Mahsa Asadi

Mahsa Asadi is a Postdoctoral Researcher in the Department of Computer Science, focusing on theoretical and applied aspects of machine learning. Her research emphasizes online learning, collaborative algorithms, and reinforcement learning, with contributions to regret minimization, state-action equivalence in RL, and distributed multitask learning. She explores intersections between statistical theory and algorithmic design, with applications in autonomous systems and multi-agent coordination.

Her work spans foundational topics like concentration inequalities and upper confidence bounds, alongside practical implementations such as the CYRUS 2D simulation team and Miss PacMan AI controller. Notable contributions include model-based reinforcement learning frameworks that exploit structural equivalence properties, and collaborative mean estimation algorithms.

• Research Themes: Online Learning, Reinforcement Learning, Distributed Systems, Statistical Learning Theory
• Key Contributions: Human-in-the-loop systems, state-action abstraction techniques, regret-optimal algorithms

No formal academic awards or grants are listed in the provided information. Her advising and collaborative work focus on advancing algorithmic foundations with practical implementations in robotics and game AI contexts.