Adjustable preference affinity propagation clustering

摘要

A new Affinity Propagation (AP) algorithm, Adjustable Preference Affinity Propagation (APAP) algorithm, is proposed in this work. The distinguishing features of APAP algorithm are that the initial value of each element preference p(k) is independently determined according to the data distribution and p(k) will be automatically adjusted during the iteration process. Experiments on synthetic data and real data are carried out. Experimental results verified the effectiveness of the proposed APAP algorithm. Compared with the standard AP algorithm, APAP algorithm has a better overall performance and can obtain better clustering results. (C) 2016 Elsevier B.V. All rights reserved.