Struct ModifiedHuber 

pub struct ModifiedHuber;

Modified Huber loss for classification.

Smooth approximation to hinge with quadratic behaviour near the margin:

L(y, p) = max(0, 1 - y*p)^2   if y*p >= -1
        = -4 * y * p            otherwise

Trait Implementations

impl Clone for ModifiedHuber

fn clone(&self) -> ModifiedHuber

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ModifiedHuber

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

Formats the value using the given formatter.

impl<F: Float> Loss<F> for ModifiedHuber

fn loss(&self, y_true: F, y_pred: F) -> F

Compute the loss for a single sample.

fn gradient(&self, y_true: F, y_pred: F) -> F

Compute the gradient of the loss with respect to y_pred.

impl Copy for ModifiedHuber