Pattern clustering is a fundamental task in machine learning. It is still challenging to cluster data that is of multiple distributions. We develop robust clustering algorithms to address such issues by means of Zeta function of a graph, maximum incremental path integral, and directed graph degree linkage.
Highlights
Agglomerative clustering via maximum incremental path integral
W. Zhang, D.L. Zhao and X.O. Tang. Pattern Recognition, 2013
PDF
Project Page
Graph Degree Linkage: Agglomerative Clustering on a Directed Graph.
W. Zhang, X.G. Wang, D.L. Zhao and X.O. Tang. In Proceedings of European Conference on Computer Vision (Poster, ECCV 2012).
PDF Project Page
Cyclizing Clusters via Zeta Function of a Graph.
D.L. Zhao and X.O. Tang. In Proceedings of Neural Information Processing Systems (Poster, NIPS 2008).
PDF Project Page