Research map / Bandits and online learning / RL theory and MDPs
Bounded Optimal Exploration in MDP
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 PAC-RMDP(h) with small h instead of PAC-MDP algorithms (MBIE, VBE) run with theoretically meaningful epsilon | Discrete PAC-RMDP · Five-state chain | average reward per time step | Helped |
| use PAC-RMDP(h) instead of near-Bayes optimal BEB and BOLT | Discrete PAC-RMDP · Five-state chain | average reward per time step | Helped |
| use PAC-RMDP(1) on chain with low probability (0.05) of leaving initial state (ablation) | Discrete PAC-RMDP(1) · Modified five-state chain | average reward per time step | Hurt |
| use PAC-RMDP(8) on chain with low-probability high-consequence transitions (ablation) | Discrete PAC-RMDP · Modified five-state chain | average reward per time step | Helped |
| use linear PAC-RMDP(10) instead of PAC-MDP and epsilon-greedy | Linear PAC-RMDP · Mountain car | total reward per episode | Helped |
| use epsilon-greedy exploration instead of PAC-RMDP | Linear PAC-RMDP · Mountain car | total reward per episode | Hurt |
| use PAC-RMDP instead of epsilon-greedy | Linear PAC-RMDP · Simulated HIV treatment | total reward per episode | Helped |
Closest papers by meaning
- Efficient PAC-Optimal Exploration in Concurrent, Continuous State MDPs with Delayed Updates (AAAI 2016 · similarity 0.91)
- Near Optimal Exploration-Exploitation in Non-Communicating Markov Decision Processes (NeurIPS 2018 · similarity 0.90)
- Improved Bounds for Reward-Agnostic and Reward-Free Exploration (ICML 2026 · similarity 0.90)
- Model-Based Exploration in Monitored Markov Decision Processes (ICML 2025 · similarity 0.88)
- Efficient Exploration in Average-Reward Constrained Reinforcement Learning: Achieving Near-Optimal Regret With Posterior Sampling (ICML 2024 · similarity 0.88)
- Solving Robust Markov Decision Processes: Generic, Reliable, Efficient (AAAI 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.