COM3: Interactive Multiobjective Optimization: Applications and Tools to Support Decision Making

About

Real-life optimization problems rarely involve only a single objective. Instead, meaningful decision-making requires optimizing multiple conflicting objectives simultaneously. To navigate these conflicts and find a satisfactory solution, preference information from a domain expert, the decision maker, is essential. Based on these preferences, we can search for solutions that best align with their goals. However, exploring such problems is often computationally and cognitively demanding, making decision support crucial.

Understanding how a decision maker expresses their preferences and how these preferences are utilized is crucial. This course introduces interactive multiobjective optimization methods, which address these key questions. Given the increasing role of data in real-world decision-making, we will also explore techniques for modeling data-driven multiobjective optimization problems. Through practical examples, we will examine and solve various real-world multiobjective problems, including both data-driven and simulator-based cases.

We will explore various interactive multiobjective optimization methods, including scalarization-based approaches and evolutionary algorithms. Additionally, we will examine how different methods can be combined and guide participants in developing their own approaches. The course will also cover graphical interfaces for interactive methods.. Hands-on experience with these concepts will be facilitated through the 2.0 version of the DESDEO framework [1], which provides the necessary tools for practical application.

Each day will focus on a central theme in interactive multiobjective optimization. The morning sessions will introduce key concepts related to the day's theme, while the afternoon sessions will provide hands-on experience applying these ideas using the DESDEO framework. To reinforce learning, an optional final project will be available, allowing participants to deepen their understanding and even contribute to the open-source DESDEO framework.

The final project is required for those seeking 4 ECTS credits. Participants who attend daily lectures and exercise sessions without the final project will receive 2 ECTS credits

Target group

Participants are expected to have prior knowledge of the following concepts:
- Python (basic programming skills)
- Fundamentals of calculus
- Optimization and mathematical programming
- Basics of single-objective optimization

Advanced Bachelor’s students, Master’s students, PhD students and post-docs.

Course aim

After completing the course, participants will have a solid understanding of multiobjective optimization and the interactive methods that support decision-making. They will be familiar with both scalarization-based and evolutionary interactive methods, as well as various visualization and GUI techniques for assisting decision makers. Additionally, students will gain the necessary prerequisites to apply the DESDEO framework in modeling and solving their own data-driven optimization problems.

Participants are encouraged to bring their own laptops and optimization problems (if any) to the course!