ConceptRobust Contextual MDPs / 鲁棒上下文MDP

Robust Contextual Markov Decision Process

The robust contextual Markov decision process is a modeling framework that adds a two-layer structure to the standard MDP. The first layer contains unobservable context latent variables representing individual characteristics, while the second seeks optimal worst-case performance within an uncertainty set over context parameters, addressing model and measurement uncertainty.

Timeline (last 90 days)

Oct 16

The framework models the patient as a hidden context and handles generalization and robustness in a unified way through Robust Contextual MDPs.

Unverified50%

All Facts (1)

Unverified

The framework models the patient as a hidden context and handles generalization and robustness in a unified way through Robust Contextual MDPs.

50%

Source Articles

Robust Reinforcement Learning Solves Uncertainty Challenges in Dynamic Treatment RegimesOct 16