Evolutionary Clustering via Message Passing

We are often interested in clustering objects that evolve over time and identifying solutions to the problem for every step. Evolutionary provides insight into cluster evolution temporal changes memberships while enabling performance superior achieved by independently data collected at different points. In this article we introduce evolutionary affinity propagation (EAP), an algorithm groups points exchanging messages on a factor graph. EAP promotes smoothness of solution time-evolving linking nodes graph associated with adjacent snapshots, introduces consensus enable tracking identification births deaths. Unlike existing methods require additional processing approximate number clusters or match them across time, determines tracks automatically. A comparison simulated experimental demonstrates effectiveness proposed algorithm.