Yonathan Efroni

Technion – Israel Institute of Technology

Papers

Total Citations

H-Index

About

Yonathan Efroni is a researcher specializing in reinforcement learning (RL) theory, with a particular focus on robust decision-making, exploration, and sample-efficient learning. His work addresses some of the most fundamental challenges in sequential decision-making under uncertainty, bridging rigorous theoretical foundations with practical algorithmic design. Efroni's most recognized contribution, "Action Robust Reinforcement Learning and Applications in Continuous Control" (2019, 66 citations), formalizes novel criteria for robustness to action uncertainty in RL agents — a critical concern when deploying learned policies in real-world environments where execution noise and adversarial perturbations are unavoidable. By introducing and analyzing scenarios where an agent's intended actions may be corrupted, he provides both theoretical grounding and practical algorithms for building more reliable control policies in continuous domains. His research has meaningfully advanced the community's understanding of how agents can learn effectively and safely, even under imperfect conditions. With work cited dozens of times across the RL community, Efroni has established himself as a thoughtful contributor to the theoretical underpinnings of modern reinforcement learning, making his findings highly relevant to students and researchers working on robust AI systems and real-world sequential decision-making applications.