题    目：多目标仿真优化理论与应用

演 讲 人：李浩斌  新加坡科技研究局高性能计算研究所科学家

主 持 人：镇  璐  上海大学管理学院教授

时    间：2016年1月7日（周四）上午10:00

地    点：管理学院467室

主办单位：上海大学管理学院、上海大学管理学院青年教师联谊会

演讲内容简介：

Simulation optimization has become an established methodology for technology providers and researchers. Nevertheless, industrial problems remain particularly hard to describe, and many of them are stochastic, large-scale and multi-objective in nature, thus hindering the spread of the developed methodologies.

The recent development of MO-COMPASS (multi-objective convergent optimization via most-promising-area stochastic search) answers the open question on how simulation optimization should be conducted in multi-objective setting. In addition, GO-POLARS (gradient-oriented polar random search) is developed to further improve the optimization efficiency, specifically when local information can be estimated from the simulator. Targeting at balancing the exploration vs. exploitation during the random search, GO-POLARS innovatively uses of polar-coordinate and distributions of a "random direction" in navigating the high-dimensional solution space. The revolutionary idea overturns the conventional concept of iterative search relying on Cartesian coordinates, and is expected to have much potential in future research.

The methodology of GO-POLARS with COMPASS is implemented as an open-source project, for both research and practical applications. The developers hope that the effort will bring closer the world of simulation optimization and large-scale industrial problems.

演讲人简介：

李浩斌博士，新加坡科技研究局高性能计算研究所科学家。分别于2009年、2014年获新加坡国立大学工业与系统工程系工程学士学位、博士学位。研究兴趣：运筹学、仿真优化及其在物流、医疗和海运业中的应用。

欢迎广大师生参加！