A simple neural network to run with Node.js.

In order to learn some of basic concepts of Artificial Neural Networks I decided to build my own one from scratch.

This isn't meant to be a highly efficient replacement for other neural networks, rather a demonstration of the core principles involved.

• Setup Multi-layer Networks
• Testing via Feed Forward
• Training via Back Propagation
• Built in Matrix math
• NeuroEvolution (genetic algorithm to evaluate & breed groups of networks)
• Zero dependency

Simple! Just clone the repo:

git clone https://github.com/BenShelton/node-nn.git

There are some examples setup for you to try using npm run or yarn:

yarn xor
yarn mnist
yarn xor-evo
yarn mnist-evo

If you want to create your own, then you can import the necessary files from the library:

const { NeuralNetwork, NeuroEvolution } = require('./lib')

Feel free to raise an issue or create a pull request. Contributions are always welcome!

• Implement Batch Gradient Descent
• Create documentation for creating your own
• Update examples
• Go through & comment on each stage of code
• Add testing for NeuralNetwork, Layer & NeuroEvolution