Struct OnlineGradientDescent 

pub struct OnlineGradientDescent { /* private fields */ }

Online Gradient Descent for convex losses.

The learning rate schedule follows η_t = η_0 / √(t + 1) when lr_decay > 0, otherwise a constant η_0 is used. Optional L2 regularisation applies weight decay at each step.

Implementations

impl OnlineGradientDescent

pub fn new(n_features: usize, loss: OGDLoss) -> Self

Create a new OGD learner for the given loss function.

pub fn with_lr(self, lr: f64) -> Self

Set the initial learning rate η_0.

pub fn with_l2(self, lambda: f64) -> Self

Set the L2 regularisation coefficient λ.

pub fn with_lr_decay(self, decay: f64) -> Self

Enable learning rate decay. When decay > 0, η_t = η_0 / √(t + 1).

pub fn stats(&self) -> &OnlineStats

Reference to running statistics.

Trait Implementations

impl Clone for OnlineGradientDescent

fn clone(&self) -> OnlineGradientDescent

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for OnlineGradientDescent

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

Formats the value using the given formatter.

impl OnlineLearner for OnlineGradientDescent

fn update( &mut self, features: &[f64], label: f64, ) -> Result<OnlineUpdateResult, OnlineError>

Update model on a single (features, label) pair. Returns update stats.

fn predict(&self, features: &[f64]) -> Result<f64, OnlineError>

Predict class label or regression value for features.

fn n_updates(&self) -> usize

Number of updates seen so far.

fn weights(&self) -> &[f64]

Current weight vector (without bias).