An acceleration method for pairwise constraints cross-entropy algorithm

Cross-entropy Semi-supervised Clustering Based on Pairwise Constraints (CE-sSC) can make full use of pairwise constraints by introducing cross-entropy as the pairwise constraints measure, which significantly boost the performance of clustering. However, CE-sSC suffers from the problems of slow convergence and local optimum, which impacts the running velocity and the searching space. The swarm intelligence clustering algorithms represented by Particle Swarm Optimization K-means algorithm (PSOKmeans) addresses these problems by parallelly computing the particles during the iteration. But in the multi-mode problem, some particles might deteriorate during the iteration, which impact the performance of these particles and impacts the searching space of the whole swarm. Therefore, in order to solve the above issues, we propose an acceleration method for pairwise constraints cross-entropy algorithm (SSACE-sSC). The algorithm takes Salp Swarm Algorithm(SSA) and the CE-sSC into account, uses the characteristics of salp swarm to optimize the clustering solution. Further experiments conducted in UCI data show that the proposed clustering algorithm can derive a better performance and velocity.