Research map / Reinforcement learning / MDPs and planning
Regularization Guarantees Generalization in Bayesian Reinforcement Learning through Algorithmic Stability
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add L2 policy regularization to the empirical risk objective | Bayesian RL ERM policy · Bayesian RL | generalization regret bound: 2λT + 2κ/(λN) + (4κ/λ + 3C_maxT)√(ln (1/δ)/(2N)) | Helped |
| Use regularization with a finite set of possible MDPs | Bayesian RL ERM policy · Finite MDP hypothesis set | generalization regret bound: 2λT + √(C_max²T²/(2Nδ) + 48C_max³T³|A|/(2δλN minprior)) | Helped |
| Use unregularized ERM without additional assumptions on the problem | Bayesian RL ERM policy · Constructed Bayesian RL problem | regret: N = 2^T | Hurt |
| Use a structural assumption that bounds differences in history visitation across MDPs | Regularized Bayesian RL · Bayesian RL with comparable history-transition probabilities | generalization regret bound: κ=2D²C_max²T²|A|; κ/(λN) | Helped |
Closest papers by meaning
- A Theory of Regularized Markov Decision Processes (ICML 2019 · similarity 0.91)
- Doubly Regularized Markov Decision Processes for Robust Reinforcement Learning (ICML 2026 · similarity 0.90)
- Leverage the Average: an Analysis of KL Regularization in Reinforcement Learning (NeurIPS 2020 · similarity 0.90)
- Demonstration-Regularized RL (ICLR 2024 · similarity 0.88)
- Convergence Theorems for Entropy-Regularized and Distributional Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- Intrinsic Benefits of Categorical Distributional Loss: Uncertainty-aware Regularized Exploration in Reinforcement Learning (NeurIPS 2025 · 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.