Recent Advances in Evolutionary Multi-Objective Optimization
Professor & Koenig Endowed Chair, Electrical and Computer Engineering BEACON Center for the Study of Evolution in Action, An NSF Science & Technology Center.Research interests: Non-linear Optimization, Many and Multi-objective Optimization, Metamodeling, Constraint Handling, Engineering Design, Evolutionary Algorithms and Metaheuristics, Innovization, Neural Networks, Data-mining and Machine learning. If you want to obtain more information about him,please go to his homepage http://www.egr.msu.edu/~kdeb/.
Abstract: Many problems in practice are ideally posed as a multi-objective optimization problem, due to simultaneous inclusion of multiple conflicting criteria in defining a problem. Despite the traditional emphasis on scalarization methods of combining multiple objectives into a single aggregate one, it is a difficult to task to combine two or more criteria, such as fabrication cost and environment damage, into a single function in an ordinal manner. Started in early nineties, evolutionary multi-objective optimization (EMO) methods utilized multiple objectives as a vector and demonstrated to solve many two and three-objective problems with a two-step strategy: (i) first find a set of trade-off efficient solutions, and (ii) then involve a decision-maker to choose a single preferred solution. This philosophy has revolutionized the act of multi-objective optimization since then and resulted in the development of a computing field involving theoretical computer scientists, computational science, and engineering. The speaker has been involved in the growth of the EMO field right from the beginning. In this IEEE Distinguished Lecture, the speaker will discuss the development of EMO field and also highlight some recent advances, so the lecture is useful to both new-comers and EMO researchers alike.