A novel rough set based dissimilarity measure and its application in multimodal optimization

摘要

Rough Set Theory (RST) is a mathematical tool for analyzing discrete data in data tables which deals with uncertainty. Dependency Degree (DD) in RST is a measure for calculating the degree of relevancy for two discrete data columns. Referring to the nature of DD, it can be used as a proximity measure in multimodal optimization. In this paper a new binary dissimilarity measure based on the concept of DD is proposed and combined with a multimodal optimization niching method called Dynamic Fitness Sharing (DFS). Experimental results on several multimodal binary benchmark functions show the effectiveness and high performance of proposed measure comparing with Hamming Distance (HD).