Skip to content

levithomason/anny

Folders and files

NameName
Last commit message
Last commit date

Latest commit

86fd05a · Dec 16, 2016
Dec 15, 2016
Dec 15, 2016
Dec 15, 2016
Dec 16, 2016
Dec 15, 2016
Dec 15, 2016
Oct 29, 2015
Jul 4, 2015
Dec 15, 2016
Dec 16, 2016
Dec 15, 2016
Dec 15, 2016
Dec 15, 2016
Oct 14, 2015
Dec 17, 2015
Dec 15, 2016
Jul 8, 2015
Dec 16, 2016
Dec 15, 2016
Dec 16, 2016

Repository files navigation

CircleCI Test Coverage

Anny is an artificial neural network, yo!

npm install anny --save

Get Started

Read the documentation or try the demo.

Train a multilayer perceptron to approximate an OR logic gate:

const {
  Network,
  Layer,
  Trainer,
  DATA,
} = require('anny')

const network = new Network([new Layer(2), new Layer(1)])
const trainer = new Trainer()

trainer.train(network, DATA.ORGate)

network.activate([0, 0]) // => 0.000836743108
network.activate([0, 1]) // => 0.998253857294

Why Anny?

I wanted to understand machine learning.

What I cannot create, I do not understand. Know how to solve every problem that has been solved. — Richard Feynman

Inspirations

  1. Watching DeepMind's DQN teach itself to play Atari.
  2. Playing with WaterWorld, a DQN implementation in JavaScript (REINFORCEjs).
  3. Realizing DeepMind's work was public, and this can be done in a browser.

Libraries

Notes

My Google Doc notes from neuralnetworksanddeeplearning.com
Good Weight Initializations
Backpropagation
Efficient Backpropagation Equation references in the source code point to this doc.
Convolution Networks