Title: Interactive Multiobjective Optimization from MCDM and EMO Perspectives

Speaker: Prof. Kaisa Miettinen
University of Jyväskylä, Finland

Summary:

Abstract: In various domains, when we must make a decision by considering several conflicting objective functions simultaneously, we need multiobjective optimization methods. Pareto optimal solutions represent different trade-offs, and we need some additional information to order them. If we can get preference information from a decision maker, a domain expert, we can integrate this information in the solution process and support the decision maker in finding the best balance among the trade-offs, that is, the most preferred solution. In interactive methods, a decision maker augments the problem formulation with domain expertise and directs an iterative solution process with preferences. In an iterative fashion, the decision maker provides preferences and obtains solutions reflecting them. In this way, the decision maker gains insight into the interdependencies and trade-offs among the conflicting objective functions and learns about the feasibility of the preferences. Based on the learning, the decision maker can update the preferences and eventually get convinced of the quality of the most preferred solution. By generating only solutions that are of interest to the decision maker, we can save in computation cost, and also decrease cognitive cost when only a limited number of solutions is to be considered at a time. I discuss the similarities and differences of some interactive methods representing multiple criteria decision making (MCDM) and evolutionary multiobjective optimization (EMO) perspectives. I pay attention to expectations related to method development when a human being is involved. I also demonstrate the advantages of some interactive methods with real examples and introduce the modular, open-source software framework DESDEO that hosts different interactive methods. The implementation of interactive methods involves many requirements. We must also pay attention to the user experience of the decision maker in graphical user interfaces. As an example of visualizations developed to support the decision maker, I show SCORE bands. All this aims at supporting a decision maker in making better decisions.

Bio:

Kaisa Miettinen is Vice-rector and Professor of Industrial Optimization at the University of Jyväskylä. Her research interests include theory, methods, applications and software of nonlinear multiobjective optimization, in particular, interactive approaches. She heads the Research Group on Multiobjective Optimization and leads the thematic research area Decision Analytics utilizing Causal Models and Multiobjective Optimization (www.jyu.fi/demo). She has authored over 230 refereed journal, proceedings and collection papers, edited 20 proceedings, collections and special issues and written a monograph Nonlinear Multiobjective Optimization. She is a member of the Finnish Academy of Science and Letters and has been the President of the International Society on Multiple Criteria Decision Making (MCDM) and Finnish Operations Research Society. She has served in the editorial boards of ten international journals and belongs to the Steering Committee of Evolutionary Multiobjective Optimization. She has worked at IIASA, International Institute for Applied Systems Analysis in Austria, KTH Royal Institute of Technology in Stockholm, Sweden and Helsinki School of Economics, Finland. She has received the Georg Cantor Award of the International Society on MCDM for developing innovative ideas in the theory and methodology and was appointed as the OR Person of the year in 2023 by the Finnish Operations Research Society. Web: http://users.jyu.fi/~miettine/engl.html

—

SEMINAR: WED 28 MAY 2025 16:00-17:00 (AEST, Melbourne Time)
ZOOM MEETING ID: 873 1557 5255; PASSWORD: 778635