Simultaneously Learning Stochastic and Adversarial Episodic MDPs with Known Transition

Nov-15-2025, 06:06:35 GMT–Neural Information Processing Systems 

We study the problem of learning episodic Markov Decision Processes (MDPs).