Struct CompetitiveNeuralClusterer 

pub struct CompetitiveNeuralClusterer { /* private fields */ }

Bio-inspired competitive learning for spatial clustering

This clusterer uses winner-take-all dynamics with lateral inhibition to discover clusters in spatial data. Neurons compete for activation, with the winner (neuron with strongest response) being updated while others are inhibited.

Features

• Winner-take-all competitive dynamics
• Lateral inhibition for neural competition
• Adaptive learning rates
• Neighborhood function for topological organization
• Distance-based neuron activation

Example

use scirs2_core::ndarray::Array2;
use scirs2_spatial::neuromorphic::algorithms::CompetitiveNeuralClusterer;

let points = Array2::from_shape_vec((4, 2), vec![
    0.0, 0.0, 1.0, 0.0, 0.0, 1.0, 1.0, 1.0
]).unwrap();

let mut clusterer = CompetitiveNeuralClusterer::new(2, 2);
let assignments = clusterer.fit(&points.view(), 100).unwrap();
println!("Cluster assignments: {:?}", assignments);

Implementations

impl CompetitiveNeuralClusterer

pub fn new(num_clusters: usize, input_dims: usize) -> Self

Create new competitive neural clusterer

Arguments

• num_clusters - Number of clusters to discover
• input_dims - Number of input dimensions

Returns

A new CompetitiveNeuralClusterer with randomly initialized neurons

pub fn with_competition_params(self, sigma: f64, wta_threshold: f64) -> Self

Configure neighborhood parameters

Arguments

• sigma - Neighborhood function width
• wta_threshold - Winner-take-all threshold

pub fn fit( &mut self, points: &ArrayView2<'_, f64>, epochs: usize, ) -> SpatialResult<Array1<usize>>

Train competitive network on spatial data

Applies competitive learning dynamics where neurons compete for activation and the winner adapts towards the input pattern while inhibiting neighbors.

Arguments

• points - Input spatial points (n_points × n_dims)
• epochs - Number of training epochs

Returns

Cluster assignments for each input point

pub fn get_cluster_centers(&self) -> Array2<f64>

Get cluster centers (neuron weights)

Returns

Array containing the current neuron weight vectors as cluster centers

pub fn learning_rates(&self) -> &[f64]

Get current learning rates

pub fn inhibition_strengths(&self) -> &Array2<f64>

Get inhibition strength matrix

pub fn num_clusters(&self) -> usize

Get number of clusters

pub fn reset(&mut self)

Reset the clusterer to initial state

Trait Implementations

impl Clone for CompetitiveNeuralClusterer

fn clone(&self) -> CompetitiveNeuralClusterer

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for CompetitiveNeuralClusterer

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

Formats the value using the given formatter.