Research map / Learning theory and algorithms / Prediction and decision losses
Beyond Confidence Regions: Tight Bayesian Ambiguity Sets for Robust MDPs
Marek Petrik, Reazul Hasan Russel
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use RSVF instead of BCI to construct value-adapted ambiguity sets | Bayesian robust MDP safe return estimation · Single-state Bellman update, uniform Dirichlet prior | expected regret | Improved |
| Use RSVF instead of Hoeffding ambiguity sets | robust MDP safe return estimation · Single-state Bellman update, uniform Dirichlet prior | expected regret | Improved |
| Use BCI instead of Hoeffding ambiguity sets | robust MDP safe return estimation · Single-state Bellman update, informative Gaussian prior | expected regret | Improved |
| Use RSVF instead of BCI to construct value-adapted ambiguity sets | Bayesian robust MDP safe return estimation · Single-state Bellman update, informative Gaussian prior | expected regret | Improved |
| Use BCI instead of Hoeffding ambiguity sets | robust MDP safe return estimation · RiverSwim, uninformative prior | expected regret | Improved |
| Use RSVF instead of BCI to construct value-adapted ambiguity sets | Bayesian robust MDP safe return estimation · RiverSwim, uninformative prior | expected regret | Improved |
| Use BCI instead of Hoeffding ambiguity sets | robust MDP safe return estimation · ExpPopulation, informative prior | expected regret | Improved |
| Use RSVF instead of BCI to construct value-adapted ambiguity sets | Bayesian robust MDP safe return estimation · ExpPopulation, informative prior | expected regret | Improved |
| Use the mean transition model instead of robust ambiguity sets | safe return estimation · Single-state Bellman update | expected regret | Mixed |
| Use RSVF with value-adapted sets instead of a confidence-region method | robust MDP safe return estimation · RiverSwim, uninformative prior | safety violation rate: 0% | No clear effect |
| Use RSVF rather than BCI or Hoeffding for the single-state Bellman update | robust MDP safe return estimation · Single-state Bellman update, uniform Dirichlet prior | safety violation rate: 0% -> 5% | Mixed |
Closest papers by meaning
- Near-Optimal Distributionally Robust Reinforcement Learning with General $L_p$ Norms (NeurIPS 2024 · similarity 0.90)
- Time-Constrained Robust MDPs (NeurIPS 2024 · similarity 0.89)
- Provable Policy Gradient for Robust Average-Reward MDPs Beyond Rectangularity (ICML 2025 · similarity 0.89)
- On the Complexity of Discounted Robust MDPs with $L_p$ Uncertainty Sets (NeurIPS 2026 · similarity 0.88)
- Online MDP with Prototypes Information: A Robust Adaptive Approach (AAAI 2025 · similarity 0.88)
- Efficient Value Iteration for s-rectangular Robust Markov Decision Processes (ICML 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
