Struct neural_network::back_prop::BackProp

pub struct BackProp {
    pub parameters: LearningParameters,
    pub hidden_derivative: fn(_: f64) -> f64,
    pub output_derivative: fn(_: f64) -> f64,
    // some fields omitted
}

Fields

parameters: LearningParameters

hidden_derivative: fn(_: f64) -> f64

output_derivative: fn(_: f64) -> f64

Methods

impl BackProp

fn new(
    input_count: usize,
    hidden_counts: &[usize],
    output_count: usize,
    learning_rate: f64,
    momentum: f64,
    weight_decay: f64,
    hidden_activation: fn(_: f64) -> f64,
    hidden_derivative: fn(_: f64) -> f64,
    output_activation: fn(_: f64) -> f64,
    output_derivative: fn(_: f64) -> f64
) -> Self

fn pulse(&mut self, input: &[f64]) -> Vec<f64>

fn back_prop(&mut self, target: &[f64])

fn test(&mut self, input: &[f64], target: &[f64]) -> f64

fn train(
    &mut self,
    min_error: f64,
    max_epochs: Option<u64>,
    max_duration: Option<Duration>,
    train_inputs: &[&[f64]],
    train_targets: &[&[f64]],
    test_inputs: &[&[f64]],
    test_targets: &[&[f64]]
) -> TrainingResult

fn save<T: CopyIO>(&self, f: &mut T) -> Result<()>

fn load<T: CopyIO>(
    f: &mut T,
    hidden_activation: fn(_: f64) -> f64,
    hidden_derivative: fn(_: f64) -> f64,
    output_activation: fn(_: f64) -> f64,
    output_derivative: fn(_: f64) -> f64
) -> Result<Self>

Trait Implementations

impl PartialEq for BackProp

fn eq(&self, __arg_0: &BackProp) -> bool

This method tests for self and other values to be equal, and is used by ==.

fn ne(&self, __arg_0: &BackProp) -> bool

This method tests for !=.

impl Debug for BackProp

fn fmt(&self, __arg_0: &mut Formatter) -> Result

Formats the value using the given formatter.