Tadashi Kozuno

Emergence of Exploration in Policy Gradient Reinforcement Learning via Retrying

Provably Efficient RL under Episode-Wise Safety in Constrained MDPs with Linear Function Approximation

Self Iterative Label Refinement via Robust Unlabeled Learning

Near-Optimal Policy Identification in Robust Constrained Markov Decision Processes via Epigraph Form

The Harder Path: Last Iterate Convergence for Uncoupled Learning in Zero-Sum Games with Bandit Feedback

Foundation of Scalable Constraint Learning from Human Feedback