Seminar Learning in Games

(IN2107, IN0014, IN2396)

Supervision: Prof. Dr. Martin Bichler, Dr. Artem Tsikiridis.

Game Theory studies the strategic interactions of multiple decision-makers. In this seminar, we will explore the subfield of learning in games, in which these decision-makers are agents that learn to adapt their strategies in order to maximize their own expected utility. As such, this topic is located at the intersection of game theory and machine learning, particularly multi-agent Learning. In recent years, there has been renewed interest as well as big breakthroughs in this field, fueled by the advent of more powerful computation and deep learning on the one hand, as well as the increasing deployment of autonomous systems (bots) in the economy. Topics covered in this seminar may range from theory (e.g. Equilibria and Game Dynamics, Complexity, Evaluation of Agent Strength), algorithms (e.g. Counterfactual Regret Minimization, AlphaZero), and applications (e.g. superhuman AI for Board Games; economical applications such as Automated Mechanism Design, or Equilibrium Computation in Auctions).

Credit: Stock picture, pexels.com

Introduction and Q&A meeting

There will be a brief Teams meeting for an introduction to the contents / structure of the seminar, as well as questions and answers, on February 4th at 2:00 pm.

Teilnehmen: https://teams.microsoft.com/meet/339753465666?p=OoFgSm1MZKJxl4VjD1

Besprechungs-ID: 339 753 465 666

Passcode: Zx9Qe33s

You are welcome to join the pre-course meeting for our Practical Course: Deep Reinforcement Learning & Agentic AI Systems, which will take place via the same Teams meeting at 2:30 pm.

Previous Knowledge Expected

Basic knowledge of Game Theory (e.g. from IN2239 Algorithmic Game Theory), machine learning (e.g. IN2028 Business Analytics & Machine Learning, or IN2064 Machine Learning). Students will also need a firm grasp of mathematical foundations like calculus (MA0902 / MA0001), and probability theory (IN0018, MA0009 / MA1109).

Objective

IN2107 (Master Seminar in the Master program Informatik).
IN0014 (Seminar in the Bachelor programs Informatik, Wirtschaftsinformatik).
For all other programs: Please check first whether this seminar fits in your curriculum.

Teaching and Learning Method

The emphasis in this seminar lies in the independent study of classic papers, as well as recent new results in the fields of Algorithmic Game Theory and Multi-Agent Learning. Each student will be assigned an individual topic and prepare a presentation as well as a short summary paper. We will have biweekly presentation meetings throughout the semester with two topics being presented in each meeting. Attendance of all meetings is mandatory, and interaction with the other students' work is expected. All meetings will be in 01.10.011.

Course Criteria and Registration

The seminar is primarily aimed at MSc students in Computer Science and adjacent degree programs such as Information Systems, Data Science, or Mathematics. In case there is remaining capacity available, applications from BSc students will also be considered.

All students (from all schools) must apply for the seminar via email (see below) AND via the School of CIT's matching system in the "Seminar SS26" matching instance. (Additional information here.)

Additionally, please send a current transcript of records to Artem Tsikiridis via email (artem.tsikiridis at tum.de) by 17/02/26. If you want to let us know anything else, feel free to add it in the email (e.g., your motivation to participate in this seminar, any experience relevant to the seminar outside of university courses, relevant classes in WS25/26 that aren't listed on your ToR.)

Grading

The grade will be calculated as follows:

60% your presentation
25% your handout / abstract paper
15% your score on the quizzes