Data and MATLAB Code for Reinforcement Learning Framework based Shuffled Multi Opposition-based Learning Evolutionary Algorithm

Shuffled algorithms partition a population into groups, each undergoing an independent evolutionary process. To enhance search diversity and prevent premature convergence, the shuffled multi evolutionary algorithm (SMEA) was introduced, allowing each group to evolve using a randomly selected evolutionary process. This paper presents an improved version of SMEA, called the reinforcement learning-based shuffled multi opposition-based learning evolutionary algorithm (RLSMOBEA). The proposed algorithm integrates opposition-based versions of three evolutionary processes: shuffled frog leaping (SFL), shuffled complex evolution (SCE), and shuffled differential evolution (SDE), within its local search stage. Unlike SMEA, which selects the evolutionary process randomly, RLSMOBEA incorporates reinforcement learning (RL) and SMEA to intelligently choose the most suitable evolutionary process at each iteration among the available candidates. Furthermore, the RLSMOBEA algorithm is enhanced by incorporating linear population size reduction (LPSR), leading to the development of the L-RLSMOBEA variant. This modification employs a linear function to gradually decrease both the population size and the number of groups during the optimization process. The performance of L-RLSMOBEA is rigorously assessed using the CEC2014 and CEC2017 benchmark test suites