AGNOSTIC Q-LEARNING WITH FUNCTION APPROXIMATION IN DETERMINISTIC SYSTEMS: NEAR-OPTIMAL BOUNDS ON APPROXIMATION ERROR AND SAMPLE COMPLEXITY

von Du, Simon S.
Veröffentlicht in NeurIPS (34. : 2020 : Online) 34th Conference on Neural Information Processing Systems (NeurIPS 2020) ; Volume 27 of 27 (2021)

BILINEAR CLASSES: A STRUCTURAL FRAMEWORK FOR PROVABLE GENERALIZATION IN RL

von DU, SIMON S.
Veröffentlicht in International Conference on Machine Learning (38. : 2021 : Online) International Conference on Machine Learning (ICML 2021 ; Part 4 of 16 (2022)