Bora Yongacoglu

I’m an applied mathematician and machine learning researcher affiliated with Université de Montréal and Mila, the Quebec Artificial Intelligence Institute.

My research interests centre on multi-agent systems and game theory, recently with a focus on learning to cooperate in social dilemmas. These problems are interesting to me as they require agents to learn to cooperate with other players that act in good faith, but agents must also learn to resist exploitation by players who attempt to deceive and manipulate them. Naturally, understanding how agents can and should learn such policies will have important implications for designing robust, safe multi-agent AI systems.

Before moving to Mila, I was a postdoctoral fellow with the Department of Electrical and Computer Engineering at the University of Toronto. Prior to that, I received my PhD in Mathematics from Queen’s University.

Research Interests:

• learning in games
• multi-agent reinforcement learning
• mean-field games

For a complete list of papers and preprints, please check out my publications page.

Contact:

If you’re interested in collaborating or if you have any questions/comments you’d like to share, you can reach me by email at bora.yongacoglu (at) mila.quebec

News:

• (May 2025) I’ve moved to Montreal!
• (February 2025) Our paper “Unsynchronized Decentralized Q-Learning: Two Timescale Analysis By Persistence” has been accepted for publication in SIAM Journal on Control and Optimization! Stay tuned for updates!
• (October 2024) Our paper “Mean-Field Games With Finitely Many Players: Independent Learning and Subjectivity” was accepted for publication in the Journal of Machine Learning Research!
• (September 2024) Our paper on mathematical structure relevant to learning in multi-agent systems, “Paths to Equilibrium in Games”, was accepted to NeurIPS 2024. See you in Vancouver!
• (July 2024) Our paper “Generalizing Better Response Paths and Weakly Acyclic Games” was accepted for publication at this year’s Conference on Decision and Control in Milan, Italy!
• (April 2024) I delivered a talk in the Machine Learning and Mean Field Games seminar series. A recording is available on YouTube.
• (March 2024) My co-authors and I have written two new papers on game theory, and their pre-prints are now available on arXiv: link to paper 1; link to paper 2. These papers explore how the idea of “satisficing” appears in multi-agent reinforcement learning.
• (December 2023) I travelled to Singapore to present work on asynchronous decentralized Q-learning at the 2023 Conference on Decision and Control.
• (December 2023) I presented some results on $n$-player mean-field games at the Canadian Mathematics Society Winter Meeting in Montreal.
• Our paper “Independent Learning and Subjectivity in Mean-Field Games” has been awarded an Outstanding Student Paper Prize by the 2023 Networks and Communication Systems Technical Committee!
• (May 2023) I gave a talk on learning in stochastic and mean-field games at 9th Meeting on System and Control Theory in Waterloo.
• Our paper on the structure and algorithmic consequences of $\epsilon$-satisficing paths in stochastic games was accepted for publication to SIAM Journal on Mathematics of Data Science.
• (February 2023) I delivered a talk in the GERAD seminar at Polytechnique Montreal about some work on learning dynamics in games.