Seminar: Graduate Seminar
Safe and Efficient Reinforcement Learning
Date:
August,16,2026
Start Time:
11:30 - 12:30
Location:
506, Zisapel Building
Add to:
Lecturer:
Navdeep Kumar
Research Areas:
| Reinforcement learning (RL) has achieved remarkable empirical success, yet deploying RL systems in real-world environments remains challenging due to uncertainty, limited data, and reliability requirements. In this seminar, I will present my research on the theoretical foundations of robust and efficient reinforcement learning. The talk will cover recent advances in robust Markov decision processes, policy-gradient methods, and actor-critic algorithms, with a focus on uncertainty modeling, computational tractability, and sample complexity. I will discuss how these results contribute to the development of reliable decision-making algorithms with provable guarantees and conclude with future directions toward robust, safe, and trustworthy reinforcement learning. |
| Navdeep Kumar is a final-year PhD candidate in the Department of Electrical and Computer Engineering at the Technion – Israel Institute of Technology. His research focuses on reinforcement learning, robust Markov decision processes, and optimization under uncertainty. He is interested in developing reliable and sample-efficient learning algorithms with theoretical guarantees for decision-making in uncertain environments.
Ph.D. student Under the supervision of Prof. Kfir Levy and Prof. Shie Mannor.
|