Research map / Bandits and online learning / RL theory and MDPs
Tightening Exploration in Upper Confidence Reinforcement Learning
Hippolyte Bourel, Odalric Maillard, Mohammad Sadegh Talebi
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 UCRLnew instead of UCRL | Average-reward RL · 6-, 12-, 20-, 40-, 70-, and 100-state RiverSwim | normalized regret bound: 124.9 -> 30.0; 589.3 -> 59.5; 1736.3 -> 94.9; 7243.9 -> 176.9; 22576 -> 293.6; 46394 (UCRL value; UCRLnew 100-state valu | Improved |
| Use UCRLnew instead of SCAL+ | Average-reward RL · 6-, 12-, 20-, 40-, and 70-state RiverSwim | normalized regret bound: 69.1 -> 30.0; 235.5 -> 59.5; 542.2 -> 94.9; 1609.6 -> 176.9; 3802.4 -> 293.6 | Improved |
| Use UCRLnew instead of UCRLB | Average-reward RL · 6-, 12-, 20-, 40-, and 70-state RiverSwim | normalized regret bound: 38.6 -> 30.0; 85.8 -> 59.5; 148.5 -> 94.9; 305.3 -> 176.9; 540.0 -> 293.6 | Improved |
| Replace UCRL with UCRLnew | UCRL · 6-state RiverSwim | regret | Improved |
| Replace UCRL with UCRLnew | UCRL · 25-state RiverSwim | regret | Improved |
| Replace UCRL with UCRLnew | UCRL · 2-room and 4-room grid worlds | regret | Improved |
| Replace UCRL with UCRLnew | UCRL · 100-state randomly generated MDP | regret | Improved |
| Add tight confidence sets and EVI-NOSS planning (ablation) | UCRLnew · Numerical environments | regret | Improved |
| Use tight confidence sets instead of UCRLB-style confidence intervals (ablation) | UCRLnew · Numerical environments | regret | Improved |
| Use EVI-NOSS instead of EVI (ablation) | UCRLnew · Numerical environments | regret | Improved |
Closest papers by meaning
- Why is Posterior Sampling Better than Optimism for Reinforcement Learning? (ICML 2017 · similarity 0.91)
- Reinforcement Learning for Non-Stationary Markov Decision Processes: The Blessing of (More) Optimism (ICML 2020 · similarity 0.91)
- Optimistic Posterior Sampling for Reinforcement Learning with Few Samples and Tight Guarantees (NeurIPS 2022 · similarity 0.90)
- NeoRL: Efficient Exploration for Nonepisodic RL (NeurIPS 2024 · similarity 0.89)
- Beyond Optimism: Exploration With Partially Observable Rewards (NeurIPS 2024 · similarity 0.89)
- Efficient Exploration in Average-Reward Constrained Reinforcement Learning: Achieving Near-Optimal Regret With Posterior Sampling (ICML 2024 · similarity 0.89)
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.
