P1: MNIST with TensorFlow

a) 1 linear layer, followed by a softmax

code: tf_P1_a.py

b) 1 hidden layer (128 units) with a ReLU non-linearity, followed by a softmax

code: tf_P1_b.py

c) 2 hidden layers (256 units) each, with ReLU non-linearity, follow by a softmax

code: tf_P1_c.py

d) 3 layer convolutional model (2 convolutional layers followed by max pooling) + 1 non-linear layer (256 units), followed by softmax.

code: tf_P1_c.py

P2: MNIST without TensorFlow

a) Derivations in report

b) Implement and train the model in (P1:a)

code: tf_P2_a.py

c) Implement and train the model in (P1:b)

code: tf_P2_b.py

d) Implement and train the model in (P1:c)

code: tf_P2_c.py

All the final trained weights can be found in the trained_weights folder

Setup to run source code

1. Install TensorFlow on Anaconda environment (gpu version prefered for speed of execution), setup for windows
2. Install numpy, sklearn, matplotlib if not installed by default.
3. Activate tensforflow environment. e.g. activate tensorflow-gpu
4. Navigate to source code directory and run each python file. (pycharm prefered)

Note: Each python file has a flag called TRAIN_MODE. By default it is set to False to run code in TEST mode with saved parameters. You can toggle it to run in TRAIN mode
P.S. The saved parameters have been trained using tensorflow-gpu version r0.12.