Rapid Biogeography-based Silhouette Optimization for Data Clustering
Sung-Soo Kim, Bum-Su Kang · 한국경영과학회지 · 2020
Heuristic algorithm can be used for the NP-complete data clustering problem. This research aims to develop the rapid biogeography-based silhouette optimization (RBBSO) to search for the optimal data clustering solution. Valid index silhouette with sufficient clusters for the unsupervised data with considerable computation time reduction is used for this exploration. Suitable migration and mutation are also proposed to balance the converged and diversified search for data clustering by using valid index silhouette to adapt new good solutions while keeping good ones. Specifically, the computational time of RBBSO is reduced by migration based on the centroid of habitat (solution) and mutation, considering the silhouette value of each data. Experimental simulation results show that the performance of the proposed RBBSO is competitive in data clustering which uses benchmark problems.