Leveraging symbolic models is a fundamental concept in Artificial Intelligence, enabling expressive representation of relational data, effective inference on structured data, and efficient sequential decision-making. Model-based AI methods assume that problems are specified symbolically using a logical framework that allows for compact representation. These models can be derived automatically from observations, provided explicitly by users, or developed through a combination of both. Regardless of their origin, reasoning with symbolic models is a powerful approach to solving a wide range of problems.

In this course, students will develop a deep understanding of model-based approaches for sequential decision-making through automated planning, as well as statistical relational learning for graphically representing relational information derived from observations. The course combines theoretical foundations with hands-on problem-solving, equipping participants with the skills needed to design robust, adaptable, and intelligent symbolic AI systems for complex real-world applications.

Course type:

Advanced course for PhD students

  • AI track: elective
  • AS track: elective
  • Joint curriculum: advanced

Time: Given even years, spring (first time spring 2026)

Teachers: Daniel Gnad (LiU). Jendrik Seipp (LiU), Vera Koponen (UU)

The course requires solid programming experience in a high-level language. The participants are assumed to have a basic background in propositional logic and first-order logic.

Upon completion of the course the students should be able to:

  • explain and analyze state-of-the-art planning methods.
  • implement planning algorithms for solving sequential decision-making problems.
  • model simple planning problems in a suitable symbolic representation.
  • use basic methods of inductive logic programming for learning and inference.
  • apply the concept of “probabilities on domains” to learn a parametrized graphical model from a relational structure.
  • apply the concept of “probabilities on possible worlds” with parametrized graphical model to make inferences on arbitrary domains.

This course provides an introduction to two key fields in Artificial Intelligence:

  1.  symbolic sequential decision-making and
  2. statistical relational learning.

The course is structured into independent modules, with each field introduced separately before participants explore their integration in a final group project.

Recommended:

  • Artificial Intelligence: A Modern Approach; Stuart Russell, Peter Norvig; Pearson (2020)
  • Notes on Logic, Probability and Statistical Relational Learning; Vera Koponen (distributed by the author)
  • (optional) Statistical Relational Artificial Intelligence: Logic, Probability, and Computation, Synthesis; Luc De Raedt, Kristian Kersting, Sriraam Natarajan, David Poole; Lectures on Artificial Intelligence and Machine Learning #32, Morgan & Claypool Publishers (2016)
  •  (optional) Introduction to Statistical Relational Learning; Lise Getoor, Ben Taskar (editors); The MIT Press (2007)
  • (optional) Logical and Relational Learning; Luc De Raedt; Springer-Verlag (2008)

One individual hand-in assignment per 2-day meeting and a group project that is presented in a seminar.