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(10, activation=\"relu\"))\n",
        "model.add(Dense(2, activation=\"softmax\"))\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=\"adam\", metrics=[\"acc\"])\n",
        "model.summary()\n",
        "\n",
        "history = model.fit(train_x, train_y, epochs=10, verbose=1, validation_split=0.1, batch_size=128)\n",
        "\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.show()\n",
        "\n",
        "\n",
        "\n",
        "loss, acc = model.evaluate(test_x, test_y)\n",
        "print(\"loss=\", loss)\n",
        "print(\"acc=\", acc)\n",
        "\n",
        "\n",
        "y_ = model.predict(test_x)\n",
        "predicted = np.argmax(y_, axis=-1)\n",
        "plt.plot(test_y[:30], 'o')\n",
        "plt.plot(predicted[:30], '.')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_3\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " bidirectional (Bidirectiona  (None, 256)              133120    \n",
            " l)                                                              \n",
            "                                                                 \n",
            " flatten_3 (Flatten)         (None, 256)               0         \n",
            "                                                                 \n",
            " dense_6 (Dense)             (None, 10)                2570      \n",
            "                                                                 \n",
            " dense_7 (Dense)             (None, 2)                 22        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 135,712\n",
            "Trainable params: 135,712\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "7/7 [==============================] - 0s 4ms/step - loss: 0.0527 - acc: 0.9800\n",
            "loss= 0.052710674703121185\n",
            "acc= 0.9800000190734863\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAD4CAYAAAATpHZ6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAQPElEQVR4nO3df6zdd13H8efLtnMNIAV7JawtdJoyXZAwvJkaiC4CrlvMBorQEhIxhGlkBgNZ3NQMnDEiVSIJExxKYMRRJ8zZxJpKdAY1DHfLBmNbinUO125ulx9DFzr2g7d/nNPlcHd/fE97Tu/9fvZ8JM095/P99Hve7+/39LWz7/d77jdVhSSpDd+z2gVIkibHUJekhhjqktQQQ12SGmKoS1JD1q/WC2/evLm2b9++Wi8vSb108ODBr1bVzFLLVy3Ut2/fztzc3Gq9vCT1UpKvLLfcwy+S1BBDXZIaYqhLUkMMdUlqiKEuSQ1Z8eqXJB8Bfg54sKpevMjyAO8HLgS+Bby5qj4/6ULXghtvPcqeA4e476FjnLFpI5edfxavOWfLapf1FF3r7Es/45hG79PYTn3Z9q3VOY39vta2UZdLGj8KfAC4donlFwA7hn9+HPjg8GdTbrz1KFfccDvHHnsCgKMPHeOKG24HWFNv8q519qWfcUyj92lsp75s+9bqnMZ+X4vbaMXDL1X1GeDry0y5GLi2Bm4GNiV5/qQKXCv2HDj05I675/Q3AnDssSfYc+DQapb1FF3r7Es/45hG79PYTn3Z9q3VOY39vha30SSOqW8B7h15fmQ49hRJLkkyl2Rufn5+Ai996tz30LEnH29/5LpFx9eCrnX2pZ9xTKP3aWynvmz71uqcxn5fi9volJ4oraprqmq2qmZnZpb8luuadMamjWONr5audfaln3FMo/dpbKe+bPvW6pzGfl+L22gSoX4U2DbyfOtwrCmXnX8WGzes+66xjRvWcdn5Z61SRYvrWmdf+hnHNHqfxnbqy7Zvrc5p7Pe1uI0m8btf9gGXJtnL4ATpN6vq/gmsd005ftJjLZ3lXkzXOvvSzzim0fs0tlNftn1rdU5jv6/FbZSV7lGa5BPAecBm4AHgXcAGgKr60PCSxg8AOxlc0vjLVbXib+qanZ0tf6GXJI0nycGqml1q+Yqf1Ktq9wrLC3jbCdQmSZowv1EqSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDOoV6kp1JDiU5nOTyRZa/IMlNSW5N8sUkF06+VEnSSlYM9STrgKuBC4Czgd1Jzl4w7XeA66vqHGAX8KeTLlSStLIun9TPBQ5X1d1V9SiwF7h4wZwCvm/4+NnAfZMrUZLUVZdQ3wLcO/L8yHBs1LuBNyU5AuwHfn2xFSW5JMlckrn5+fkTKFeStJxJnSjdDXy0qrYCFwIfT/KUdVfVNVU1W1WzMzMzE3ppSdJxXUL9KLBt5PnW4diotwDXA1TVZ4HTgc2TKFCS1F2XUL8F2JHkzCSnMTgRum/BnP8GXgmQ5EcYhLrHVyTpFFsx1KvqceBS4ABwF4OrXO5IclWSi4bT3gm8NckXgE8Ab66qmlbRkqTFre8yqar2MzgBOjp25cjjO4GXT7Y0SdK4/EapJDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJakinUE+yM8mhJIeTXL7EnNcnuTPJHUmum2yZkqQu1q80Ick64Grg1cAR4JYk+6rqzpE5O4ArgJdX1TeS/MC0CpYkLa3LJ/VzgcNVdXdVPQrsBS5eMOetwNVV9Q2AqnpwsmVKkrroEupbgHtHnh8Zjo16EfCiJP+W5OYkOxdbUZJLkswlmZufnz+xiiVJS5rUidL1wA7gPGA38OEkmxZOqqprqmq2qmZnZmYm9NKSpOO6hPpRYNvI863DsVFHgH1V9VhV/RfwZQYhL0k6hbqE+i3AjiRnJjkN2AXsWzDnRgaf0kmymcHhmLsnWKckqYMVQ72qHgcuBQ4AdwHXV9UdSa5KctFw2gHga0nuBG4CLquqr02raEnS4lJVq/LCs7OzNTc3tyqvLUl9leRgVc0utdxvlEpSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1JBOoZ5kZ5JDSQ4nuXyZeb+QpJLMTq5ESVJXK4Z6knXA1cAFwNnA7iRnLzLvWcDbgc9NukhJUjddPqmfCxyuqrur6lFgL3DxIvN+D/hD4JEJ1idJGkOXUN8C3Dvy/Mhw7ElJXgZsq6q/W25FSS5JMpdkbn5+fuxiJUnLO+kTpUm+B3gf8M6V5lbVNVU1W1WzMzMzJ/vSkqQFuoT6UWDbyPOtw7HjngW8GPjnJPcAPwHs82SpJJ16XUL9FmBHkjOTnAbsAvYdX1hV36yqzVW1vaq2AzcDF1XV3FQqliQtacVQr6rHgUuBA8BdwPVVdUeSq5JcNO0CJUndre8yqar2A/sXjF25xNzzTr4sSdKJ8BulktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGdQj3JziSHkhxOcvkiy9+R5M4kX0zyj0leOPlSJUkrWTHUk6wDrgYuAM4Gdic5e8G0W4HZqnoJ8EngvZMuVJK0si6f1M8FDlfV3VX1KLAXuHh0QlXdVFXfGj69Gdg62TIlSV10CfUtwL0jz48Mx5byFuDvF1uQ5JIkc0nm5ufnu1cpSepkoidKk7wJmAX2LLa8qq6pqtmqmp2ZmZnkS0uSgPUd5hwFto083zoc+y5JXgX8NvDTVfXtyZQnSRpHl0/qtwA7kpyZ5DRgF7BvdEKSc4A/Ay6qqgcnX6YkqYsVQ72qHgcuBQ4AdwHXV9UdSa5KctFw2h7gmcBfJ7ktyb4lVidJmqIuh1+oqv3A/gVjV448ftWE65IknQC/USpJDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkPWd5mUZCfwfmAd8OdV9Z4Fy78XuBb4MeBrwBuq6p7Jlgo33nqUPQcOcd9Dxzhj00YuO/8sXnPOlpOaO846W6tzHNOosy+9d/V0fi/1pc7Vdip6SlUtPyFZB3wZeDVwBLgF2F1Vd47M+TXgJVX1q0l2Aa+tqjcst97Z2dmam5vrXOiNtx7lihtu59hjTzw5tnHDOv7g53/0KRul69xx1tlaneOYRp196b2rp/N7qS91rrZJ9ZTkYFXNLrW8y+GXc4HDVXV3VT0K7AUuXjDnYuBjw8efBF6ZJJ2r7GDPgUNPbox7Tn8jAMcee4I9Bw6d8Nxx1tlaneOYRp196b2rp/N7qS91rrZT1VOXUN8C3Dvy/MhwbNE5VfU48E3g+xeuKMklSeaSzM3Pz49V6H0PHXvy8fZHrlt0fNy546yztTrHMY06+9J7V0/n91Jf6lxtp6qnU3qitKquqarZqpqdmZkZ6++esWlj5/Guc8dZZ1d9qXMc06izL7139XR+L/WlztV2qnrqEupHgW0jz7cOxxadk2Q98GwGJ0wn5rLzz2LjhnXfNbZxwzouO/+sE547zjpbq3Mc06izL7139XR+L/WlztV2qnrqcvXLLcCOJGcyCO9dwBsXzNkH/BLwWeB1wD/VSmdgx3T8REKXM8dd546zztbqHMc06uxL7109nd9LfalztZ2qnla8+gUgyYXAnzC4pPEjVfX7Sa4C5qpqX5LTgY8D5wBfB3ZV1d3LrXPcq18kSStf/dLpOvWq2g/sXzB25cjjR4BfPNEiJUmT4TdKJakhhrokNcRQl6SGGOqS1JBOV79M5YWTeeArJ/jXNwNfnWA5a0FrPbXWD7TXU2v9QHs9LdbPC6tqyW9vrlqon4wkc8td0tNHrfXUWj/QXk+t9QPt9XQi/Xj4RZIaYqhLUkP6GurXrHYBU9BaT631A+311Fo/0F5PY/fTy2PqkqTF9fWTuiRpEYa6JDWkd6GeZGeSQ0kOJ7l8tes5WUnuSXJ7ktuS9PLXVib5SJIHk3xpZOy5ST6d5D+GP5+zmjWOY4l+3p3k6HA/3Tb8zaW9kWRbkpuS3JnkjiRvH473cj8t009v91OS05P8e5IvDHv63eH4mUk+N8y8v0py2rLr6dMx9S43we6bJPcAs1XV2y9MJPkp4GHg2qp68XDsvcDXq+o9w//4PqeqfnM16+xqiX7eDTxcVX+0mrWdqCTPB55fVZ9P8izgIPAa4M30cD8t08/r6el+Gt7X+RlV9XCSDcC/Am8H3gHcUFV7k3wI+EJVfXCp9fTtk3qXm2DrFKuqzzD4PfqjRm9G/jEG/+B6YYl+eq2q7q+qzw8f/x9wF4N7C/dyPy3TT2/VwMPDpxuGfwr4GeCTw/EV91HfQr3LTbD7poB/SHIwySWrXcwEPa+q7h8+/h/geatZzIRcmuSLw8MzvThMsZgk2xnc0OZzNLCfFvQDPd5PSdYluQ14EPg08J/AQ1X1+HDKipnXt1Bv0Suq6mXABcDbhv/r35ThrQ37c5xvcR8Efgh4KXA/8MerW86JSfJM4FPAb1TV/44u6+N+WqSfXu+nqnqiql7K4F7Q5wI/PO46+hbqXW6C3StVdXT480HgbxjsyBY8MDzuefz454OrXM9JqaoHhv/gvgN8mB7up+Fx2k8Bf1lVNwyHe7ufFuunhf0EUFUPATcBPwlsSnL8LnUrZl7fQv3Jm2APzwDvYnDT615K8ozhSR6SPAP4WeBLy/+t3jh+M3KGP/92FWs5aceDb+i19Gw/DU/C/QVwV1W9b2RRL/fTUv30eT8lmUmyafh4I4MLQu5iEO6vG05bcR/16uoXWPwm2Ktc0glL8oMMPp3D4H6x1/WxnySfAM5j8GtCHwDeBdwIXA+8gMGvWH59VfXi5OMS/ZzH4H/pC7gH+JWRY9FrXpJXAP8C3A58Zzj8WwyOQ/duPy3Tz256up+SvITBidB1DD5wX19VVw1zYi/wXOBW4E1V9e0l19O3UJckLa1vh18kScsw1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JD/h8d3aAbRcu3JgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TfY_p_SJA5Rf"
      },
      "source": [
        "## Bidirectional + GRU"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OcjycCe8wqEH",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 878
        },
        "outputId": "cc86d265-2526-40f1-f03c-11cc745e4688"
      },
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import Dense, RepeatVector, Flatten, Input\n",
        "from tensorflow.keras.layers import Bidirectional, LSTM, GRU\n",
        "\n",
        "input_shape = train_x.shape[1:]\n",
        "# input_shape = (5,1)\n",
        "\n",
        "model = keras.Sequential()\n",
        "model.add(Input(input_shape))\n",
        "model.add(Bidirectional(GRU(128)))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(10, activation=\"relu\"))\n",
        "model.add(Dense(2, activation=\"softmax\"))\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=\"adam\", metrics=[\"acc\"])\n",
        "model.summary()\n",
        "\n",
        "history = model.fit(train_x, train_y, epochs=10, verbose=1, validation_split=0.1, batch_size=128)\n",
        "\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.show()\n",
        "\n",
        "\n",
        "\n",
        "loss, acc = model.evaluate(test_x, test_y)\n",
        "print(\"loss=\", loss)\n",
        "print(\"acc=\", acc)\n",
        "\n",
        "\n",
        "y_ = model.predict(test_x)\n",
        "predicted = np.argmax(y_, axis=-1)\n",
        "plt.plot(test_y[:30], 'o')\n",
        "plt.plot(predicted[:30], '.')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_4\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " bidirectional_1 (Bidirectio  (None, 256)              100608    \n",
            " nal)                                                            \n",
            "                                                                 \n",
            " flatten_4 (Flatten)         (None, 256)               0         \n",
            "                                                                 \n",
            " dense_8 (Dense)             (None, 10)                2570      \n",
            "                                                                 \n",
            " dense_9 (Dense)             (None, 2)                 22        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 103,200\n",
            "Trainable params: 103,200\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "7/7 [==============================] - 0s 3ms/step - loss: 0.0438 - acc: 0.9750\n",
            "loss= 0.043835509568452835\n",
            "acc= 0.9750000238418579\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "P3aO3WEwA-lR"
      },
      "source": [
        "## stacked"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xIYi9GUDwqA6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 895
        },
        "outputId": "9aaeb304-87fb-4498-e58b-f20f9d0cc8a8"
      },
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import Dense, RepeatVector, Flatten, Input\n",
        "from tensorflow.keras.layers import Bidirectional, LSTM, GRU\n",
        "\n",
        "input_shape = train_x.shape[1:]\n",
        "# input_shape = (5,1)\n",
        "\n",
        "model = keras.Sequential()\n",
        "model.add(Input(input_shape))\n",
        "model.add(LSTM(128, return_sequences=True))\n",
        "model.add(LSTM(128))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(10, activation=\"relu\"))\n",
        "model.add(Dense(2, activation=\"softmax\"))\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=\"adam\", metrics=[\"acc\"])\n",
        "model.summary()\n",
        "\n",
        "history = model.fit(train_x, train_y, epochs=10, verbose=1, validation_split=0.1, batch_size=128)\n",
        "\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.show()\n",
        "\n",
        "\n",
        "\n",
        "loss, acc = model.evaluate(test_x, test_y)\n",
        "print(\"loss=\", loss)\n",
        "print(\"acc=\", acc)\n",
        "\n",
        "\n",
        "y_ = model.predict(test_x)\n",
        "predicted = np.argmax(y_, axis=-1)\n",
        "plt.plot(test_y[:30], 'o')\n",
        "plt.plot(predicted[:30], '.')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_5\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " lstm_2 (LSTM)               (None, 5, 128)            66560     \n",
            "                                                                 \n",
            " lstm_3 (LSTM)               (None, 128)               131584    \n",
            "                                                                 \n",
            " flatten_5 (Flatten)         (None, 128)               0         \n",
            "                                                                 \n",
            " dense_10 (Dense)            (None, 10)                1290      \n",
            "                                                                 \n",
            " dense_11 (Dense)            (None, 2)                 22        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 199,456\n",
            "Trainable params: 199,456\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "7/7 [==============================] - 0s 4ms/step - loss: 0.1163 - acc: 0.9350\n",
            "loss= 0.11634407937526703\n",
            "acc= 0.9350000023841858\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAD4CAYAAAATpHZ6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAQVklEQVR4nO3df+xdd13H8efLtnMNIAX7lbC20GnKdEHC8Jupgegi4LrFbKAILSERQ5hGZjCQxU3NmDNGpEokYYJFCYw46oQ5m1hTic6ghuG+ZYOxLcU6h2s3ty8/Nl3o2A/e/nFvl8t33x/ntvf22/PZ85E033s/59Nz3+9zbl+7O+fc70lVIUlqw/esdgGSpMkx1CWpIYa6JDXEUJekhhjqktSQtav1whs3bqytW7eu1stLUi8dOHDga1U1s9TyVQv1rVu3Mjc3t1ovL0m9lOSryy338IskNcRQl6SGGOqS1BBDXZIaYqhLUkNWvPolyUeBnwMerKqXLrI8wAeAC4FvAW+tqi9MutBTwY23HmHX/oPc99BRztiwnsvOP4vXnbNptct6mq519qWfcUyj92lsp75s+9bqnMZ+P9W2UZdLGj8GfBC4donlFwDbhn9+HPjQ8GdTbrz1CFfccDtHH38SgCMPHeWKG24HOKXe5F3r7Es/45hG79PYTn3Z9q3VOY39fipuoxUPv1TVZ4FvLDPlYuDaGrgZ2JDkhZMq8FSxa//Bp3bcPae/GYCjjz/Jrv0HV7Osp+laZ1/6GcdoT8ecaO/T2E592fat1TmN/d71PXcyTeKY+ibg3pHnh4djT5PkkiRzSebm5+cn8NInz30PHX3q8dZHr1t0/FTQtc6+9DOOpWo/kd6nsZ36su1bq3Ma+73re+5kOqknSqtqd1XNVtXszMyS33I9JZ2xYf1Y46ula5196Wcc0+h9GtupL9u+tTqnsd9PxW00iVA/AmwZeb55ONaUy84/i/Xr1nzX2Pp1a7js/LNWqaLFda2zL/2MYxq9T2M79WXbt1bnNPb7qbiNJvG7X/YClybZw+AE6cNVdf8E1ntKOXbS41Q6y72YrnX2pZ9xTKP3aWynvmz71uqcxn4/FbdRVrpHaZJPAucBG4EHgPcA6wCq6sPDSxo/CGxncEnjL1fVir+pa3Z2tvyFXpI0niQHqmp2qeUrflKvqp0rLC/gHcdRmyRpwvxGqSQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDekU6km2JzmY5FCSyxdZ/qIkNyW5NcmXklw4+VIlSStZMdSTrAGuAS4AzgZ2Jjl7wbTfAa6vqnOAHcCfTrpQSdLKunxSPxc4VFV3V9VjwB7g4gVzCvi+4ePnAvdNrkRJUlddQn0TcO/I88PDsVFXAW9JchjYB/z6YitKckmSuSRz8/Pzx1GuJGk5kzpRuhP4WFVtBi4EPpHkaeuuqt1VNVtVszMzMxN6aUnSMV1C/QiwZeT55uHYqLcB1wNU1eeA04GNkyhQktRdl1C/BdiW5MwkpzE4Ebp3wZz/Bl4NkORHGIS6x1ck6SRbMdSr6gngUmA/cBeDq1zuSHJ1kouG094NvD3JF4FPAm+tqppW0ZKkxa3tMqmq9jE4ATo6duXI4zuBV062NEnSuPxGqSQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWpIp1BPsj3JwSSHkly+xJw3JrkzyR1JrptsmZKkLtauNCHJGuAa4LXAYeCWJHur6s6ROduAK4BXVtU3k/zAtAqWJC2tyyf1c4FDVXV3VT0G7AEuXjDn7cA1VfVNgKp6cLJlSpK66BLqm4B7R54fHo6NegnwkiT/luTmJNsXW1GSS5LMJZmbn58/voolSUua1InStcA24DxgJ/CRJBsWTqqq3VU1W1WzMzMzE3ppSdIxXUL9CLBl5Pnm4diow8Deqnq8qv4L+AqDkJcknURdQv0WYFuSM5OcBuwA9i6YcyODT+kk2cjgcMzdE6xTktTBiqFeVU8AlwL7gbuA66vqjiRXJ7loOG0/8PUkdwI3AZdV1denVbQkaXGpqlV54dnZ2Zqbm1uV15akvkpyoKpml1ruN0olqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWpIp1BPsj3JwSSHkly+zLxfSFJJZidXoiSpqxVDPcka4BrgAuBsYGeSsxeZ9xzgncDnJ12kJKmbLp/UzwUOVdXdVfUYsAe4eJF5vwf8IfDoBOuTJI2hS6hvAu4deX54OPaUJK8AtlTV3y23oiSXJJlLMjc/Pz92sZKk5Z3widIk3wO8H3j3SnOrandVzVbV7MzMzIm+tCRpgS6hfgTYMvJ883DsmOcALwX+Ock9wE8Aez1ZKkknX5dQvwXYluTMJKcBO4C9xxZW1cNVtbGqtlbVVuBm4KKqmptKxZKkJa0Y6lX1BHApsB+4C7i+qu5IcnWSi6ZdoCSpu7VdJlXVPmDfgrErl5h73omXJUk6Hn6jVJIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDWkU6gn2Z7kYJJDSS5fZPm7ktyZ5EtJ/jHJiydfqiRpJSuGepI1wDXABcDZwM4kZy+YdiswW1UvAz4FvG/ShUqSVtblk/q5wKGquruqHgP2ABePTqiqm6rqW8OnNwObJ1umJKmLLqG+Cbh35Pnh4dhS3gb8/WILklySZC7J3Pz8fPcqJUmdTPREaZK3ALPArsWWV9XuqpqtqtmZmZlJvrQkCVjbYc4RYMvI883Dse+S5DXAbwM/XVXfnkx5kqRxdPmkfguwLcmZSU4DdgB7RyckOQf4M+Ciqnpw8mVKkrpYMdSr6gngUmA/cBdwfVXdkeTqJBcNp+0Cng38dZLbkuxdYnWSpCnqcviFqtoH7FswduXI49dMuC5J0nHwG6WS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDVkbZdJSbYDHwDWAH9eVe9dsPx7gWuBHwO+Drypqu6ZbKlw461H2LX/IPc9dJQzNqznsvPP4nXnbDqhueOss7U6xzGNOvvSe1fP5PdSX+pcbSejp1TV8hOSNcBXgNcCh4FbgJ1VdefInF8DXlZVv5pkB/D6qnrTcuudnZ2tubm5zoXeeOsRrrjhdo4+/uRTY+vXreEPfv5Hn7ZRus4dZ52t1TmOadTZl967eia/l/pS52qbVE9JDlTV7FLLuxx+ORc4VFV3V9VjwB7g4gVzLgY+Pnz8KeDVSdK5yg527T/41Ma45/Q3A3D08SfZtf/gcc8dZ52t1TmOadQ59d6vem7n/ibhmfxe6kudU7fCe+5k9dQl1DcB9448PzwcW3ROVT0BPAx8/8IVJbkkyVySufn5+bEKve+ho0893vrodYuOjzt3nHW2Vuc4plHn1Hu/6uHll0/YM/m91Jc6p26F99zJ6umkniitqt1VNVtVszMzM2P93TM2rO883nXuOOvsqi91jmMadfal966eye+lvtS52k5WT11C/QiwZeT55uHYonOSrAWey+CE6cRcdv5ZrF+35rvG1q9bw2Xnn3Xcc8dZZ2t1jmMadfal966eye+lvtS52k5WT12ufrkF2JbkTAbhvQN484I5e4FfAj4HvAH4p1rpDOyYjp1I6HLmuOvccdbZWp3jmEadfem9q2fye6kvda62k9XTile/ACS5EPgTBpc0frSqfj/J1cBcVe1NcjrwCeAc4BvAjqq6e7l1jnv1iyRp5atfOl2nXlX7gH0Lxq4cefwo8IvHW6QkaTL8RqkkNcRQl6SGGOqS1BBDXZIa0unql6m8cDIPfPU4//pG4GsTLOdU0FpPrfUD7fXUWj/QXk+L9fPiqlry25urFuonIsnccpf09FFrPbXWD7TXU2v9QHs9HU8/Hn6RpIYY6pLUkL6G+u7VLmAKWuuptX6gvZ5a6wfa62nsfnp5TF2StLi+flKXJC3CUJekhvQu1JNsT3IwyaEkl692PScqyT1Jbk9yW5Je/trKJB9N8mCSL4+MPT/JZ5L8x/Dn81azxnEs0c9VSY4M99Ntw99c2htJtiS5KcmdSe5I8s7heC/30zL99HY/JTk9yb8n+eKwp98djp+Z5PPDzPurJKctu54+HVPvchPsvklyDzBbVb39wkSSnwIeAa6tqpcOx94HfKOq3jv8j+/zquo3V7POrpbo5yrgkar6o9Ws7XgleSHwwqr6QpLnAAeA1wFvpYf7aZl+3khP99Pwvs7PqqpHkqwD/hV4J/Au4Iaq2pPkw8AXq+pDS62nb5/Uu9wEWydZVX2Wwe/RHzV6M/KPM/gH1wtL9NNrVXV/VX1h+Pj/gLsY3Fu4l/tpmX56qwYeGT5dN/xTwM8AnxqOr7iP+hbqXW6C3TcF/EOSA0kuWe1iJugFVXX/8PH/AC9YzWIm5NIkXxoenunFYYrFJNnK4IY2n6eB/bSgH+jxfkqyJsltwIPAZ4D/BB6qqieGU1bMvL6FeoteVVWvAC4A3jH8X/+mDG9t2J/jfIv7EPBDwMuB+4E/Xt1yjk+SZwOfBn6jqv53dFkf99Mi/fR6P1XVk1X1cgb3gj4X+OFx19G3UO9yE+xeqaojw58PAn/DYEe24IHhcc9jxz8fXOV6TkhVPTD8B/cd4CP0cD8Nj9N+GvjLqrphONzb/bRYPy3sJ4Cqegi4CfhJYEOSY3epWzHz+hbqT90Ee3gGeAeDm173UpJnDU/ykORZwM8CX17+b/XGsZuRM/z5t6tYywk7FnxDr6dn+2l4Eu4vgLuq6v0ji3q5n5bqp8/7KclMkg3Dx+sZXBByF4Nwf8Nw2or7qFdXv8DiN8Fe5ZKOW5IfZPDpHAb3i72uj/0k+SRwHoNfE/oA8B7gRuB64EUMfsXyG6uqFycfl+jnPAb/S1/APcCvjByLPuUleRXwL8DtwHeGw7/F4Dh07/bTMv3spKf7KcnLGJwIXcPgA/f1VXX1MCf2AM8HbgXeUlXfXnI9fQt1SdLS+nb4RZK0DENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNeT/AWaunhsDtsrpAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OVGRQYg5beg_"
      },
      "source": [
        "## CNN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "zIc_XMibgSiB",
        "outputId": "a2590cd3-8507-45ca-a684-d297c261e626"
      },
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import Dense, RepeatVector, Flatten, Input\n",
        "from tensorflow.keras.layers import Bidirectional, LSTM, GRU\n",
        "from tensorflow.keras.layers import Conv2D, MaxPooling2D, Conv1D, MaxPooling1D, Input, Reshape\n",
        "\n",
        "input_shape = train_x.shape[1:]\n",
        "# input_shape = (5,1)\n",
        "\n",
        "model = keras.Sequential()\n",
        "model.add(Input(input_shape))\n",
        "model.add(Conv1D(32, 3, padding=\"same\"))\n",
        "model.add(Conv1D(32, 3, padding=\"same\"))\n",
        "model.add(MaxPooling1D())\n",
        "model.add(Conv1D(32, 3, padding=\"same\"))\n",
        "model.add(Conv1D(32, 3, padding=\"same\"))\n",
        "model.add(MaxPooling1D())\n",
        "model.add(Flatten())\n",
        "model.add(Dense(10, activation=\"relu\"))\n",
        "model.add(Dense(2, activation=\"softmax\"))\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=\"adam\", metrics=[\"acc\"])\n",
        "model.summary()\n",
        "\n",
        "history = model.fit(train_x, train_y, epochs=10, verbose=1, validation_split=0.1, batch_size=128)\n",
        "\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.show()\n",
        "\n",
        "\n",
        "\n",
        "loss, acc = model.evaluate(test_x, test_y)\n",
        "print(\"loss=\", loss)\n",
        "print(\"acc=\", acc)\n",
        "\n",
        "\n",
        "y_ = model.predict(test_x)\n",
        "predicted = np.argmax(y_, axis=-1)\n",
        "plt.plot(test_y[:30], 'o')\n",
        "plt.plot(predicted[:30], '.')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_6\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv1d (Conv1D)             (None, 5, 32)             128       \n",
            "                                                                 \n",
            " conv1d_1 (Conv1D)           (None, 5, 32)             3104      \n",
            "                                                                 \n",
            " max_pooling1d (MaxPooling1D  (None, 2, 32)            0         \n",
            " )                                                               \n",
            "                                                                 \n",
            " conv1d_2 (Conv1D)           (None, 2, 32)             3104      \n",
            "                                                                 \n",
            " conv1d_3 (Conv1D)           (None, 2, 32)             3104      \n",
            "                                                                 \n",
            " max_pooling1d_1 (MaxPooling  (None, 1, 32)            0         \n",
            " 1D)                                                             \n",
            "                                                                 \n",
            " flatten_6 (Flatten)         (None, 32)                0         \n",
            "                                                                 \n",
            " dense_12 (Dense)            (None, 10)                330       \n",
            "                                                                 \n",
            " dense_13 (Dense)            (None, 2)                 22        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 9,792\n",
            "Trainable params: 9,792\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "7/7 [==============================] - 0s 4ms/step - loss: 0.0184 - acc: 1.0000\n",
            "loss= 0.01840193197131157\n",
            "acc= 1.0\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0MW7OTLyfph4"
      },
      "source": [
        "## CNN + LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 930
        },
        "id": "RHQL-koAgiXH",
        "outputId": "1a60bfe6-4000-4ec6-956c-fe9c01f385b9"
      },
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import Dense, RepeatVector, Flatten, Input\n",
        "from tensorflow.keras.layers import Bidirectional, LSTM, GRU\n",
        "from tensorflow.keras.layers import Conv2D, MaxPooling2D, Conv1D, MaxPooling1D, Input, Reshape\n",
        "\n",
        "input_shape = train_x.shape[1:]\n",
        "# input_shape = (5,1)\n",
        "\n",
        "model = keras.Sequential()\n",
        "model.add(Input(input_shape))\n",
        "model.add(Conv1D(32, 3, padding=\"same\"))\n",
        "model.add(Conv1D(32, 3, padding=\"same\"))\n",
        "model.add(LSTM(128))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(10, activation=\"relu\"))\n",
        "model.add(Dense(2, activation=\"softmax\"))\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=\"adam\", metrics=[\"acc\"])\n",
        "model.summary()\n",
        "\n",
        "history = model.fit(train_x, train_y, epochs=10, verbose=1, validation_split=0.1, batch_size=128)\n",
        "\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.show()\n",
        "\n",
        "\n",
        "\n",
        "loss, acc = model.evaluate(test_x, test_y)\n",
        "print(\"loss=\", loss)\n",
        "print(\"acc=\", acc)\n",
        "\n",
        "\n",
        "y_ = model.predict(test_x)\n",
        "predicted = np.argmax(y_, axis=-1)\n",
        "plt.plot(test_y[:30], 'o')\n",
        "plt.plot(predicted[:30], '.')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_7\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv1d_4 (Conv1D)           (None, 5, 32)             128       \n",
            "                                                                 \n",
            " conv1d_5 (Conv1D)           (None, 5, 32)             3104      \n",
            "                                                                 \n",
            " lstm_4 (LSTM)               (None, 128)               82432     \n",
            "                                                                 \n",
            " flatten_7 (Flatten)         (None, 128)               0         \n",
            "                                                                 \n",
            " dense_14 (Dense)            (None, 10)                1290      \n",
            "                                                                 \n",
            " dense_15 (Dense)            (None, 2)                 22        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 86,976\n",
            "Trainable params: 86,976\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "7/7 [==============================] - 0s 4ms/step - loss: 0.0284 - acc: 0.9950\n",
            "loss= 0.028431681916117668\n",
            "acc= 0.9950000047683716\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAD4CAYAAAATpHZ6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAQPElEQVR4nO3df6zdd13H8efLtnMNIAV7JawtdJoyXZAwvJkaiC4CrlvMBorQEhIxhGlkBgNZ3NQMnDEiVSIJExxKYMRRJ8zZxJpKdAY1DHfLBmNbinUO125ulx9DFzr2g7d/nNPlcHd/fE97Tu/9fvZ8JM095/P99Hve7+/39LWz7/d77jdVhSSpDd+z2gVIkibHUJekhhjqktQQQ12SGmKoS1JD1q/WC2/evLm2b9++Wi8vSb108ODBr1bVzFLLVy3Ut2/fztzc3Gq9vCT1UpKvLLfcwy+S1BBDXZIaYqhLUkMMdUlqiKEuSQ1Z8eqXJB8Bfg54sKpevMjyAO8HLgS+Bby5qj4/6ULXghtvPcqeA4e476FjnLFpI5edfxavOWfLapf1FF3r7Es/45hG79PYTn3Z9q3VOY39vta2UZdLGj8KfAC4donlFwA7hn9+HPjg8GdTbrz1KFfccDvHHnsCgKMPHeOKG24HWFNv8q519qWfcUyj92lsp75s+9bqnMZ+X4vbaMXDL1X1GeDry0y5GLi2Bm4GNiV5/qQKXCv2HDj05I675/Q3AnDssSfYc+DQapb1FF3r7Es/45hG79PYTn3Z9q3VOY39vha30SSOqW8B7h15fmQ49hRJLkkyl2Rufn5+Ai996tz30LEnH29/5LpFx9eCrnX2pZ9xTKP3aWynvmz71uqcxn5fi9volJ4oraprqmq2qmZnZpb8luuadMamjWONr5audfaln3FMo/dpbKe+bPvW6pzGfl+L22gSoX4U2DbyfOtwrCmXnX8WGzes+66xjRvWcdn5Z61SRYvrWmdf+hnHNHqfxnbqy7Zvrc5p7Pe1uI0m8btf9gGXJtnL4ATpN6vq/gmsd005ftJjLZ3lXkzXOvvSzzim0fs0tlNftn1rdU5jv6/FbZSV7lGa5BPAecBm4AHgXcAGgKr60PCSxg8AOxlc0vjLVbXib+qanZ0tf6GXJI0nycGqml1q+Yqf1Ktq9wrLC3jbCdQmSZowv1EqSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDOoV6kp1JDiU5nOTyRZa/IMlNSW5N8sUkF06+VEnSSlYM9STrgKuBC4Czgd1Jzl4w7XeA66vqHGAX8KeTLlSStLIun9TPBQ5X1d1V9SiwF7h4wZwCvm/4+NnAfZMrUZLUVZdQ3wLcO/L8yHBs1LuBNyU5AuwHfn2xFSW5JMlckrn5+fkTKFeStJxJnSjdDXy0qrYCFwIfT/KUdVfVNVU1W1WzMzMzE3ppSdJxXUL9KLBt5PnW4diotwDXA1TVZ4HTgc2TKFCS1F2XUL8F2JHkzCSnMTgRum/BnP8GXgmQ5EcYhLrHVyTpFFsx1KvqceBS4ABwF4OrXO5IclWSi4bT3gm8NckXgE8Ab66qmlbRkqTFre8yqar2MzgBOjp25cjjO4GXT7Y0SdK4/EapJDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJakinUE+yM8mhJIeTXL7EnNcnuTPJHUmum2yZkqQu1q80Ick64Grg1cAR4JYk+6rqzpE5O4ArgJdX1TeS/MC0CpYkLa3LJ/VzgcNVdXdVPQrsBS5eMOetwNVV9Q2AqnpwsmVKkrroEupbgHtHnh8Zjo16EfCiJP+W5OYkOxdbUZJLkswlmZufnz+xiiVJS5rUidL1wA7gPGA38OEkmxZOqqprqmq2qmZnZmYm9NKSpOO6hPpRYNvI863DsVFHgH1V9VhV/RfwZQYhL0k6hbqE+i3AjiRnJjkN2AXsWzDnRgaf0kmymcHhmLsnWKckqYMVQ72qHgcuBQ4AdwHXV9UdSa5KctFw2gHga0nuBG4CLquqr02raEnS4lJVq/LCs7OzNTc3tyqvLUl9leRgVc0utdxvlEpSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1JBOoZ5kZ5JDSQ4nuXyZeb+QpJLMTq5ESVJXK4Z6knXA1cAFwNnA7iRnLzLvWcDbgc9NukhJUjddPqmfCxyuqrur6lFgL3DxIvN+D/hD4JEJ1idJGkOXUN8C3Dvy/Mhw7ElJXgZsq6q/W25FSS5JMpdkbn5+fuxiJUnLO+kTpUm+B3gf8M6V5lbVNVU1W1WzMzMzJ/vSkqQFuoT6UWDbyPOtw7HjngW8GPjnJPcAPwHs82SpJJ16XUL9FmBHkjOTnAbsAvYdX1hV36yqzVW1vaq2AzcDF1XV3FQqliQtacVQr6rHgUuBA8BdwPVVdUeSq5JcNO0CJUndre8yqar2A/sXjF25xNzzTr4sSdKJ8BulktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGdQj3JziSHkhxOcvkiy9+R5M4kX0zyj0leOPlSJUkrWTHUk6wDrgYuAM4Gdic5e8G0W4HZqnoJ8EngvZMuVJK0si6f1M8FDlfV3VX1KLAXuHh0QlXdVFXfGj69Gdg62TIlSV10CfUtwL0jz48Mx5byFuDvF1uQ5JIkc0nm5ufnu1cpSepkoidKk7wJmAX2LLa8qq6pqtmqmp2ZmZnkS0uSgPUd5hwFto083zoc+y5JXgX8NvDTVfXtyZQnSRpHl0/qtwA7kpyZ5DRgF7BvdEKSc4A/Ay6qqgcnX6YkqYsVQ72qHgcuBQ4AdwHXV9UdSa5KctFw2h7gmcBfJ7ktyb4lVidJmqIuh1+oqv3A/gVjV448ftWE65IknQC/USpJDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkMMdUlqiKEuSQ0x1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JDDHVJaoihLkkNMdQlqSGGuiQ1xFCXpIYY6pLUEENdkhpiqEtSQwx1SWqIoS5JDTHUJakhhrokNcRQl6SGGOqS1BBDXZIaYqhLUkPWd5mUZCfwfmAd8OdV9Z4Fy78XuBb4MeBrwBuq6p7Jlgo33nqUPQcOcd9Dxzhj00YuO/8sXnPOlpOaO846W6tzHNOosy+9d/V0fi/1pc7Vdip6SlUtPyFZB3wZeDVwBLgF2F1Vd47M+TXgJVX1q0l2Aa+tqjcst97Z2dmam5vrXOiNtx7lihtu59hjTzw5tnHDOv7g53/0KRul69xx1tlaneOYRp196b2rp/N7qS91rrZJ9ZTkYFXNLrW8y+GXc4HDVXV3VT0K7AUuXjDnYuBjw8efBF6ZJJ2r7GDPgUNPbox7Tn8jAMcee4I9Bw6d8Nxx1tlaneOYRp196b2rp/N7qS91rrZT1VOXUN8C3Dvy/MhwbNE5VfU48E3g+xeuKMklSeaSzM3Pz49V6H0PHXvy8fZHrlt0fNy546yztTrHMY06+9J7V0/n91Jf6lxtp6qnU3qitKquqarZqpqdmZkZ6++esWlj5/Guc8dZZ1d9qXMc06izL7139XR+L/WlztV2qnrqEupHgW0jz7cOxxadk2Q98GwGJ0wn5rLzz2LjhnXfNbZxwzouO/+sE547zjpbq3Mc06izL7139XR+L/WlztV2qnrqcvXLLcCOJGcyCO9dwBsXzNkH/BLwWeB1wD/VSmdgx3T8REKXM8dd546zztbqHMc06uxL7109nd9LfalztZ2qnla8+gUgyYXAnzC4pPEjVfX7Sa4C5qpqX5LTgY8D5wBfB3ZV1d3LrXPcq18kSStf/dLpOvWq2g/sXzB25cjjR4BfPNEiJUmT4TdKJakhhrokNcRQl6SGGOqS1JBOV79M5YWTeeArJ/jXNwNfnWA5a0FrPbXWD7TXU2v9QHs9LdbPC6tqyW9vrlqon4wkc8td0tNHrfXUWj/QXk+t9QPt9XQi/Xj4RZIaYqhLUkP6GurXrHYBU9BaT631A+311Fo/0F5PY/fTy2PqkqTF9fWTuiRpEYa6JDWkd6GeZGeSQ0kOJ7l8tes5WUnuSXJ7ktuS9PLXVib5SJIHk3xpZOy5ST6d5D+GP5+zmjWOY4l+3p3k6HA/3Tb8zaW9kWRbkpuS3JnkjiRvH473cj8t009v91OS05P8e5IvDHv63eH4mUk+N8y8v0py2rLr6dMx9S43we6bJPcAs1XV2y9MJPkp4GHg2qp68XDsvcDXq+o9w//4PqeqfnM16+xqiX7eDTxcVX+0mrWdqCTPB55fVZ9P8izgIPAa4M30cD8t08/r6el+Gt7X+RlV9XCSDcC/Am8H3gHcUFV7k3wI+EJVfXCp9fTtk3qXm2DrFKuqzzD4PfqjRm9G/jEG/+B6YYl+eq2q7q+qzw8f/x9wF4N7C/dyPy3TT2/VwMPDpxuGfwr4GeCTw/EV91HfQr3LTbD7poB/SHIwySWrXcwEPa+q7h8+/h/geatZzIRcmuSLw8MzvThMsZgk2xnc0OZzNLCfFvQDPd5PSdYluQ14EPg08J/AQ1X1+HDKipnXt1Bv0Suq6mXABcDbhv/r35ThrQ37c5xvcR8Efgh4KXA/8MerW86JSfJM4FPAb1TV/44u6+N+WqSfXu+nqnqiql7K4F7Q5wI/PO46+hbqXW6C3StVdXT480HgbxjsyBY8MDzuefz454OrXM9JqaoHhv/gvgN8mB7up+Fx2k8Bf1lVNwyHe7ufFuunhf0EUFUPATcBPwlsSnL8LnUrZl7fQv3Jm2APzwDvYnDT615K8ozhSR6SPAP4WeBLy/+t3jh+M3KGP/92FWs5aceDb+i19Gw/DU/C/QVwV1W9b2RRL/fTUv30eT8lmUmyafh4I4MLQu5iEO6vG05bcR/16uoXWPwm2Ktc0glL8oMMPp3D4H6x1/WxnySfAM5j8GtCHwDeBdwIXA+8gMGvWH59VfXi5OMS/ZzH4H/pC7gH+JWRY9FrXpJXAP8C3A58Zzj8WwyOQ/duPy3Tz256up+SvITBidB1DD5wX19VVw1zYi/wXOBW4E1V9e0l19O3UJckLa1vh18kScsw1CWpIYa6JDXEUJekhhjqktQQQ12SGmKoS1JD/h8d3aAbRcu3JgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "emE8KhR9BF-Z"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}