An Improved Spectral Clustering Algorithm Based on Random Walk
The construction of similarity matrix has an important im pact on the performance of spectral clustering algorithm. In this paper, we propose a random walk based approach to process the Gaussian ker nel similarity matrix. In this way, the pair-wise similarity between two data points is not only related to the two points, but also related to their neighbors, and the resulting matrix is closer to the ideal matrix which provides the ideal clustering result. We give a theoretical analysis of the similarity matrix and have applied this similarity matrix to spec tral clustering. Experimental results on real-world data sets show that the proposed spectral clustering algorithm improves over the traditional and other improved spectral clustering algorithms significantly.
Xianchao Zhang Quanzeng You
School of Software,Dalian University of Technology
国际会议
南宁
英文
54-66
2009-12-04（万方平台首次上网日期，不代表论文的发表时间）