format

Author: Michael Lou

display_network.py

@@ -1,7 +1,8 @@
 from math import sqrt, ceil, floor
+
+import numpy as np
 import matplotlib.pyplot as plt
 import matplotlib.cm as cm
-import numpy as np
 
 
 def display_network(data, cols=-1, opt_normalize=True, opt_graycolor=True):

sample_images.py

@@ -1,8 +1,7 @@
-__author__ = 'ylou'
+from random import randint
 
-import scipy.io
+from scipy.io import loadmat
 import numpy as np
-from random import randint
 #from matplotlib import pyplot as plt
 
 
@@ -31,7 +30,7 @@ def sample_images():
     :return: 10000 patches for training
     """
     ## Get IMAGES.mat from http://ufldl.stanford.edu/wiki/resources/sparseae_exercise.zip
-    images = scipy.io.loadmat('IMAGES.mat')['IMAGES']    # load images from disk
+    images = loadmat('IMAGES.mat')['IMAGES']    # load images from disk
     num_images = images.shape[2]
 
     patch_size = 8          # we'll use 8x8 patches
@@ -58,4 +57,4 @@ def sample_images():
     # the range of pixel values is also bounded between [0,1]
     patches = normalize_data(patches)
 
-    return patches
+    return patches

sparse_autoencoder_cost.py

@@ -1,4 +1,3 @@
-
 import numpy as np
 
 
@@ -65,4 +64,4 @@ def sparse_autoencoder_cost(theta, visible_size, hidden_size, decay_lambda, spar
 
     grad = np.concatenate((w1grad.flatten(), w2grad.flatten(), b1grad.flatten(), b2grad.flatten()))
 
-    return cost, grad
+    return cost, grad

train.py

@@ -1,11 +1,11 @@
-__author__ = 'ylou'
+from random import randint
+from math import sqrt
+
+import numpy as np
 
 from sample_images import sample_images
 from display_network import display_network
 from sparse_autoencoder_cost import sparse_autoencoder_cost
-from math import sqrt
-import random
-import numpy as np
 
 
 def initialize_parameters(hidden_size, visible_size):
@@ -43,7 +43,7 @@ def train():
     # display a random sample of 200 patches from the dataset
 
     patches = sample_images()
-    list = [random.randint(0, patches.shape[0]-1) for i in xrange(64)]
+    list = [randint(0, patches.shape[0]-1) for i in xrange(64)]
     display_network(patches[list, :], 8)
 
     # Obtain random parameters theta
@@ -81,4 +81,4 @@ def train():
 
 
 if __name__ == "__main__":
-    train()
+    train()