edf壹定发手机网页版

学术报告通知(编号:2017-44)

2017-11-13

       报告题目:Evolutionary Multimodal Optimization: Decision-Making and Visualization
       报 告 人:Dr. Ran Cheng(程然) Research Fellow
       单    位:英国伯明翰大学计算智能与应用研发中心
       报告时间:2017年11月15日(周三)上午8:30
       报告地点:逸夫楼408会议室

       报告摘要:
       Multimodal optimization generally refers to single-objective optimization involving multiple optimal (or near-optimal) solutions. Thanks to the population based nature of evolutionary algorithms, which are able to obtain a set of candidate solutions in a single run, the evolutionary multimodal optimization has attracted increasing interest recently. In this talk, I will introduce our recent research in evolutionary multimodal optimization. On one hand, I will present how to perform decision-makings in multimodal optimization using evolutionary multiobjective optimization based techniques. On the other hand, I will introduce a visualization method for benchmark studies of multimodal optimization.

 

       

       报告题目:Evolutionary Many-Objective Optimisation: Pushing the Boundaries
       报 告 人:Dr. Miqing Li(李密青) Research Fellow
       单    位:英国伯明翰大学计算智能与应用研发中心
       报告时间:2017年11月15日(周三)上午9:15
       报告地点:逸夫楼408会议室

       报告摘要:
       Many-objective optimisation refers to a class of optimisation problems that have more than three objectives. The last decade has witnessed the emergence of many-objective optimisation as a booming topic in a wide range of complex modern real-world scenarios. However, in contrast to conventional multi-objective optimisation which involves two or three objectives, many-objective optimisation poses far great challenges to the area of nature-inspired search algorithms. In this talk, I will introduce several pieces of our work in solving the challenges from perspectives of algorithm design, performance assessment, test problem construction and visualisation in many-objective optimisation. In particular, I will present a simple but effective method to make Pareto-based algorithm well suited to many-objective optimisation and then a test problem suite to aid the visual examination of many-objective optimisers.

 

       

       报告题目:Multi-objective evolutionary algorithms for solving complex optimization problems
       报 告 人:张兴义 教授、博士生导师
       单    位:安徽大学计算机科学与技术学院生物智能与知识发现研究所
       报告时间:2017年11月15日(周三)上午10:00
       报告地点:逸夫楼408会议室

       报告摘要:
       Multi-objective evolutionary algorithms have been verified to be a useful technology for solving optimization problems during the last two decades, however, much work still deserves further investigations when addressing complex optimization tasks. In this talk, I will first briefly introduce the multi-objective evolutionary algorithms, and then mainly focus on threemulti-objective evolutionary algorithms recently suggested by us to tackle complex optimization problems. The three works included in this presentation are: 1) a knee point driven evolutionary algorithm for many-objective optimization problems, 2) a decision variable clustering based evolutionary algorithm for large-scale optimization problems, and 3) a multi-objective evolutionary algorithm for task-oriented pattern mining task.

 

       

       报告题目:“分而治之”的协同演化策略在大规模优化中的应用

       报 告 人:杨鸣 副教授 硕士生导师
       单    位: 中国地质大学计算机学院
       报告时间:2017年11月15日(周三)上午10:45
       报告地点: 逸夫楼408会议室

 

       报告摘要:
       大规模优化是大数据优化的重要组成部分,也是当今优化研究领域的难点和前沿。协同演化(Cooperative Co-evolution)算法将问题分解为若干个子问题,并分别对每个子问题进行优化求解。这种“分而治之”的优化策略降低了问题的求解难度。本报告主要讲解如何将大规模优化问题中的变量进行分组及如何对每个子问题进行优化求解。

 

 

edf壹定发手机版

本文章已浏览次数:551