معرفی
Martin Müller is a Professor in the Department of Computing Science at the University of Alberta, where he conducts research in artificial intelligence, game theory, and heuristic search. He holds the Canada CIFAR AI Chair at Amii and is an Amii Fellow, underscoring his leadership in AI. His research group focuses on Monte Carlo tree search, reinforcement learning, combinatorial game theory, and automated planning, with applications in games such as Go, Hex, and NoGo.
His research interests span Monte Carlo and exact methods in game-tree search, exploration in heuristic search and machine learning, and algorithms in combinatorial game theory. He has developed open-source software like MCGS (Minimax-based Combinatorial Game Solver) and contributes to game-playing systems such as Fuego for Go. His work bridges theoretical foundations with practical implementations in AI-driven game solvers.
Recent publications show a strong trend in reinforcement learning, particularly in deep Q-learning, policy gradient methods, and anomaly detection in deep RL. His team also explores combinatorial game solving, sparse reward environments, and imperfect information games. The research integrates machine learning with classical AI techniques, emphasizing empirical validation and algorithmic innovation.
- Canada CIFAR AI Chair
- Amii Fellow
- Best student paper award at IEEE Conference on Games 2024
- Best paper award at IEEE COG 2021
- Outstanding paper award at AAAI-18
- Faculty of Science Dissertation Award (2016)
- Dissertation Award from the Canadian Artificial Intelligence Association (2013)
Müller has supervised numerous PhD and MSc students, including Hongming Zhang, Henry Du, and Timo Bertram, many of whose theses focus on game AI, reinforcement learning, and combinatorial optimization. He is funded by NSERC, Mitacs, and Compute Canada. His group collaborates on projects involving neural networks for game playing, SAT solving, and planning algorithms. He is currently on sabbatical but remains academically active, teaching a graduate course on combinatorial games in 2025 and hosting visiting researchers.
His lab is involved in the development of MCGS, a solver for sum games, and contributes to open-source AI software. The team publishes regularly in top venues such as NeurIPS, ICML, AAAI, and IEEE Transactions on Games. Future work includes advancing combinatorial game solvers, improving deep RL robustness, and exploring generalization in game representations.
