This project includes implementation of a simple neural network for classification tasks and backpropagation mechanism.
The main building blocks are the following modules:
• Tests folder - includes gradient based tests for backpropagation mechanism.
• Neural Network.py - standard neural network.
• function.py - includes all activation functions and its gradients.
• loss function.py - objective loss function "soft-max" and its gradient.
• optimizer.py - implementation of stochastic gradient decent optimizer.
• train.py - implementation of training functions.
• Utils.py - util functions.
• main.py - full network experiments implementation.