Struct SpectralClustering 

pub struct SpectralClustering { /* private fields */ }

Spectral Clustering using graph Laplacian and eigendecomposition.

Uses graph theory to find clusters by analyzing the spectrum (eigenvalues) of the graph Laplacian. Effective for non-convex cluster shapes.

Algorithm

1. Construct affinity matrix W (RBF or k-NN)
2. Compute graph Laplacian: L = D - W (D = degree matrix)
3. Find k smallest eigenvectors of L
4. Cluster rows of eigenvector matrix using K-Means

Examples

use aprender::prelude::*;

// Non-convex clusters (concentric circles)
let data = Matrix::from_vec(
    8,
    2,
    vec![
        0.0, 1.0, 1.0, 0.0, 0.0, -1.0, -1.0, 0.0,  // Inner circle
        0.0, 3.0, 3.0, 0.0, 0.0, -3.0, -3.0, 0.0,  // Outer circle
    ],
)
.expect("Valid matrix dimensions and data length");

let mut sc = SpectralClustering::new(2).with_gamma(0.5);
sc.fit(&data).expect("Fit succeeds with valid data");

let labels = sc.predict(&data);

Performance

• Time complexity: O(n² + n³) for eigendecomposition
• Space complexity: O(n²) for affinity matrix

Implementations

impl SpectralClustering

pub fn new(n_clusters: usize) -> Self

Create a new Spectral Clustering with default parameters.

Default: RBF affinity, gamma=1.0, n_neighbors=10

pub fn with_affinity(self, affinity: Affinity) -> Self

Set the affinity type.

pub fn with_gamma(self, gamma: f32) -> Self

Set gamma for RBF kernel (higher = more local similarity).

pub fn with_n_neighbors(self, n_neighbors: usize) -> Self

Set number of neighbors for KNN affinity.

pub fn is_fitted(&self) -> bool

Check if model has been fitted.

pub fn labels(&self) -> &Vec<usize>

Get cluster labels (panics if not fitted).

pub fn fit(&mut self, x: &Matrix<f32>) -> Result<()>

Fit the Spectral Clustering model.

Trait Implementations

impl Clone for SpectralClustering

fn clone(&self) -> SpectralClustering

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SpectralClustering

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

Formats the value using the given formatter.

impl Default for SpectralClustering

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for SpectralClustering

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for SpectralClustering

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl UnsupervisedEstimator for SpectralClustering

type Labels = Vec<usize>

The type of labels/clusters produced.

fn fit(&mut self, x: &Matrix<f32>) -> Result<()>

Fits the model to data.

fn predict(&self, _x: &Matrix<f32>) -> Self::Labels

Predicts cluster assignments or transforms data.