Semi-supervised clustering via multi-level random walk
作者：
Highlights：
• We propose a new semi-supervised clustering algorithm given pairwise constraints.
• It can expand the influence of the pairwise constraints locally and proportionally.
• It utilizes the pairwise constraints in an “edge→vertex→edge” manner.
• It introduces an intermediate structure to uncover the underlying sub-structures.
• Given a k-NN sparse similarity matrix, its time complexity is approximately linear.
摘要
Highlights•We propose a new semi-supervised clustering algorithm given pairwise constraints.•It can expand the influence of the pairwise constraints locally and proportionally.•It utilizes the pairwise constraints in an “edge→vertex→edge” manner.•It introduces an intermediate structure to uncover the underlying sub-structures.•Given a k-NN sparse similarity matrix, its time complexity is approximately linear.
论文关键词：Semi-supervised clustering,Pairwise constraint,Influence expansion,Multi-level random walk,Spectral clustering
论文评审过程：Received 12 February 2013, Revised 17 June 2013, Accepted 31 July 2013, Available online 12 August 2013.
论文官网地址：https://doi.org/10.1016/j.patcog.2013.07.023
