TIEJ6802 COM3: Multicriterial Design Optimization in the Age of Data Science - Fundamentals and Case Studies (JSS30) (4 cr)

Study level:
Postgraduate studies
Grading scale:
Pass - fail
Responsible organisation:
Faculty of Information Technology
Curriculum periods:
2020-2021, 2021-2022


Multicriteria Design optimization can be viewed as the challenge to find an optimal plan or specification for the construction of an object or system concerning multiple design criteria. Performance criteria that have to be met in contemporary design endeavors are various, for instance, economic efficiency, robust functioning on a high-quality level, domain-specific quality criteria, environmental sustainability, safety, health, aesthetics, and elegance. Computer-aided design has provided tools for designers to simulate, visualize, and digitally pre-evaluate a large number of virtual prototypes. To find optimal solutions is still a big challenge, especially when it comes to combining multiple design objectives.

In this class, we will discuss modern techniques in multicriteria design optimization. In particular, we will focus on newly available techniques from data science that can be useful in optimal product design.

The course will cover the topics:

  • Problem Formulations: Formalizing Multicriteria Optimization Tasks in the Language of Mathematical Programming
  • Optimization Methods: Fundamental Principles and Limits of Exact Methods and Heuristics
  • Machine Learning: Data-driven Modeling of Expensive Evaluations & Bayesian Optimization
  • Human in the Loop: Preference Elicitation, Visualization, and Navigation in Multicriteria Design Space

Throughout the course, we will work with modern optimization software (DESDEO framework, Google OR tools) and hands-on case studies from consumer product design, organizational/logistics design, and building architecture design. Here, course participants can choose from a number of case studies to get hands-on experience. The focus will be on state-of-the-art methods and we will utilize public domain/open source software packages.

Learning outcomes

Skill how to use Optimization problem solvers. Understand basic principles of optimization solvers, understand how to use machine learning in optimization contexts, know how to model user preferences and design user interaction for exploring data from optimization, practical experience with modern data-driven optimization software.

Description of prerequisites

Basic knowledge in Python/Numpy, Calculus or Analysis + Linear Algebra Bachelor Level, Basic Concepts of Statistics (correlation, mean, variance, random variables).

Completion methods

Method 1

Select all marked parts
Parts of the completion methods

Participation in teaching (4 cr)

Participation in teaching
Grading scale:
Pass - Pass with distinction
Study methods:
Attendance in lectures; Active participation in Case Study group work and presentation of results on last day (lab sessions), Exam (based on a set of review questions)
No published teaching