Ezgi Korkmaz

Fair Reinforcement Learning for Just AI

How to Lose Inherent Counterfactuality in Reinforcement Learning

Provable Guarantees from Practical Regularization for Alignment with Human Preferences

Stable Basis Deep Neural Policy Training

The Consequences of the Intrinsic Gap Between Reward Beliefs and MDP Rewards

Intriguing Properties of Deep Neural Policy Manifold: Intrinsic Correlation and Deep Neural Policy Curvature

AI Alignment with Provable Protection of Human Judgements

Counteractive RL: Rethinking Core Principles for Efficient and Scalable Deep Reinforcement Learning

Aligning With Human Values Without Revealing Human Judgements

HAL: Harmonic Learning in High-Dimensional MDPs

JuxtAlign: A Foundational Analysis on Alignment of Certified Reinforcement Learning

Counterintuitive RL: The Hidden Value of Acting Bad

A Large-Scale Analysis on Methodological Choices in Deep Reinforcement Learning