Converging Towards Preferred Solutions under Distance-based Value Functions

Gülşah Karakaya, Business Administration, Middle East Technical University

Abstract

One of the goals of multi-objective optimization problems is converging towards preferred solution(s) of a decision maker (DM). To reach preferred solutions, many interactive approaches where the DM is involved in the search process, assume an underlying value function that represents the preferences of the DM. Generally, these approaches obtain preference information from the DM progressively, update the search space accordingly, and continue until converging to preferred solutions. Throughout the search process, the approaches exploit the properties of the assumed value function. Linear, quasiconcave/quasiconvex, and general monotone functions are among the commonly used value functions. In this talk, we address distance-based value functions that can represent a wide variety of preference structures. We demonstrate the performances of the interactive approaches under different distance-based value functions.

This is a joint work with Murat Köksalan.

Gülşah Karakaya is an assistant professor in the Business Administration Department of Middle East Technical University (METU). She received her BS, MS, and PhD degrees from the Industrial Engineering Department of METU where she worked as a research\teaching assistant. Before joining the Business Administration Department, she worked as a postdoctoral researcher at Singapore University of Technology and Design. Her research interests are multiple criteria decision making, evolutionary algorithms, combinatorial optimization, and feature selection.

Venue

Friday, December 30, 2022, 4.00 pm - IE Building, Blue Auditorium (IE 03)