Research map / Bandits and online learning / RL theory and MDPs
Near-Optimal Representation Learning for Linear Bandits and Linear RL
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 |
|---|---|---|---|
| Share a learned k-dimensional representation across tasks instead of solving them independently | MTLR-OFUL · Multi-task linear bandits | regret: \tilde{O}(Md\sqrt{T}) -> \tilde{O}(M\sqrt{dkT} + d\sqrt{kMT} ) | Helped |
| Share a low-rank representation across tasks instead of running a single-task RL algorithm for each task | MTLR-LSVI · Multi-task episodic linear RL | regret: \tilde{O} (HM d\sqrt{T} + HMT\sqrt{d} \caI) -> \tilde{O} \left(H M \sqrt{dkT} + Hd\sqrt{kMT} + HMT\sqrt{d} \caI\right) | Helped |
| Allow reward-model misspecification bounded by ζ while adjusting the confidence-region parameter | MTLR-OFUL · Misspecified multi-task linear bandits | regret: An additional term \tilde{O}(MT \sqrt{d }\zeta) | Mixed |
| Use a global confidence-bonus variable to keep the value estimator linear, rather than adding the bonus directly to the least-squares estimate | MTLR-LSVI · Multi-task episodic linear RL | regret | Helped |
Closest papers by meaning
- Impact of Representation Learning in Linear Bandits (ICLR 2021 · similarity 0.92)
- Scalable Representation Learning in Linear Contextual Bandits with Constant Regret Guarantees (NeurIPS 2022 · similarity 0.92)
- Beyond task diversity: provable representation transfer for sequential multitask linear bandits (NeurIPS 2024 · similarity 0.91)
- Fast and Sample Efficient Multi-Task Representation Learning in Stochastic Contextual Bandits (ICML 2024 · similarity 0.90)
- Causal Bandits: The Pareto Optimal Frontier of Adaptivity, a Reduction to Linear Bandits, and Limitations around Unknown Marginals (ICML 2024 · similarity 0.87)
- Generalized Linear Bandits: Almost Optimal Regret with One-Pass Update (NeurIPS 2025 · similarity 0.87)
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.