Rough Cuckoo Search: A Novel Mathematics Based Optimization Approach Based on Rough Set
Swarnajit Ray, Krishna Gopal Dhal, Prabir Kumar Naskar · Pattern Recognition and Image Analysis · 2022
Abstract Cuckoo Search Algorithm (CSA) is a robust, flexible, and easily implementable Nature-inspired Optimization Algorithm (NIOA) with fewer control parameters. Cuckoo Search has good capability on global search, but easy to suffer from local optima problem. Hence, it may possible to enhance the optimization ability of the classical Cuckoo Search algorithm. This paper presents an improved Cuckoo Search variant based on the rough set theory. Here, the population of the solutions has been considered as a rough set and partitioned into two subsets and each solution belongs to the sets depending on their fitness. Centroids of the subsets, guidance by global best solution, and Lévy distribution-based mutation are utilized to improve the solutions of the population. The experimental study has been conducted over CEC-2014 test suite and image multi-level thresholding domain with well-accepted objective functions namely Kapur’s and Tsallis entropy. Comparative study of the proposed rough Cuckoo Search (RCS) algorithm has been performed with classical Cuckoo Search, Particle Swarm Optimization, Firefly Algorithm, and Bat Algorithm. The results of such a comparative study show that proposed rough cuckoo search outperforms the other tested nature-inspired optimization algorithm in terms of optimization ability and consistency.