TIEJ6810 COM1: Interactive Multiobjective Optimization (Applications and Tools to Support Decision Making) (2–4 cr)

Study level:
Postgraduate studies
Grading scale:
Pass - fail
Responsible organisation:
Faculty of Information Technology
Curriculum periods:
2021-2022, 2022-2023, 2023-2024


Real-life optimization problems are seldom single-objective. Instead, to make meaningful decisions, we must account for multiple objectives to be optimized simultaneously. Furthermore, the considered objectives are often in conflict. To resolve these conflicts and find a satisfactory solution, we need preference information from a domain expert, also known as a decision maker. Based on these preferences, we can start searching for solution(s) that best match the wishes of the decision maker. However, exploring such problems can be computationally and cognitively challenging. Thus, the decision maker requires support.
How a decision maker provides their preferences and how these preferences are used are important questions. In this course, we will introduce interactive multiobjective optimization methods, which provide answers to the two questions presented. Moreover, as many real-life problems are based on data nowadays, we will also explore how to model data-driven multiobjective optimization problems. We will give examples of (and solve) various kinds of real-life multiobjective problems, including data-driven and simulator-based problems.

We will explore various kinds of interactive multiobjective optimization methods, such as, scalarization based methods and methods based on evolutionary algorithms. We will also explore the possibility of combining different types of methods, and how to create your own method. Graphical interfaces for interactive methods, and how to implement them, will be also considered during the course. All of this will be made possible by the DESDEO framework [1], which offers the necessary tools for us to get a hands-on experience on the methods and ideas discussed in this course. Each day will revolve around a central theme in interactive multiobjective optimization. The first half of each day will start with an introduction to the central concepts related to the day's theme. During the second half of each day, we will get a hands-on experience in applying the ideas discussed by utilizing the DESDEO framework. At the end of the course, you will be tasked with an optional final project to really test your understanding of the discussed topics. You will have an opportunity to contribute to the open source software framework DESDEO.
The final project will be mandatory for those willing to get 4 ECTS. For those who will only attend the daily lectures and complete the daily assignments, 2 ECTS will be rewarded.

[1] G. Misitano, B. S. Saini, B. Afsar, B. Shavazipour and K. Miettinen, "DESDEO: The Modular and Open Source Framework for Interactive Multiobjective Optimization," IEEE Access, vol. 9, pp. 148277-148295, 2021, doi: 10.1109/ACCESS.2021.3123825.

Learning outcomes

After the course, the student will be familiar with the central concepts of multiobjective optimizaiton in general and interactive methods that support decision making in particular. The student will be familiar with both scalarization and evolutionary based interactive methods. The students will also learn how to apply various visualization and GUI techniques to support a decision maker. Moreover, the student will have the pre-requisites to start applying the DESDEO framework to model and solve their own (data-driven) problems.

Description of prerequisites

The student is expected to be familiar with the following concepts:
- Python, Basics of calculus, Optimization and mathematical programming, Basics of single-objective optimization.

Completion methods

Method 1

Select all marked parts
Parts of the completion methods

Participation in teaching (2–4 cr)

Participation in teaching
Grading scale:
Pass - fail