Struct ritenn::NN

pub struct NN { /* fields omitted */ }

Neural network

Implementations

impl NN

pub fn new(
    layers_sizes: &[u32],
    hidden_activation: Activation,
    output_activation: Activation
) -> NN

Each number in the layers_sizes parameter specifies a layer in the network. The number itself is the number of nodes in that layer. The first number is the input layer, the last number is the output layer, and all numbers between the first and last are hidden layers. There must be at least two layers in the network. The activation function can be Sigmoid, SELU, PELU, LRELU, Linear or Tanh. Important: Take care of inputs/outputs for the individual activation functions! Do not use linear activation for hidden layers.

pub fn run(&self, inputs: &[f64]) -> Vec<f64>

Runs the network on an input and returns a vector of the results. The number of f64s in the input must be the same as the number of input nodes in the network. The length of the results vector will be the number of nodes in the output layer of the network.

pub fn train<'b>(
    &'b mut self,
    examples: &'b [(Vec<f64>, Vec<f64>)]
) -> Trainer<'_, '_>

Takes in vector of examples and returns a Trainer struct that is used to specify options that dictate how the training should proceed. No actual training will occur until the go() method on the Trainer struct is called.

Trait Implementations

impl Clone for NN

fn clone(&self) -> NN

Returns a copy of the value.

pub fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for NN

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.