Struct LearningVectorQuantization 

pub struct LearningVectorQuantization {
    pub n_prototypes_per_class: usize,
    pub lr_init: f64,
    pub lr_final: f64,
    pub max_epochs: usize,
    pub seed: u64,
}

Supervised prototype learning via LVQ-1.

Each prototype is associated with a class label. For every training sample the nearest prototype is found:

• If it carries the correct class label it is moved toward the sample.
• Otherwise it is moved away from the sample.

Prototypes are initialised by sampling one or more points per class.

Fields

n_prototypes_per_class: usize

Number of prototypes per class. Default 1.

lr_init: f64

Initial learning rate. Default 0.1.

lr_final: f64

Final learning rate (annealed). Default 0.001.

max_epochs: usize

Number of training epochs. Default 50.

seed: u64

RNG seed. Default 42.

Implementations

impl LearningVectorQuantization

pub fn new( n_prototypes_per_class: usize, lr_init: f64, lr_final: f64, max_epochs: usize, seed: u64, ) -> Self

Create a new LVQ instance with all hyperparameters.

pub fn fit( &self, data: ArrayView2<'_, f64>, labels: &[usize], ) -> Result<LvqModel>

Train LVQ-1 on labelled data.

Arguments

• data – Feature matrix (n_samples, n_features).
• labels – Integer class labels; values must be in 0..n_classes.

Returns

A trained LvqModel.

pub fn predict(model: &LvqModel, data: ArrayView2<'_, f64>) -> Array1<usize>

Convenience: predict class labels for all rows in data using a trained model.

Trait Implementations

impl Clone for LearningVectorQuantization

fn clone(&self) -> LearningVectorQuantization

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LearningVectorQuantization

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

Formats the value using the given formatter.

impl Default for LearningVectorQuantization

fn default() -> Self

Returns the “default value” for a type.