<ann.py> Built a neural network for deep learning from scratch without using any machine learning libraries. Introducing gradient desent, forward/backward propagation.
<model.py> A binary classification instance to use the library. Also introduced a comparision of prediction accurarcy with simple deep learning model in tensorflow. Result: same # of hidden layers, # of neurons each layer, cost function and activation functions, the model created from ann.py library has the same accuracy as tensorflow model(actually 1% higher sometimes).