NeuralNetwork.md

Class NeuralNetwork

@Constructor NeuralNetwork(input, hiddens, output)

Creates a new neural network with input number of input neurons, output number of output neurons and #hiddens number of hidden layers with hidden[n] neurons in each layer.
All weights and biases are randomized on network creation.
For example, NeuralNetwork(4, {5, 6}, 1) will create a NeuralNetwork with 4 input neurons, 2 hidden layers with 5 neurons in the first and 6 neurons in the second layer and 1 output neuron.

@Static NeuralNetwork.deserialize(data)

Static method that creates a new NeuralNetwork from given data.

@Method NeuralNetwork:serialize()

Serializes the network into json format and returns the json data that can be saved to disk and loaded by .deserialize(data)

@Method NeuralNetwork:feedForward(inputArr)

Takes in an array of numerical inputs and returns an array of numerical outputs.

@Method NeuralNetwork:mutate(rate)

Mutate the weights and biases of the network with the given rate between 0 and 1. (A rate of 0.1 means 10% mutation rate) Returns the network itself for possible chaining.

@Method NeuralNetwork:copy()

Returns a copy of the network.