Add bare bones implementations

.gitignore

@@ -65,4 +65,4 @@ target/
 .DS_Store
 
 # PyCharm
-idea/
+.idea/

__init__.py

@@ -0,0 +1 @@
+__author__ = 'sidd'

langmod_nn/__init__.py

@@ -0,0 +1 @@
+__author__ = 'sidd'

mnist_cnn/mnist_cnn.py

@@ -0,0 +1,118 @@
+"""
+deep_mnist.py
+
+Code for following Tensorflow Tutorial.
+"""
+from tensorflow.examples.tutorials.mnist import input_data
+import tensorflow as tf
+
+# Load MNIST Data
+mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
+
+# Start Tensorflow Session
+sess = tf.InteractiveSession()
+
+# Build Input/Output placeholders
+x = tf.placeholder(tf.float32, shape=[None, 784])
+y_ = tf.placeholder(tf.float32, shape=[None, 10])
+
+# # Single Layer Model
+#
+# # Set up Model Parameters
+# W = tf.Variable(tf.zeros([784, 10]))
+# b = tf.Variable(tf.zeros([10]))
+#
+# # Initialize all variables in the session
+# sess.run(tf.initialize_all_variables())
+#
+# # Actually implement the model
+# y = tf.nn.softmax(tf.matmul(x, W) + b)
+#
+# # Set up loss function
+# cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1]))
+#
+# # Start the training process
+# train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)
+#
+# # Train the model
+# for i in range(1000):
+#     batch = mnist.train.next_batch(50)
+#     train_step.run(feed_dict={x: batch[0], y_: batch[1]})
+#
+# # Set up evaluations
+# correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1))
+# accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
+
+# Evaluate model
+# print "Single Layer Accuracy:", accuracy.eval(feed_dict={x: mnist.test.images, y_: mnist.test.labels})
+
+
+# Convolution Model
+# Define variables
+def weight_variable(shape):
+    initial = tf.truncated_normal(shape, stddev=0.1)
+    return tf.Variable(initial)
+
+def bias_variable(shape):
+    initial = tf.constant(0.1, shape=shape)
+    return tf.Variable(initial)
+
+# Define special layers
+def conv2d(x, W):
+    return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
+
+def max_pool_2x2(x):
+    return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
+
+# Build model
+# Reshape Input
+x_image = tf.reshape(x, [-1, 28, 28, 1])
+
+# Convolution + Max Pool (1)
+W_conv1 = weight_variable([5, 5, 1, 32])
+b_conv1 = bias_variable([32])
+
+h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
+h_pool1 = max_pool_2x2(h_conv1)
+
+# Convolution + Max Pool (2)
+W_conv2 = weight_variable([5, 5, 32, 64])
+b_conv2 = bias_variable([64])
+
+h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
+h_pool2 = max_pool_2x2(h_conv2)
+
+# Dense Layer
+W_fc1 = weight_variable([7 * 7 * 64, 1024])
+b_fc1 = bias_variable([1024])
+
+h_pool2_flat = tf.reshape(h_pool2, [-1, 7 * 7 * 64])
+h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
+
+# Dropout
+keep_prob = tf.placeholder(tf.float32) # Placeholder lets us turn on during train, off during test
+h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
+
+# Softmax Layer
+W_fc2 = weight_variable([1024, 10])
+b_fc2 = bias_variable([10])
+
+y_conv = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)
+
+# Train and Evaluate
+cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y_conv), reduction_indices=[1]))
+train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
+correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
+accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
+sess.run(tf.initialize_all_variables())
+
+for i in range(2000):
+    batch = mnist.train.next_batch(50)
+    if i % 100 == 0:
+        train_accuracy = accuracy.eval(feed_dict={x: batch[0], y_: batch[1], keep_prob: 1.0})
+        print("Step %d, training accuracy %g" % (i, train_accuracy))
+
+    train_step.run(feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})
+
+print("Test accuracy: %g" % accuracy.eval(feed_dict={x: mnist.test.images, y_: mnist.test.labels,
+                                                     keep_prob: 1.0}))

mnist_nn/README.md

@@ -0,0 +1,28 @@
+# MNIST NN #
+Simple MNIST Handwriting Classification Problem. Given a 28 x 28 pixel image, pick which digit class
+it belongs to.
+
+## Model Setup ##
+
+    + Input: 28 x 28 Pixel Image, flattened into a single vector of size 784. The input "x" is then
+             a matrix of size (Batch-Size, 784).
+
+    + Output: Vector of size 10, corresponding to the probability that the given image belongs to
+              each of the 10 digit classes (0, 1, 2, ... 9). The output "y" is then a matrix of 
+              size (Batch-Size, 10).
+              
+    + Layers: The model only has one fully-connected feed-forward layer, with a bias vector. This
+              layer is followed by a Softmax Layer.
+              
+        1) Feed-Forward Layer: This is represented by a weight matrix "W" of size (784, 10). 
+                              The bias vector is of size (10). The transformation is of the form 
+                              (xW + b), a simple matrix multiplication between the input and the 
+                              weight matrix, and the addition of the bias.
+         
+        2) Softmax Layer: Normal softmax over the output of size (10). Represents probability that
+                          the current example image belongs to each of the 10 digit classes.
+
+    + Loss: We use cross-entropy loss. The true predictions are encoded as one-hot versions of the 
+            true classes (vectors of size 10).
+            
+    + Optimizer: Simple SGD Optimizer with learning rate 0.5.

mnist_nn/mnist_nn.py

@@ -0,0 +1,47 @@
+"""
+mnist_nn.py
+
+Code for MNIST Tensorflow Tutorial. Builds a simple one layer network with a simple softmax.
+"""
+from tensorflow.examples.tutorials.mnist import input_data
+import tensorflow as tf
+
+# Load MNIST Data
+mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
+
+# Start Tensorflow Session
+sess = tf.InteractiveSession()
+
+# Build Input/Output placeholders
+x = tf.placeholder(tf.float32, shape=[None, 784])  # X is input, of shape (Batch, 784)
+y_ = tf.placeholder(tf.float32, shape=[None, 10])  # Y is output, of shape (Batch, 10)
+
+# Single Layer Model
+
+# Set up Model Parameters
+W = tf.Variable(tf.zeros([784, 10]))
+b = tf.Variable(tf.zeros([10]))
+
+# Initialize all variables in the session
+sess.run(tf.initialize_all_variables())
+
+# Actually implement the model
+y = tf.nn.softmax(tf.matmul(x, W) + b)
+
+# Set up loss function
+cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1]))
+
+# Start the training process
+train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)
+
+# Train the model
+for i in range(1000):
+    batch = mnist.train.next_batch(50)
+    train_step.run(feed_dict={x: batch[0], y_: batch[1]})
+
+# Set up evaluations
+correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
+accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
+
+# Evaluate model
+print "Single Layer Accuracy:", accuracy.eval(feed_dict={x: mnist.test.images, y_: mnist.test.labels})