Nenet

vectorized neural network library for browser and node

Easy to use

const x = [[2, 4, 6], 
           [3, 6, 9]];

const y = [[0.25, 0.5, 0.7], 
           [0.5, 0.7, 0.9]];

const nn = new Nenet([
		["input", 3],
		["hidden", 6],
		["hidden", 6],
		["hidden", 6],
		["output", 3]
	])
	.options({
		dataSet: {x,y}
	})
	.train(100);

console.log(nn.y_pred);

Examples

Please check the examples folder

1. iris classification

2. mnist classification

3. house price regression

4. boolean operations XOR, NXOR, OR, AND

Roadmap

Seperate neural network and trainer -

Implement online and mini-batch training +

Ipmlement momentum -

Lab

• Web app (redux, react, d3, nenet)
• Visualize neural network and workflow.
• Create, run
• Plot error and weights.
• Visualize propagated errors