Clustering by evidence accumulation on affinity propagation

If there are more clusters than the ideal, each intrinsic cluster will be split into several subsets. Theoretically, this split can be arbitrary and neighboring data points have a certain probability to be co-located into same cluster. Based on this observation, a method using evidence accumulation through majority voting scheme with the k-means algorithm is proposed in [3] to achieve a clustering result of an appropriate number of arbitary shaped clusters. However, the value k is not easy to choose to make it effective. Affinity propagation (AP) is a clustering algorithm which has much better performance than traditional clustering approach such as k-means algorithm. In this paper, we present an algorithm called voting partition affinity propagation (voting-PAP) which is a method for clustering using evidence accumulation based on AP. Resulting clusters by voting-PAP are not constrained to be hyper-spherically shaped. VotingPAP consists of three parts: Partition Affinity propagation (...

• Type: Conference
• Year: 2008
• Where: ICPR
• Authors: Xuqing Zhang, Fei Wu, Yueting Zhuang
• Keywords: Computer Vision, Evidence Accumulation, ICPR 2008, Majority Voting, Partition Affinity Propagation