협력 공진화 알고리즘에서 최적 부분 문제 선택을 위한 Non-Stationary UCB-Tuned 알고리즘의 활용 방법 연구 = Study on Utilizing the Non-Stationary UCB-Tuned Algorithm to Choose Optimal Subproblems in Cooperative Coevolutionary Algorithms

다국어 초록 (Multilingual Abstract)

In Cooperative Coevolutionary (CC) algorithms used to optimize an objective function with a high-dimensional domain, the subproblem selection task serves to restrict the solution space in which the sub-optimal solution is searched, thereby significantly affecting the performance of the solution search. Therefore, it is important to identify a subproblem that can significantly contribute to finding an optimal solution by balancing exploration and exploitation. Accordingly, in this paper, we propose a subproblem selection method that utilizes the sliding window-based non-stationary UCB-Tuned algorithm to effectively maintain the exploration-exploitation trade-off in the subproblem selection task. In the practical experiments with 1,000-dimensional benchmark functions, the CC algorithm with our subproblem selection method exhibited the best performance in terms of optimal solution search.