Improving Cuckoo Search: Incorporating Changes for CEC 2017 and CEC 2020 Benchmark Problems

Cuckoo search (CS) is a highly competitive single objective optimization technique. The algorithm has been widely applied in various diverse application domains and has been found to be efficient in solving various real-life problems. In the present work, we have proposed a new enhanced version of CS algorithm and tested its performance on recently proposed CEC 2017 and CEC 2020 benchmark test problems. The proposed algorithm has been named as CSsin and it employs four major modifications, i) new techniques for global and local search are devised, ii) dual search strategy is followed to enhance exploration and exploitation properties of CS algorithm, iii) a linearly decreasing switch probability has been used to add a balance between local and global search, and iv) linearly decreasing population size is used to reduce the computational burden. Apart from these modifications, the division of iterations has been employed as a further modification. The CSsin algorithm has been tested on IEEE CEC 2017 and CEC 2020 benchmark test problems having various dimension sizes and a comparative study has been performed with respect to state-of-the-art optimization algorithms for single objective bound constraint optimization problems. The results of statistical significance test affirm the competitiveness of the proposed algorithm with respect to state-of-the-art techniques.