摘要: We propose a simple clustering framework on graphs encoding pairwise data similarities. Unlike usual similarity-based methods, the approach softly assigns to clusters in probabilistic way. More importantly, hierarchical is naturally derived this gradually merge lower-level into higher-level ones. A random walk analysis indicates that algorithm exposes structures various resolutions, i.e., higher level statistically models longer-term diffusion and thus discovers more global structure. Finally we provide very encouraging experimental results.
nips.cc参考文章(7)
Sepandar D. Kamvar, Christopher D. Manning, Dan Klein, Interpreting and Extending Classical Agglomerative Clustering Algorithms using a Model-Based approach international conference on machine learning. pp. 283- 290 ,(2002)
Jianbo Shi, J. Malik, Normalized cuts and image segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence. ,vol. 22, pp. 888- 905 ,(2000) , 10.1109/34.868688
Katherine A Heller, Zoubin Ghahramani, None, Bayesian hierarchical clustering Proceedings of the 22nd international conference on Machine learning - ICML '05. pp. 297- 304 ,(2005) , 10.1145/1102351.1102389
H. Sebastian Seung, Daniel D. Lee, Algorithms for Non-negative Matrix Factorization neural information processing systems. ,vol. 13, pp. 556- 562 ,(2000)
Dengyong Zhou, Bernhard Schölkopf, Thomas Hofmann, None, Semi-supervised Learning on Directed Graphs neural information processing systems. ,vol. 17, pp. 1633- 1640 ,(2004)
Jacob Goldberger, Sam T. Roweis, Hierarchical Clustering of a Mixture Model neural information processing systems. ,vol. 17, pp. 505- 512 ,(2004)
L Lee, D Seung, ALGORITHMS FOR NON-NEGATIVE MATRIX FACTORIZATION ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS. ,vol. 13, pp. 556- 562 ,(2001)