ArtificialNN.js
JavaScript
A library for artificial neural networks in JavaScript [Programmed from scratch]. When training with 8,000 MNIST images, the network detects 7,971 [99.64%]. The library is implemented without recursions and only has 50 lines of sourcecode for feedforward und backpropagation.
Features:
- Calculates artificial neural networks [Feedforward]
- Trains artificial neural networks [Backpropagation]
- Visualizes artificial neural networks [SVG exporter]
Example:
Programming language
JavaScript [without additional libraries - from scratch]
Release date
February 2019