ConceptEpisodic Confounded MDP
Episodic Confounded Markov Decision Process
An extension of the standard MDP in which state transitions and rewards are affected by unobserved confounding variables and episodes are independent; structural and completeness conditions are required to achieve causal identification.
Timeline (last 90 days)
Oct 30
The framework is further extended to episodic confounded Markov decision processes, where under structural and completeness conditions, complementary source data can identify the reward and transition mechanisms of the target domain.
Unverified50%