聚类算法 Resources

Showing items tagged with "聚类算法"

DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a representative density-based clustering algorithm. Unlike partitioning and hierarchical clustering methods, it defines clusters as the largest sets of density-connected points. DBSCAN can identify clusters of arbitrary shapes in spatial databases with noise by grouping regions with sufficient density. The algorithm implementation typically involves calculating point densities using neighborhood radius (eps) and minimum points (minPts) parameters, followed by expanding clusters through density-reachable connections.

MATLAB 264 views Tagged
Download

Spectral clustering identifies arbitrarily shaped sample spaces and converges to global optimal solutions by performing eigen decomposition on similarity matrices to obtain eigenvectors for clustering. This program implements multiple clustering algorithms: Q-matrix clustering, k-means clustering, first eigencomponent clustering, second generalized eigencomponent clustering, shared data generation, and neighborhood matrix generation. Code implementation includes similarity matrix construction using Gaussian kernel functions, eigenvalue decomposition via scipy.linalg.eig, and comparative evaluation metrics.

MATLAB 254 views Tagged
Download

The AP clustering algorithm, originally published in Science magazine in 2008, represents one of the most effective clustering methods I've encountered during my decade-long research in image processing. This MATLAB implementation (compatible with MATLAB 7.0) provides comprehensive clustering demonstrations and includes practical applications for image segmentation, featuring optimized parameter configuration and efficient similarity matrix computation.

MATLAB 258 views Tagged
Download