Q-Learning based Metaheuristic Optimization Algorithms: A short review and perspectives

Abstract In recent years, reinforcement learning (RL) has garnered a great deal of interest from researchers because of its success in handling some complicated issues. Specifically, Q-learning as a model of RL is used a lot in various fields, and it has given an attractive result in games. In recent years, some researchers have tried to exploit the power of Q-learning to improve the results of optimization algorithms by guiding the optimization algorithm search agents based on the data saved in Q-table during the search process. The best search agent is chosen based on its accumulated performance, in other words, how well it has done overall, not how well it has done at each iteration. It is important to note that this review does not focus on reinforcement learning algorithms collaborating with metaheuristic optimization algorithms because there are so many reinforcement learning algorithms and to narrow the scope of the review, this paper will only discuss Q-learning used to enhance metaheuristic optimization algorithms. In this study will look at the huge progress made in the research community by looking at 32 different algorithms proposed on the subject from 2009 to 2022, with a focus on studies published in the last five years. As a result of the surveys conducted in this study, researchers (novices and experts) in the field of metaheuristic optimization algorithms research are expected to gain a better understanding of current research trends involving the use of Q-Learning and new motivations for outlining appropriate strategic plans for future development work as a result of the surveys conducted in this study.