Struct MaximumMeanDiscrepancyDetector 

pub struct MaximumMeanDiscrepancyDetector { /* private fields */ }

Maximum Mean Discrepancy (MMD) for multivariate drift detection.

Uses a Gaussian (RBF) kernel with bandwidth sigma. The biased MMD² estimator is:

MMD²(P,Q) = E[k(x,x')] - 2E[k(x,y)] + E[k(y,y')]

where x,x’ ~ P and y,y’ ~ Q.

For multivariate data, each sample is a slice of length dim.

Implementations

impl MaximumMeanDiscrepancyDetector

pub fn new(dim: usize, sigma: f64, threshold: f64) -> Result<Self>

Create a new MMD detector.

• dim – dimensionality of each sample
• sigma – RBF kernel bandwidth
• threshold – MMD² value above which drift is declared

pub fn detect_multivariate( &self, reference: &[f64], test: &[f64], ) -> Result<DriftResult>

Detect drift on multivariate data.

reference and test are flattened arrays where every dim consecutive elements form one sample.

Trait Implementations

impl Clone for MaximumMeanDiscrepancyDetector

fn clone(&self) -> MaximumMeanDiscrepancyDetector

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MaximumMeanDiscrepancyDetector

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

Formats the value using the given formatter.

impl DriftDetector for MaximumMeanDiscrepancyDetector

fn detect(&self, reference: &[f64], test: &[f64]) -> Result<DriftResult>

For the 1-D DriftDetector trait, each element is treated as a 1-D sample.