Struct SupportSetManager 

pub struct SupportSetManager<T: Float + Debug + Send + Sync + 'static> { /* private fields */ }

Support set manager

Implementations

impl<T: Float + Debug + Send + Sync + 'static> SupportSetManager<T>

pub fn select_support_set( &self, candidates: &[Array1<T>], _labels: &[T], budget: usize, ) -> Result<Vec<usize>>

Select a diverse subset of candidate indices for a support set.

Uses a greedy farthest-point sampling strategy: the first point is selected as the one with the largest norm, then each subsequent point is the one that is most distant from all already-selected points (measured by squared Euclidean distance).

pub fn augment_support_set( &self, support: &[Array1<T>], noise_scale: T, ) -> Result<Vec<Array1<T>>>

Augment a support set by adding Gaussian noise to each example.

For each input vector, produces a copy with noise ~ N(0, noise_scale^2) added to each element (using a simple deterministic hash-based approach for reproducibility without requiring rand).

pub fn evaluate_quality(&self, support: &[Array1<T>]) -> Result<T>

Evaluate the quality/diversity of a support set.

Returns the average pairwise squared Euclidean distance between all support vectors (higher = more diverse = better quality).

impl<T: Float + Debug + Send + Sync + 'static> SupportSetManager<T>

pub fn new(config: SupportSetManagerConfig) -> Result<Self>

Create a new support set manager

pub fn from_max_size(max_support_size: usize) -> Result<Self>

Create from max support size (convenience)

pub fn max_support_size(&self) -> usize

Get the max support size

pub fn config(&self) -> &SupportSetManagerConfig

Get the config