Solving large-scale multiobjective optimization via the probabilistic prediction model

The characteristic of large-scale multiobjective optimization problems (LSMOPs) is optimizing multiple conflicting objectives while considering thousands decision variables at the same time. An efficient algorithm for LSMOPs should have ability to search a large space and find global optimum in objective space. Maintaining diversity population one effective ways locate Pareto optimal set In this paper, we propose based on probabilistic prediction model, called LMOPPM, establish generating-filtering strategy tackle LSMOP. proposed method improves through importance sampling enhances convergence via trend model. Furthermore, due adoption individual-based evolutionary mechanism, computational costs are less relevant number variables, thus avoiding high time complexity. We compared with several state-of-the-art algorithms benchmark functions. experimental results complexity analysis demonstrate that provides significant improvements terms performance efficiency optimization.