Đang chuẩn bị liên kết để tải về tài liệu:
Multi-objective evolutionary algorithms: Foundation, development and open issues

Không đóng trình duyệt đến khi xuất hiện nút TẢI XUỐNG Tải xuống

This paper focuses on a special class problem called multi-objective optimization problems and evolutionary algorithms designed for it. We will overview the development of multi-objective evolutionary algorithms (MOEAs) over the years and problem difficulties and then indicate the open problems in this area. | Journal of Computer Science and Cybernetics, V.33, N.3 (2017), 193–212 DOI 10.15625/1813-9663/33/3/11111 REVIEW MULTI-OBJECTIVE EVOLUTIONARY ALGORITHMS: FOUNDATION, DEVELOPMENT AND OPEN ISSUES LAM THU BUI Le Quy Don Technical University; lambt@lqdtu.edu.vn Abstract. Evolutionary computation (EC) has been a fascinating branch of computation inspired by a natural phenomenal of evolution. EC enables computer scientists to design effective algorithms dealing with difficult problems. This paper focuses on a special class problem called multi-objective optimization problems and evolutionary algorithms designed for it. We will overview the development of multi-objective evolutionary algorithms (MOEAs) over the years and problem difficulties and then indicate the open problems in this area. Our chief goal is to provide readers with reference material in the area of multi-objective evolutionary algorithms. Keywords. Evolutionary computation, multi-objective evolutionary algorithms, optimization. 1. INTRODUCTION Evolutionary algorithms (EAs) [3, 28, 40] have emerged as major heuristic search paradigms. With the usage of a population for the search in each iteration, EAs are naturally suitable for solving multi-objective problems, which often have multiple conflicting objectives. They have attracted significant attention from the research community over the last 30 years. We can observe this fact by the number of publications produced over time [13, 16, 17]. Obviously, in these problems, there is no single solution that is the best when measured on all objectives (note that the terms solution, individual and point are used interchangeably in this paper). Instead we usually find several trade-off solutions (called the Pareto optimal set (POS) to honor Vilfredo Pareto [44], or Pareto optimal front (POF) for the image of the vectors corresponding to these solutions). In that sense, the search for an optimal solution has fundamentally changed from what we see in the case