← Search

Itai Shufaro

2 accepted papers

2025

Global Convergence of Policy Gradient in Average Reward MDPs

ICLR 2025poster

We present the first comprehensive finite-time global convergence analysis of policy gradient for infinite horizon average reward Markov decision processes (MDPs). Specifically, we focus on ergodic tabular MDPs with finite state and action spaces. Our analysis shows that the policy gradient iterates…

Cited by 0SourcePDFScholar
2025

On Bits and Bandits: Quantifying the Regret-Information Trade-off

ICLR 2025poster

In many sequential decision problems, an agent performs a repeated task. He then suffers regret and obtains information that he may use in the following rounds. However, sometimes the agent may also obtain information and avoid suffering regret by querying external sources. We study the trade-off be…