NNLearn
This project aims to develop a learning prototype for various concepts involved in Neural Network and Deep Learning
Usage
Project follows various components with various stages of development. At present, following components exist:
- Simple demonstration for
backpropagation algorithm
- A sample experiment to classify digits based on MNIST dataset:
digit_demo.py
Dependencies
- Python 3+
- numpy
TODO
- Converting to a functional library
- Adding features such as Regularisation, Momentum, Dropout