JassTheRipper - the Jass-playing AI bot

Check out the jass-bot here! All the code can be found on GitHub.

• Challenging Human Supremacy: Evaluating Monte Carlo Tree Search and Deep Learning for the Trick Taking Card Game Jass
  Niklaus, Joel and Alberti, Michele and Ingold, Rolf and Stolze, Markus and Koller, Thomas
  Accepted at: AAAI-20 Workshop on Reinforcement Learning in Games
  
  Despite the recent successful application of Artificial Intelligence (AI) to games, the performance of cooperative agents in imperfect information games is still far from surpassing humans. Cooperating with teammates whose play-styles are not previously known poses additional challenges to current state-of-the-art algorithms. In the Swiss card game Jass, coordination within the two opposing teams is crucial for winning. Since verbal communication is forbidden, the only way to transmit information within the team is through a player’s play-style. This makes the game a particularly suitable candidate subject to continue the research on AI in cooperation games with hidden information. In this work, we analyse the effectiveness and shortcomings of several state-of-the-art algorithms (Monte Carlo Tree Search ( MCTS) variants and Deep Neural Networks (DNNs)) at playing the Jass game. Our key contributions are two-fold. First, we provide a performance overview for state-of-the-art algorithms, thus, setting a strong foundation for further research on the subject. Second, we implement an  open-source a platform, where different agents (both humans and AI) can play Jass in an automated fashion, effectively reducing the overhead for other researchers who want to perform further experiments

• Comparing Learning and Search Algorithms in the Swiss Card Game Jass
  Niklaus, Joel and Alberti, Michele and Ingold, Rolf and Stolze, Markus and Koller, Thomas
  Accepted at: Applied Machine Learning Days 2020
  
  In this work, we analyse the effectiveness and shortcomings of several state-of-the-art algorithms (Monte Carlo Tree Search (MCTS) variants and Deep Neural Networks (DNNs)) at playing the Jass game. We implement and open-source a platform where different agents (both humans and AI) can play Jass in an automated fashion, effectively reducing the overhead for other researchers who want to perform further experiments
  
  
• Survey of Artificial Intelligence for Card Games and Its Application to the Swiss Game Jass
  Read the paper published at SDS 2019 about this topic here.

Reproducibility Award at SDS Conference (14.6.2019)

Link to the awarded paper: Improving Reproducible Deep Learning Workflows with DeepDIVA

Link to the conference: SDS Conference 2019

Publications, workshop  and tutorial accepted for ICDAR 2019 (Sydney)

The DIVA reserach group is well represented at this year's International Conference on Document Analysis and Recognition (ICDAR).

Publications

• A Comprehensive Study of ImageNet Pre-Training for Historical Document Image Analysis
• Labeling, Cutting, Grouping: an Efficient Text Line Segmentation Method for Medieval Manuscripts
• Offline Signature Verification using Structural Dynamic Time Warping

Tutorial

• Creating a Reproducible Research Environment

Workshop

• 2nd International Workshop on Open Services and Tools for Document Analysis (ICDAR-OST)

ICDAR 2021

Update: The location has been moved from Montrex, Switzerland to Lausanne, Switzerland.

View the proposal

Un seul pixel trompe l’intelligence artificielle

Des chercheurs ont trouvé une nouvelle manière d’attaquer les algorithmes de reconnaissance visuelle: en faussant les données utilisées pour leur apprendre à classifier les images.

publié dans "horizons magazine"

ICDAR 2021

We are happy to announce that our proposal to host ICDAR 2021 in Montreux, Switzerland has been accepted by the ICDAR committee.

View the proposal