Simple-NeuralNetwork-Py

A very simple fully connected Neural Network implementation in Python based on the tutorials and the NN.js library by Danial Shiffman in this playlist.

If you want some resources to learn about Neural Networks, I suggest checking these playlists and book:

• Neural networks By 3Blue1Brown
• Neural Networks - The Nature of Code By The Coding Train
• Make Your Own Neural Network By Tariq Rashid

To-Do List

• Implmeneting the basic NN itself with only 1 hidden layer.
• Train the library to solve XOR problem
• Replace matrix.py with numpy
• Add more activation functions
• Train the library on the MNIST dataset
• Add support for multiple hidden layers
• Train the library to play games
  • Flappy Bird

Getting started

• Install required libraries to run the neural_network module and examples using:
  • pip install -r requirements.txt
• To run examples, run the example py from its directory, for example:
  • $ ..\Simple-NeuralNetwork-Py\examples\XOR> python xor.py

Documentation

• NeuralNetwork - The neural network class
  • predict(input_list) - Returns the output of the NeuralNetwork
  • train(input_list, output_list) - Trains the NeuralNetwork on the given input.

License

This project is licensed under the terms of the MIT license, see LICENSE.