Struct Network

pub struct Network<const X: usize, const H: usize, const Y: usize> { /* private fields */ }

Holds a neural network.

Implementations

impl<const X: usize, const H: usize, const Y: usize> Network<X, H, Y>

pub fn new( input_activation: Box<dyn Activation<H>>, output_activation: Box<dyn Activation<Y>>, learning_rate: f64, ) -> Self

Constructs a new neural network.

pub fn add_hidden_layer(&mut self, activation: Box<dyn Activation<H>>)

Add a hidden layer.

pub fn evaluate(&mut self, vector: Vector<X>) -> Vector<Y>

Evaluates a vector input.

pub fn train_once(&mut self, input: Vector<X>, expected: Vector<Y>) -> f64

Train this network based on an input and an expected output.

pub fn train_all(&mut self, dataset: &TrainingDataset<X, Y>) -> f64

Train this network based on a list of inputs and expected outputs.

pub fn train(&mut self, dataset: &TrainingDataset<X, Y>, generations: usize)

Train this network based on a dataset for a given number of generations.