Study on Modified Affinity Propagation Clustering Based on Simulated Annealing Algorithm

Affinity Propagation Clustering Algorithm which can obtain ideal results fast and effectively is a new classification algorithm.Taking consideration of randomness of initial setting bias parameters and it is difficult to assign bias parameters to get optimal clustering results, this paper propose two methods of how to set initial bias parameters according to the local distribution of raw data and regards Silhouette as objective function to use Simulated Annealing Algorithm to automatically search for the optimal bias parameters.Experimental results on real data sets show that algorithm can obtain better classification results.