Struct Mixup 

pub struct Mixup { /* private fields */ }

Mixup data augmentation (Zhang et al., 2018). Creates virtual training examples: x’ = λx_i + (1-λ)x_j

Implementations

impl Mixup

pub fn new(alpha: f32) -> Self

Create new Mixup with alpha parameter for Beta distribution.

pub fn sample_lambda(&self) -> f32

Sample mixing coefficient from Beta(alpha, alpha).

pub fn mix_samples( &self, x1: &Vector<f32>, x2: &Vector<f32>, lambda: f32, ) -> Vector<f32>

Mix two samples: x’ = λ*x1 + (1-λ)*x2

pub fn mix_labels( &self, y1: &Vector<f32>, y2: &Vector<f32>, lambda: f32, ) -> Vector<f32>

Mix labels: y’ = λ*y1 + (1-λ)*y2

pub fn alpha(&self) -> f32

Trait Implementations

impl Clone for Mixup

fn clone(&self) -> Mixup

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Mixup

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

Formats the value using the given formatter.