Skip to content

Latest commit

 

History

History
36 lines (23 loc) · 973 Bytes

README.md

File metadata and controls

36 lines (23 loc) · 973 Bytes
Raw

Dimag

What is it?

Dimag is a löve library that provides a simple Neural Network implementation. Use it for whatever you feel like.

Installatiom

Copy and paste the dimag folder in your löve project and do,

NeuralNetwork = require("dimag").NeuralNetwork

API Documentation

See docs/NeuralNetwork.md

Example

See main.lua in the repository root for an commented example of a flappy bird game being played by AI

License

This library is licensed under the MIT license, do whatever you want with it.

To Do

  • Implement crossover.
  • Think about a Population class.
  • Add backpropagation (supervised learning).
  • Add a score counter to the flappy bird example.
  • Add more activation functions.
  • Add unit tests for the matrix class.
  • Implement random.randNormal in lua, so this library can work without löve.
  • Add support for easy visualization of NN. (?)