Effectively Guiding Multiobjective Evolutionary Search by Discovering the Regularity Property
报告题目：Effectively Guiding Multiobjective Evolutionary Search by Discovering the Regularity Property
报告人：孙建永 教授（西安交通大学****）
主持人：周爱民 副教授
报告时间：2016年9月12日周一10:00-11:00
报告地点：中北校区理科大楼B816室
报告摘要：
The regularity property of multi-objective optimisation problem (MOP) states that the distribution of the Pareto optimal solutions (PSs) of a m-objective optimization problem (MOP) exhibits a (m-1)-dimensional manifold structure under mild conditions. In this paper, we propose an advanced multi-objective evolutionary algorithm, called AMEA, in which a clustering analysis is employed to discover the PS's manifold structure. An advanced sampling strategy is then developed to effectively generate promising offspring from the learned structure. The developed sampling strategy generates offspring by Gaussian perturbation on individual non-dominated solutions using the variance-covariance matrix within its cluster. The other new features include 1) an adaptive hybridisation of the developed sampling strategy with a DE recombination operator which aims to combine local and global information; 2) a re-using scheme which is to reduce the computational cost on the clustering; and 3) an adaptive strength Pareto based approach which is to adaptively determine the contribution of the developed sampling strategy and the DE generator for balancing exploration and exploitation. AMEA was empirically compared with four state-of-the-art MOEAs on a number of test instances with complex PS structure and complicated Pareto fronts. Experimental results suggest that AMEA outperforms the compared algorithms on these test instances in terms of commonly-used measure metrics. The effectiveness of the developed sampling strategy, the reusing scheme, the hybrid strategy, and the adaptive strategy was also empirically validated.
报告人简介：
Dr Jianyong Sun is now a professor at School of Mathematics and Statistics, Xi’an Jiaotong University. He is one of the award winners of the 12th “1000 Young Talent” programme. Before this post, he was a senior lecturer in Faculty of Engineering, University of Greenwich. His research spans both theoretical and practical aspects of artificial intelligence, mainly on machine learning, statistical modelling, meta-heuristic, evolutionary optimisation, computational biology/bioinformatics, and image processing. His current research interests include, but not limited to, machine learning (algorithms and learning theories) on big data; and evolutionary optimization for large-scale (problem dimension ≥ 1000) optimization problems. He has published more than 40 journal and conference papers on prestigious international journals such as IEEE Trans on Evolutionary Computation, IEEE Trans. on Cybernetics, IEEE Trans. On Neural Networks and Learning Systems, Proceedings of the National Academic Sciences (PNAS), IEEE/ACM Trans on Computational Biology and Bioinformatics, etc, and top-tier conferences such as International Conference on Machine Learning, and Congress on Evolutionary Computation, etc. He serves as PC members for more than 15 conferences, and regular reviewer/editor for many prestigious international journals.