Struct linfa::composing::platt_scaling::PlattParams

pub struct PlattParams<F, O>(_);

Implementations

impl<F: Float, O> PlattParams<F, O>

pub fn maxiter(self, maxiter: usize) -> Self

Set the maximum number of iterations in the optimization process

The Newton’s method is an iterative optimization process, which uses the first and second order gradients to find optimal A and B. This function caps the maximal number of iterations.

pub fn minstep(self, minstep: F) -> Self

Set the minimum stepsize in the line search

After estimating the Hessian matrix, a line search is performed to find the optimal step size in each optimization step. In each attempt the stepsize is halfed until this threshold is reached. After reaching the threshold the algorithm fails because the desired precision could not be achieved.

pub fn sigma(self, sigma: F) -> Self

Set the Hessian’s sigma value

The Hessian matrix is regularized with H’ = H + sigma I to avoid numerical issues. This function set the amount of regularization.

Trait Implementations

impl<F: Float, O> Default for PlattParams<F, O>

fn default() -> Self

Returns the “default value” for a type.

impl<F: Float, O> ParamGuard for PlattParams<F, O>

type Checked = PlattValidParams<F, O>

The checked hyperparameters

type Error = PlattError

Error type resulting from failed hyperparameter checking

fn check_ref(&self) -> Result<&Self::Checked, PlattError>

Checks the hyperparameters and returns a reference to the checked hyperparameters if successful

fn check(self) -> Result<Self::Checked, PlattError>

Checks the hyperparameters and returns the checked hyperparameters if successful

fn check_unwrap(self) -> Self::Checked where
    Self: Sized, 

Calls check() and unwraps the result