Offline Reinforcement Learning with Differentiable Function Approximation is Provably Efficient.

Ming Yin, Mengdi Wang, Yu-Xiang Wang

Browse the full ICLR paper archive.