手法を比較

	アフィニティ伝播クラスタリング	DBSCAN	スペクトラルクラスタリング
分野	機械学習	機械学習	機械学習
系統	Machine learning	Machine learning	Machine learning
提唱年	2007	1996	2002
提唱者	Brendan Frey & Delbert Dueck	Ester, M., Kriegel, H.-P., Sander, J. & Xu, X.	Ng, A. Y.; Jordan, M. I.; Weiss, Y.
種類	Exemplar-based clustering via message passing	Density-based clustering algorithm	Graph-based clustering (spectral method)
原典	Frey, B. J., & Dueck, D. (2007). Clustering by passing messages between data points. Science, 315(5814), 972–976.	Ester, M., Kriegel, H.-P., Sander, J. & Xu, X. (1996). A Density-Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise. Proceedings of the 2nd KDD, 226–231.	Ng, A. Y., Jordan, M. I., & Weiss, Y. (2002). On Spectral Clustering: Analysis and an Algorithm. Advances in Neural Information Processing Systems, 14, 849–856.
別名	affinity propagation clustering, message-passing clustering, exemplar-based clustering, yakınlık yayılımı kümeleme	DBSCAN Kümeleme, density-based clustering, density-based spatial clustering	NJW spectral clustering, graph Laplacian clustering, normalized spectral clustering, spectral graph clustering
関連	4	3	5
概要	Affinity propagation, introduced by Brendan Frey and Delbert Dueck in 2007, is a clustering algorithm that identifies representative 'exemplars' among the data by exchanging messages between every pair of points until a consistent set of clusters emerges. Unlike k-means it does not require the number of clusters to be specified in advance — that number arises from the data and a 'preference' parameter — and it works directly from pairwise similarities, which need not be a metric.	DBSCAN is a density-based clustering algorithm, introduced by Ester, Kriegel, Sander and Xu in 1996, that groups together points lying in dense regions and flags points in sparse regions as noise. It is effective on noisy data and on clusters of irregular, non-spherical shapes.	Spectral Clustering is a graph-based unsupervised learning algorithm, formalized by Ng, Jordan, and Weiss in 2002, that maps data points into a low-dimensional eigenspace derived from the similarity graph's Laplacian before applying k-means. This spectral embedding makes it possible to recover clusters of arbitrary shape — rings, crescents, interleaved spirals — that Euclidean distance-based methods consistently fail to separate.