报告题目：基于学习的多目标优化——以自组织演化多目标优化算法为例
报告日期及时间：2017年10月18号上午10：00
报告地点： E202
报告人：周爱民
报告人国籍： 中国
报告人单位：华东师范大学
报告人简介：
周爱民，博士，研究员，博士生导师，主要研究方向为演化计算与最优化、机器学习、图像处理和应用。分别于2001年和2003年在武汉大学获得计算机学士和硕士学位，2009年在英国Essex大学获得计算机博士学位，2009年起在华东师范大学工作。在IEEE TEVC、IEEE TCYB等权威期刊和会议发表50余篇学术论文，获IES 2014最优论文。担任Swarm and Evolutionary Computation、Complex & Intelligent Systems及Swarm Intelligence and Numerical Methods等期刊副编辑或编委，参与创办演化计算与优化(ECOLE)研讨会并担任2016年会议主席。
报告摘要：
Learning guided evolutionary optimization utilizes statistical & machine learning techniques to assist the evolutionary algorithms. The learning techniques can be used to extract the problem and algorithm information online and thus to improve the algorithm performance. When using learning techniques in evolutionary algorithms, there arises a variety questions, such as why using learning techniques, which learning techniques to use, and how to use learning techniques. In this talk, we try to answer some of these questions by using a self-organizing multiobjective evolutionary algorithm (SMEA). Under mild conditions, the Pareto front (Pareto set) of a continuous m-objective optimization problem forms an (m−1)-dimensional piecewise continuous manifold. Based on this property, we proposedSMEA. At each generation, a self-organizing mapping method with (m−1) latent variables is applied to establish the neighborhood relationship among current solutions. A solution is only allowed to mate with its neighboring solutions to generate a new solution. To reduce the computational overhead, the self-organizing training step and the evolution step are conducted in an alternative manner. In other words, the self-organizing training is performed only one single step at each generation. SMEA has been applied to a number of test instances and compared with some state-of-the-art multiobjective evolutionary methods. The results have demonstrated its advantages over other approaches.
邀请人： 应时 教授，王峰 副教授
