corelay.pipeline.spectral

SprAy-specific spectral clustering pipelines.

Classes

SpectralClustering	Clustering on a spectral embedding
SpectralEmbedding	Spectral Embedding with custom pipeline

class corelay.pipeline.spectral.SpectralClustering(*args, **kwargs)

Bases: SpectralEmbedding

Clustering on a spectral embedding

Parameters:

• clustering_fn (callable, optional) – label-returning clustering function of signature: (embedding : numpy.ndarray,) -> numpy.ndarray

• **kwargs – Keyword arguments for SpectralEmbedding

Returns:

• numpy.ndarray – Eigenvalues for spectral embedding

• numpy.ndarray – Spectral embedding (eigenvectors)

• numpy.ndarray – Labels of clustering on spectral embedding

class corelay.pipeline.spectral.SpectralEmbedding(*args, **kwargs)

Bases: Pipeline

Spectral Embedding with custom pipeline

Parameters:

• preprocessing (callable, optional) – data pre-processing function of signature: (data : numpy.ndarray,) -> numpy.ndarray

• pairwise_distance (callable, optional) – pairwise distance function of signature: (data : numpy.ndarray,) -> numpy.ndarray

• affinity (callable, optional) – affinity function of signature: (distance : numpy.ndarray,) -> numpy.ndarray

• laplacian (callable, optional) – laplacian function of signature: (distance : numpy.ndarray,) -> numpy.ndarray

Notes

Pre-computed distance matrices can be supplied by passing pairwise_distance=(lambda x: x). Pre-computed affinity matrices can be supplied by additionally passing affinity=(lambda x: x). Pre-computed graph laplacian matrices can be supplied by further passing laplacian=(lambda x: x).