Esther Derman

Technion – Israel Institute of Technology

Papers

Total Citations

H-Index

About

Esther Derman is a researcher specializing in reinforcement learning and sequential decision-making, with a particular focus on the challenges that arise when standard theoretical assumptions break down in real-world environments. Her most notable work addresses the problem of action delays in reinforcement learning systems — a critical yet often overlooked issue in the field. In her influential paper "Acting in Delayed Environments with Non-Stationary Markov Policies," Derman tackles the fundamental limitation of standard Markov Decision Process formulations, which unrealistically assume that actions are executed instantaneously. This assumption, she demonstrates, can lead to catastrophic failures in high-stakes domains such as robotic manipulation, cloud computing, and financial systems. By developing frameworks that account for non-stationary Markov policies in delayed environments, her work bridges the gap between idealized theoretical models and the messy realities of practical deployment. Though her citation record is still growing — reflecting her status as an emerging voice in the field — the practical relevance of her contributions to robotics, finance, and distributed computing signals a researcher whose work is poised to have lasting influence on how reinforcement learning systems are designed and evaluated in real-world settings.