1.제목: Multiobjective Genetic Algorithms for Network Design Problems
2.연사: Mitsuo Gen (Kwang Nam Hyun)
Graduate School of Information Production and System, Waseda University, Japan
1971.4 – 1974.3: Ph.D. degree, Dept. of Information Engineering, Graduate School of Eng.
Kogakuin University, Tokyo, Japan
1969.4 – 1971.3: M.S. degree, Dept. of Electronic Engineering, Graduate School of Eng.
Kogakuin University, Tokyo, Japan
1965.4 – 1969.3: B.S. degree, Dept. of Electronic Engineering, Kogakuin University, Tokyo, Japan
3.일시: 2004년 10월 27일(수요일) 오후 3:30
4.장소: 서울대학교 302동 317-2호
5.Abstract
Network design problems are fundamental issue in the various fields such as applied mathematics, computer science, engineering, management, and operations research. Networks provide a useful way to modeling real world problems and are extensively used in many different types of systems: communications, hydraulic, mechanical, electronic and logistics. At the same time, Genetic Algorithm (GA) has received one of great deal of attention regarding their potential as optimization techniques for network design problems and is often used to solve many real world problems, including the effective approaches on the multiobjective optimization problems.
In this talk I will review the recent network design techniques using GAs and introduce a new Multiobjective Genetic Algorithm (MOGA) approach for designing a Bicriteria Network Design (BND) Problem. The objectives are to maximize flow and minimize cost in the BND model. The proposed method adopts priority-based encoding method to represent a path in the network. Different from other encoding methods, such as path oriented encoding method, priority-based encoding method can be applied for different network design problems, i.e., Shortest Path Problem (SPP), Maximum Flow Problem (MXF), Minimum Cost Flow Problem (MCF), etc. In the proposed method, while weighted-sum approach is employed to evaluate solutions found in the search process, nondominated sorting technique is used to obtain Pareto optimal solutions. Numerical experiments shows the efficiency and effectiveness of the MOGA approach on the BND problem when comparing with several traditional methods.