Research map / Bandits and online learning / RL theory and MDPs
Learning Near Optimal Policies with Low Inherent Bellman Error
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 |
|---|---|---|---|
| Use ELEANOR's global optimization over linear value-function parameters | Online exploration in finite-horizon MDPs with low inherent Bellman error · Finite-horizon MDPs with low inherent Bellman error | regret: rOp∑_{t=1}^H d_t√K + ∑_{t=1}^H√(d_t I K)q | Helped |
| Use ELEANOR's misspecification-adjusted confidence interval | LINUCB for contextual misspecified linear bandits · Contextual misspecified linear bandits | regret: rO(d√K + √dIK) | Helped |
| Allow positive inherent Bellman error I | ELEANOR · Finite-horizon MDPs with low inherent Bellman error | regret: ∑_{t=1}^H√(d_t I K) | Hurt |
Closest papers by meaning
- Contextual Decision Processes with low Bellman rank are PAC-Learnable (ICML 2017 · similarity 0.91)
- Computationally Efficient RL under Linear Bellman Completeness for Deterministic Dynamics (ICLR 2025 · similarity 0.90)
- Provably Efficient Reward-Agnostic Navigation with Linear Value Iteration (NeurIPS 2020 · similarity 0.90)
- Bellman-consistent Pessimism for Offline Reinforcement Learning (NeurIPS 2021 · similarity 0.89)
- Exploiting the Replay Memory Before Exploring the Environment: Enhancing Reinforcement Learning Through Empirical MDP Iteration (NeurIPS 2024 · similarity 0.88)
- Bellman Unbiasedness: Toward Provably Efficient Distributional Reinforcement Learning with General Value Function Approximation (ICML 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.