Struct OneClassSVM 

pub struct OneClassSVM<F, K> {
    pub nu: F,
    pub kernel: K,
    pub tol: F,
    pub max_iter: usize,
    pub cache_size: usize,
    pub shrinking: bool,
}

One-Class SVM for novelty detection.

Learns a decision boundary around the training data. New points are classified as inliers (+1) or outliers (-1).

Type Parameters

• F: The floating-point type (f32 or f64).
• K: The kernel type (e.g., RbfKernel).

Fields

nu: F

The nu parameter: upper bound on the fraction of outliers. Must be in (0, 1]. Default: 0.5.

kernel: K

The kernel function.

tol: F

Convergence tolerance.

max_iter: usize

Maximum number of SMO iterations. 0 is the sklearn max_iter=-1 sentinel (“no iteration limit”; libsvm runs to convergence) — the SMO loop then runs until the KKT gap closes; a non-zero value caps the iteration count (sklearn/svm/_classes.py:1721, max_iter default -1).

cache_size: usize

Size of the kernel evaluation cache (sklearn cache_size, default 200). Accepted for API parity; this module has no kernel cache.

shrinking: bool

Whether to use libsvm’s shrinking heuristic (sklearn shrinking, _classes.py:1718, default true).

Accepted for API parity. The one-class optimum is shrinking-invariant and ferrolearn’s SMO implements no shrinking heuristic, so this flag DOES NOT alter the fitted α/dual_coef_/intercept_ (R-DEV-7).

Implementations

impl<F: Float, K: Kernel<F>> OneClassSVM<F, K>

pub fn new(kernel: K) -> Self

Create a new OneClassSVM with the given kernel and default hyperparameters.

Defaults: nu = 0.5, tol = 1e-3, max_iter = 0 (= sklearn -1, no iteration limit — runs to convergence), cache_size = 200, shrinking = true (sklearn/svm/_classes.py:1712-1721).

pub fn with_nu(self, nu: F) -> Self

Set the nu parameter.

pub fn with_tol(self, tol: F) -> Self

Set the convergence tolerance.

pub fn with_max_iter(self, max_iter: usize) -> Self

Set the maximum number of SMO iterations.

pub fn with_cache_size(self, cache_size: usize) -> Self

Set the kernel cache size.

pub fn with_shrinking(self, shrinking: bool) -> Self

Set the shrinking flag (sklearn shrinking, default true).

Accepted for API parity; the one-class optimum is shrinking-invariant (ferrolearn’s SMO has no shrinking heuristic — R-DEV-7), so this does not change the fit.

Trait Implementations

impl<F: Clone, K: Clone> Clone for OneClassSVM<F, K>

fn clone(&self) -> OneClassSVM<F, K>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<F: Debug, K: Debug> Debug for OneClassSVM<F, K>

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

Formats the value using the given formatter.

impl<F: Float + Send + Sync + ScalarOperand + 'static, K: Kernel<F> + 'static> Fit<ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>, ()> for OneClassSVM<F, K>

fn fit( &self, x: &Array2<F>, _y: &(), ) -> Result<FittedOneClassSVM<F, K>, FerroError>

Fit the One-Class SVM.

Errors

• FerroError::InvalidParameter if nu is not in (0, 1].
• FerroError::InsufficientSamples if no training data is provided.

type Fitted = FittedOneClassSVM<F, K>

The fitted model type returned by fit.

type Error = FerroError

The error type returned by fit.