DEMO seminar on "Decisions, References and Achievements in MultiObjective Optimization"

Prof. Francisco Ruiz (University of Málaga, Spain) will visit the Multiobjective Optimization Group in early April. He will give a DEMO seminar talk on the 9th of April (Wednesday) from 10:15 am to 12 noon. You can join his talk in-person in Agora room Ag C421.1. You can also join online via Zoom: https://jyufi.zoom.us/j/69643413385

Event information

Event date
-
Event type
Public lectures, seminars and round tables
Event language
English
Event organizer
Faculty of Information Technology
Event payment
Free of charge
Event location category
Mattilanniemi

Title: Decisions, References and Achievements

Abstract. The concepts of references and measures are inherent to decision-making processes. This is so, first of all, because decision makers need to measure the impact of their decisions on the objectives to be taken into account. Secondly, because establishing reference or aspiration levels is a natural and intuitive way of providing preferential information. Furthermore, these two concepts are strongly linked to each other. Indeed, human beings do not understand measurement if it is not as a comparison with a certain reference unit. On the other hand, once the goals to be achieved have been established, the decision is made based on measuring the proximity of the possible solutions to these goals. For these reasons, achievement scalarizing functions play a vital role in a large part of the methods for solving multi-objective optimization problems. In this talk, a tour of the various existing achievement scalarizing functions is carried out and some of the author's contributions in this field are related.

In the field of Goal Programming, traditional achievement functions are reviewed, and the concept of Meta-Goal Programming is presented. Regarding the reference point methods, an additive achievement function is introduced, the effect of the different weight schemes (including the possibility of providing preferential weights) on the solutions obtained is studied, and the different effect that these weights may have, depending on whether the reference point is achievable or unachievable, is presented. Finally, achievement functions for problems with multiple reference points are introduced.

Keywords. Multiobjective Optimization; Goal Programming; Reference Points; Achievement Scalarizing Functions.

Add to calendar