Merge branch 'master' of https://github.com/angeloyeo/gongdols

angeloyeo · Oct 19, 2020 · 0d30cab · 0d30cab
2 parents 8f425ab + 4dd95da
commit 0d30cab
Show file tree

Hide file tree

Showing 18 changed files with 257 additions and 1 deletion.
diff --git a/머신러닝/restricted_boltzmann_machine/MNIST_data/processed/test.pt b/머신러닝/restricted_boltzmann_machine/MNIST_data/processed/test.pt
diff --git a/머신러닝/restricted_boltzmann_machine/MNIST_data/processed/training.pt b/머신러닝/restricted_boltzmann_machine/MNIST_data/processed/training.pt
diff --git a/머신러닝/restricted_boltzmann_machine/MNIST_data/raw/t10k-images-idx3-ubyte b/머신러닝/restricted_boltzmann_machine/MNIST_data/raw/t10k-images-idx3-ubyte
diff --git a/머신러닝/restricted_boltzmann_machine/MNIST_data/raw/t10k-labels-idx1-ubyte b/머신러닝/restricted_boltzmann_machine/MNIST_data/raw/t10k-labels-idx1-ubyte
diff --git a/머신러닝/restricted_boltzmann_machine/MNIST_data/raw/train-images-idx3-ubyte b/머신러닝/restricted_boltzmann_machine/MNIST_data/raw/train-images-idx3-ubyte
diff --git a/머신러닝/restricted_boltzmann_machine/MNIST_data/raw/train-labels-idx1-ubyte b/머신러닝/restricted_boltzmann_machine/MNIST_data/raw/train-labels-idx1-ubyte
diff --git a/머신러닝/restricted_boltzmann_machine/generated_testdata.png b/머신러닝/restricted_boltzmann_machine/generated_testdata.png
diff --git a/머신러닝/restricted_boltzmann_machine/main.py b/머신러닝/restricted_boltzmann_machine/main.py
@@ -137,11 +137,29 @@ def free_energy(self,v):
                        transforms.ToTensor()
                    ]))
 
-sample_data = testset.data[:32,:].view(-1, 784) # 총 32개 데이터를 받아옴.
+sample_data = testset.test_data[:32,:].view(-1, 784) # 총 32개 데이터를 받아옴.
 sample_data = sample_data.type(torch.FloatTensor)/255.
 
 v, v1 = rbm(sample_data)
 show_adn_save("real_testdata",make_grid(v.view(32,1,28,28).data))
 show_adn_save("generated_testdata",make_grid(v1.view(32,1,28,28).data))
 
 
+#%% 적절한 hidden layer의 sample을 하나 가지고 visible layer의 sample을 만들어보자.
+
+random_idx = np.random.permutation(testset.test_data.shape[0])
+random_image = Variable(testset.test_data[random_idx[0], :, :].view(-1, 784))/255.
+sample_data = random_image.bernoulli()
+
+pr_h, h = rbm.v_to_h(sample_data)
+
+new_data = torch.empty(100, 1, 28, 28)
+for i in range(100):
+    pr_v, v = rbm.h_to_v(h)
+    new_data[i, :, :, :] = v.view(28, 28)
+    # plt.imshow(v.view(28, 28).detach().numpy(), cmap = 'gray')
+
+
+plt.imshow(
+    np.transpose(make_grid(new_data, nrow = 10).detach().numpy(),(1,2,0))
+    )
diff --git a/선형대수/NMF/NMF_on_digits.py b/선형대수/NMF/NMF_on_digits.py
@@ -0,0 +1,71 @@
+from time import time
+import logging
+import pylab as pl
+
+from sklearn.decomposition import PCA, NMF
+from sklearn import datasets
+
+
+# Display progress logs on stdout
+logging.basicConfig(level=logging.INFO,
+                    format='%(asctime)s %(levelname)s %(message)s')
+
+digits = datasets.load_digits()
+
+# reshape the data using the traditional (n_samples, n_features) shape
+n_samples = len(digits.images)
+X = digits.images.reshape((n_samples, -1))
+n_features = X.shape[1]
+
+n_components = 16
+
+######################################################################
+# Compute a PCA (eigendigits) on the digit dataset
+
+print("Extracting the top %d eigendigits from %d images" % (
+    n_components, X.shape[0]))
+t0 = time()
+pca = PCA(n_components=n_components, whiten=True).fit(X)
+print("done in %0.3fs" % (time() - t0))
+
+eigendigits = pca.components_
+
+######################################################################
+# Compute a NMF on the digit dataset
+
+print("Extracting %d non-negative features from %d images" % (
+    n_components, X.shape[0]))
+t0 = time()
+nmf = NMF(n_components=n_components, init='nndsvd', tol=1e-2).fit(X)
+print("done in %0.3fs" % (time() - t0))
+
+W = nmf.fit_transform(X)
+H = nmf.components_
+
+nmfdigits = nmf.components_
+
+######################################################################
+# Plot the results
+
+n_row, n_col = 4, 4
+
+f1 = pl.figure(figsize=(1. * n_col, 1.13 * n_row))
+f1.text(.5, .95, 'Principal components', horizontalalignment='center')
+for i in range(n_row * n_col):
+    pl.subplot(n_row, n_col, i + 1)
+    pl.imshow(eigendigits[i].reshape((8, 8)), cmap=pl.cm.gray,
+              interpolation='nearest')
+    pl.xticks(())
+    pl.yticks(())
+pl.subplots_adjust(0.01, 0.05, 0.99, 0.93, 0.04, 0.)
+
+f2 = pl.figure(figsize=(1. * n_col, 1.13 * n_row))
+f2.text(.5, .95, 'Non-negative components', horizontalalignment='center')
+for i in range(n_row * n_col):
+    pl.subplot(n_row, n_col, i + 1)
+    pl.imshow(nmfdigits[i].reshape((8, 8)), cmap=pl.cm.gray,
+              interpolation='nearest')
+    pl.xticks(())
+    pl.yticks(())
+pl.subplots_adjust(0.01, 0.05, 0.99, 0.93, 0.04, 0.)
+pl.show()
diff --git a/선형대수/NMF/YaleB_32x32.mat b/선형대수/NMF/YaleB_32x32.mat
diff --git a/선형대수/NMF/html/no01_NMF_on_face_dataset.html b/선형대수/NMF/html/no01_NMF_on_face_dataset.html
@@ -0,0 +1,134 @@
+
+<!DOCTYPE html
+  PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
+<html><head>
+      <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
+   <!--
+This HTML was auto-generated from MATLAB code.
+To make changes, update the MATLAB code and republish this document.
+      --><title>no01_NMF_on_face_dataset</title><meta name="generator" content="MATLAB 9.2"><link rel="schema.DC" href="http://purl.org/dc/elements/1.1/"><meta name="DC.date" content="2020-10-15"><meta name="DC.source" content="no01_NMF_on_face_dataset.m"><style type="text/css">
+html,body,div,span,applet,object,iframe,h1,h2,h3,h4,h5,h6,p,blockquote,pre,a,abbr,acronym,address,big,cite,code,del,dfn,em,font,img,ins,kbd,q,s,samp,small,strike,strong,sub,sup,tt,var,b,u,i,center,dl,dt,dd,ol,ul,li,fieldset,form,label,legend,table,caption,tbody,tfoot,thead,tr,th,td{margin:0;padding:0;border:0;outline:0;font-size:100%;vertical-align:baseline;background:transparent}body{line-height:1}ol,ul{list-style:none}blockquote,q{quotes:none}blockquote:before,blockquote:after,q:before,q:after{content:'';content:none}:focus{outine:0}ins{text-decoration:none}del{text-decoration:line-through}table{border-collapse:collapse;border-spacing:0}
+
+html { min-height:100%; margin-bottom:1px; }
+html body { height:100%; margin:0px; font-family:Arial, Helvetica, sans-serif; font-size:10px; color:#000; line-height:140%; background:#fff none; overflow-y:scroll; }
+html body td { vertical-align:top; text-align:left; }
+
+h1 { padding:0px; margin:0px 0px 25px; font-family:Arial, Helvetica, sans-serif; font-size:1.5em; color:#d55000; line-height:100%; font-weight:normal; }
+h2 { padding:0px; margin:0px 0px 8px; font-family:Arial, Helvetica, sans-serif; font-size:1.2em; color:#000; font-weight:bold; line-height:140%; border-bottom:1px solid #d6d4d4; display:block; }
+h3 { padding:0px; margin:0px 0px 5px; font-family:Arial, Helvetica, sans-serif; font-size:1.1em; color:#000; font-weight:bold; line-height:140%; }
+
+a { color:#005fce; text-decoration:none; }
+a:hover { color:#005fce; text-decoration:underline; }
+a:visited { color:#004aa0; text-decoration:none; }
+
+p { padding:0px; margin:0px 0px 20px; }
+img { padding:0px; margin:0px 0px 20px; border:none; }
+p img, pre img, tt img, li img, h1 img, h2 img { margin-bottom:0px; } 
+
+ul { padding:0px; margin:0px 0px 20px 23px; list-style:square; }
+ul li { padding:0px; margin:0px 0px 7px 0px; }
+ul li ul { padding:5px 0px 0px; margin:0px 0px 7px 23px; }
+ul li ol li { list-style:decimal; }
+ol { padding:0px; margin:0px 0px 20px 0px; list-style:decimal; }
+ol li { padding:0px; margin:0px 0px 7px 23px; list-style-type:decimal; }
+ol li ol { padding:5px 0px 0px; margin:0px 0px 7px 0px; }
+ol li ol li { list-style-type:lower-alpha; }
+ol li ul { padding-top:7px; }
+ol li ul li { list-style:square; }
+
+.content { font-size:1.2em; line-height:140%; padding: 20px; }
+
+pre, code { font-size:12px; }
+tt { font-size: 1.2em; }
+pre { margin:0px 0px 20px; }
+pre.codeinput { padding:10px; border:1px solid #d3d3d3; background:#f7f7f7; }
+pre.codeoutput { padding:10px 11px; margin:0px 0px 20px; color:#4c4c4c; }
+pre.error { color:red; }
+
+@media print { pre.codeinput, pre.codeoutput { word-wrap:break-word; width:100%; } }
+
+span.keyword { color:#0000FF }
+span.comment { color:#228B22 }
+span.string { color:#A020F0 }
+span.untermstring { color:#B20000 }
+span.syscmd { color:#B28C00 }
+
+.footer { width:auto; padding:10px 0px; margin:25px 0px 0px; border-top:1px dotted #878787; font-size:0.8em; line-height:140%; font-style:italic; color:#878787; text-align:left; float:none; }
+.footer p { margin:0px; }
+.footer a { color:#878787; }
+.footer a:hover { color:#878787; text-decoration:underline; }
+.footer a:visited { color:#878787; }
+
+table th { padding:7px 5px; text-align:left; vertical-align:middle; border: 1px solid #d6d4d4; font-weight:bold; }
+table td { padding:7px 5px; text-align:left; vertical-align:top; border:1px solid #d6d4d4; }
+
+
+
+
+
+  </style></head><body><div class="content"><h2>Contents</h2><div><ul><li><a href="#2">NMF &#49688;&#54665;&#54616;&#44592;</a></li><li><a href="#3">PCA &#49688;&#54665;&#54616;&#44592;</a></li></ul></div><pre class="codeinput">clear; close <span class="string">all</span>; clc;
+
+load(<span class="string">'YaleB_32x32.mat'</span>); <span class="comment">% gnd&#45716; &#49324;&#46988; &#48264;&#54840;&#51064;&#46319;.</span>
+<span class="comment">% &#52636;&#52376;: http://www.cad.zju.edu.cn/home/dengcai/Data/FaceData.html</span>
+<span class="comment">% &#49324;&#50857;&#54620; &#45936;&#51060;&#53552;&#49483;&#51032; &#51060;&#47492;&#51008; Extended Yale Face Database B&#51076;.</span>
+
+figure(<span class="string">'position'</span>,[556, 237, 947, 699]);
+<span class="keyword">for</span> i= 1:25
+    subplot(5,5,i)
+    imagesc(reshape(fea(i,:), 32, 32)); colormap(<span class="string">'gray'</span>)
+<span class="keyword">end</span>
+</pre><img vspace="5" hspace="5" src="no01_NMF_on_face_dataset_01.png" alt=""> <h2 id="2">NMF &#49688;&#54665;&#54616;&#44592;</h2><pre class="codeinput">n_features = 25;
+[W, H] = nnmf(fea, n_features);
+
+figure(<span class="string">'position'</span>,[556, 237, 947, 699]);
+<span class="keyword">for</span> i_features = 1:n_features
+    subplot(5,5,i_features)
+    imagesc(reshape(H(i_features,:), 32, 32)); colormap(<span class="string">'gray'</span>);
+<span class="keyword">end</span>
+
+figure; imagesc(reshape(randn(1, 25) * H, 32, 32)); colormap(<span class="string">'gray'</span>)
+</pre><img vspace="5" hspace="5" src="no01_NMF_on_face_dataset_02.png" alt=""> <img vspace="5" hspace="5" src="no01_NMF_on_face_dataset_03.png" alt=""> <h2 id="3">PCA &#49688;&#54665;&#54616;&#44592;</h2><pre class="codeinput">[coeff, score, latent] = pca(fea);
+
+figure(<span class="string">'position'</span>,[556, 237, 947, 699]);
+<span class="keyword">for</span> i_features = 1:n_features
+    subplot(5,5,i_features)
+    imagesc(reshape(coeff(:, i_features), 32, 32)); colormap(<span class="string">'gray'</span>);
+<span class="keyword">end</span>
+</pre><img vspace="5" hspace="5" src="no01_NMF_on_face_dataset_04.png" alt=""> <p class="footer"><br><a href="http://www.mathworks.com/products/matlab/">Published with MATLAB&reg; R2017a</a><br></p></div><!--
+##### SOURCE BEGIN #####
+clear; close all; clc;
+
+load('YaleB_32x32.mat'); % gnd는 사람 번호인듯.
+% 출처: http://www.cad.zju.edu.cn/home/dengcai/Data/FaceData.html
+% 사용한 데이터셋의 이름은 Extended Yale Face Database B임.
+
+figure('position',[556, 237, 947, 699]);
+for i= 1:25
+    subplot(5,5,i)
+    imagesc(reshape(fea(i,:), 32, 32)); colormap('gray')
+end
+
+%% NMF 수행하기
+n_features = 25;
+[W, H] = nnmf(fea, n_features);
+
+figure('position',[556, 237, 947, 699]);
+for i_features = 1:n_features
+    subplot(5,5,i_features)
+    imagesc(reshape(H(i_features,:), 32, 32)); colormap('gray');
+end
+
+figure; imagesc(reshape(randn(1, 25) * H, 32, 32)); colormap('gray')
+
+%% PCA 수행하기
+
+[coeff, score, latent] = pca(fea);
+
+figure('position',[556, 237, 947, 699]);
+for i_features = 1:n_features
+    subplot(5,5,i_features)
+    imagesc(reshape(coeff(:, i_features), 32, 32)); colormap('gray');
+end
+
+##### SOURCE END #####
+--></body></html>
diff --git a/선형대수/NMF/html/no01_NMF_on_face_dataset.png b/선형대수/NMF/html/no01_NMF_on_face_dataset.png
diff --git a/선형대수/NMF/html/no01_NMF_on_face_dataset_01.png b/선형대수/NMF/html/no01_NMF_on_face_dataset_01.png
diff --git a/선형대수/NMF/html/no01_NMF_on_face_dataset_02.png b/선형대수/NMF/html/no01_NMF_on_face_dataset_02.png
diff --git a/선형대수/NMF/html/no01_NMF_on_face_dataset_03.png b/선형대수/NMF/html/no01_NMF_on_face_dataset_03.png
diff --git a/선형대수/NMF/html/no01_NMF_on_face_dataset_04.png b/선형대수/NMF/html/no01_NMF_on_face_dataset_04.png
diff --git a/선형대수/NMF/lena_std.tif b/선형대수/NMF/lena_std.tif
diff --git a/선형대수/NMF/no01_NMF_on_face_dataset.m b/선형대수/NMF/no01_NMF_on_face_dataset.m
@@ -0,0 +1,33 @@
+clear; close all; clc;
+
+load('YaleB_32x32.mat'); % gnd는 사람 번호인듯.
+% 출처: http://www.cad.zju.edu.cn/home/dengcai/Data/FaceData.html
+% 사용한 데이터셋의 이름은 Extended Yale Face Database B임.
+
+figure('position',[556, 237, 947, 699]);
+for i= 1:25
+    subplot(5,5,i)
+    imagesc(reshape(fea(i,:), 32, 32)); colormap('gray')
+end
+
+%% NMF 수행하기
+n_features = 25;
+[W, H] = nnmf(fea, n_features);
+
+figure('position',[556, 237, 947, 699]);
+for i_features = 1:n_features
+    subplot(5,5,i_features)
+    imagesc(reshape(H(i_features,:), 32, 32)); colormap('gray');
+end
+
+% figure; imagesc(reshape(randn(1, 25) * H, 32, 32)); colormap('gray')
+
+%% PCA 수행하기
+
+[coeff, score, latent] = pca(fea);
+
+figure('position',[556, 237, 947, 699]);
+for i_features = 1:n_features
+    subplot(5,5,i_features)
+    imagesc(reshape(coeff(:, i_features), 32, 32)); colormap('gray');
+end