Skip to content

Latest commit

 

History

History
44 lines (22 loc) · 1.64 KB

README.md

File metadata and controls

44 lines (22 loc) · 1.64 KB

NeuralNetwork

A simple neural network implementation which you can train easily.

To try this you must set --experimental-modules argument to node.

Two examples

xor

xor.mjs

Teaches the NN to XOR. The input is an array of two bits, the output Neuron will converge to the right answer:P

alphabet classifier

characterClassifier.mjs

You tell the NN which character is a number, consonant or a vowel, then it tries to guess it. There are 3 Neurons at the output, each of them contains the chance for that particular type.

Resources

Reference implementation Yes, I stole it from there:P

What is bias

How to build a neural network

Backpropagation

Another tutorial for noobs

How neurons work

Neural network in JS with synaptic

NN from scratch, really good

MLNN

Sigmoid function