Project Intelligent Systems

Project Intelligent Systems is the practical follow-up to the Intelligent Systems course from Period 2. You work in pairs to build AI bots that play the Austrian trick-taking card game Schnapsen, then run controlled experiments and write a proper academic research paper about your findings.

The project revolves around defining a research question (e.g. "Does an aggressive strategy outperform a passive one against advanced opponents?"), implementing your bot strategies in Python using the Schnapsen game engine, running thousands of simulated games, analysing the results statistically, and writing everything up in a paper following the Springer Lecture Notes in Computer Science (LNCS) format (~12 pages).

About Schnapsen: It is a two-player trick-taking card game with 20 cards (Aces, 10s, Kings, Queens, Jacks in four suits). The goal is to reach 66 points by winning tricks and declaring marriages (King-Queen pairs of the same suit). The game has two phases: Phase 1 with imperfect information (a face-down talon remains) and Phase 2 with perfect information (all remaining cards are known). This makes it an interesting testbed for AI techniques — you can apply minimax and alpha-beta pruning in Phase 2, and Monte Carlo sampling or machine learning approaches in Phase 1.

Bot techniques you might use: Rule-based strategies (aggressive, passive, bully-style), RDeep (Monte Carlo sampling of future game states), minimax/alpha-beta pruning for perfect-information play, and ML-based bots using scikit-learn (e.g. MLPClassifier or LogisticRegression trained on game replay data). The Schnapsen engine provides baseline bots like RandBot and RDeepBot to test against.

The course starts with a few introductory lectures refreshing relevant AI concepts and introducing the academic research process. After that, you work self-directed in your pair, consulting TAs during practical sessions.

What to expect from the individual test: The test takes place in the final week and is worth 30% of your grade. It focuses heavily on your own project — you will need to summarise your paper in your own words (research question, methodology, results, significance, and discussion of findings), explain how your work connects to the AI methods from the Intelligent Systems course, and name techniques you used as well as additional ones you could have applied. You will also be asked hypothetical "what would happen if the game rules changed?" questions (e.g. what if players alternate who starts each trick, or what if Phase 2 rules applied from the start). For these, you need to reason about whether your bot would still work, what you would change in your implementation and experiments, and what outcomes you would expect. Finally, there are questions about how work was divided in your group and your individual contributions. Make sure you deeply understand every part of your project — not just the code you personally wrote.