Struct neuronika::nn::Dropout

pub struct Dropout {
    pub status: Rc<Cell<bool>>,
    pub p: f64,
}

During training, randomly zeroes some of the elements of self with probability p using samples from a Bernoulli distribution. Each channel will be zeroed out independently on every forward call.

This has proven to be an effective technique for regularization and preventing the co-adaptation of neurons as described in the paper Improving neural networks by preventing co-adaptation of feature detectors.

Furthermore, the outputs are scaled by a factor of 1/(1 - p) during training. This means that during evaluation the resulting variable simply computes an identity function.

Fields

status: Rc<Cell<bool>>

p: f64

Implementations

impl Dropout

pub fn new(p: f64) -> Self

Creates a dropout layer.

Arguments

p - probability of an element to be zeroed.

pub fn forward<I: DropoutInput>(&self, input: I) -> I::Output

Applies the dropout to the variable in input.

Arguments

input  - variable in input to the layer.

Trait Implementations

impl Eval for Dropout

fn eval(&self)

Sets self in evaluation mode.

fn train(&self)

Sets self in training mode.

impl Register for Dropout

fn register_status(&mut self, status: Rc<Cell<bool>>)

Register self’s status to the model’s status state status.

fn register_params(&self, _: &mut Vec<RawParam>)

Registers self’s parameters to the model’s status parameters params.