modification on the build of P in order to make it work with multiple dimensions

main/convnet_random.py → main/deepfriedConvnetMnist.py

@@ -42,7 +42,7 @@ def max_pool_2x2(x):
                           strides=[1, 2, 2, 1], padding='SAME')
 
 
-def convolution(input):
+def convolution_mnist(input):
     with tf.name_scope("conv_pool_1"):
         # 32 is the number of filter we'll use. e.g. the number of different
         # shapes this layer is able to recognize
@@ -118,7 +118,7 @@ def P_variable(d, nbr_stack):
     :type nbr_stack: int
     :return: tf.Variable object containing the matrix
     """
-    idx = [(i * d) + np.random.permutation(d) for i in range(nbr_stack)]
+    idx = np.hstack([(i * d) + np.random.permutation(d) for i in range(nbr_stack)])
     P = np.random.permutation(np.eye(N=nbr_stack * d))[idx].astype(np.float32)
     return tf.Variable(P, name="P", trainable=False)
 
@@ -259,21 +259,29 @@ def fully_connected(conv_out):
     return h_fc1
 
 
+def mnist_dims():
+    input_dim = int(mnist.train.images.shape[1])
+    output_dim = int(mnist.train.labels.shape[1])
+    return input_dim, output_dim
+
 if __name__ == '__main__':
     SIGMA = 5.0
     print("Sigma = {}".format(SIGMA))
 
     with tf.Graph().as_default():
-        input_dim = int(mnist.train.images.shape[1])
-        output_dim = int(mnist.train.labels.shape[1])
-        side_size = int(np.sqrt(input_dim))
+        # todo parametrize datset
+        input_dim, output_dim = mnist_dims()
+
         x = tf.placeholder(tf.float32, shape=[None, input_dim], name="x")
         y_ = tf.placeholder(tf.float32, shape=[None, output_dim], name="labels")
+
+        # side size is width or height of the images
+        side_size = int(np.sqrt(input_dim))
         x_image = tf.reshape(x, [-1, side_size, side_size, 1])
         tf.summary.image("digit", x_image, max_outputs=3)
 
         # Representation layer
-        h_conv = convolution(x_image)
+        h_conv = convolution_mnist(x_image)
         # h_conv = x
         # out_fc = fully_connected(h_conv)  # 95% accuracy
         # out_fc = tf.nn.relu(fast_food(h_conv, SIGMA, nbr_stack=1))  # 83% accuracy (conv) | 56% accuracy (noconv)
@@ -288,8 +296,8 @@ if __name__ == '__main__':
             keep_prob = tf.placeholder(tf.float32, name="keep_prob")
             h_fc1_drop = tf.nn.dropout(out_fc, keep_prob)
             dim = np.prod([s.value for s in h_fc1_drop.shape if s.value is not None])
-            W_fc2 = weight_variable([dim, 10])
-            b_fc2 = bias_variable([10])
+            W_fc2 = weight_variable([dim, output_dim])
+            b_fc2 = bias_variable([output_dim])
             tf.summary.histogram("weights", W_fc2)
             tf.summary.histogram("biases", b_fc2)