Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Impact of Representation Learning in Linear Bandits
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 |
|---|---|---|---|
| Learn and use a shared low-dimensional representation across tasks instead of learning each task independently | Finite-action multi-task linear bandits · Finite-action multi-task linear bandits | expected total regret: \widetilde{O}(T\sqrt{d N}) -> \widetilde{O}(T \sqrt{kN} + \sqrt{dkNT}) | Helped |
| Learn and use a shared low-dimensional representation across tasks | Infinite-action multi-task linear bandits · Infinite-action multi-task linear bandits | expected total regret: \widetilde{O}(dT\sqrt{N}) -> \widetilde{O}(d^{1.5} k \sqrt{TN} + k T \sqrt N) | Helped |
| Increase the number of tasks | MLinGreedy · Synthetic finite-action simulations | regret per task | Helped |
| Increase representation dimension relative to ambient dimension | MLinGreedy · Synthetic finite-action simulations | regret per task | Mixed |
| Use MLinGreedy instead of the naive independent-task algorithm | MNIST pairwise-digit tasks · MNIST, T=10 and T=45 | regret | Helped |
Closest papers by meaning
- Fast and Sample Efficient Multi-Task Representation Learning in Stochastic Contextual Bandits (ICML 2024 · similarity 0.91)
- Beyond task diversity: provable representation transfer for sequential multitask linear bandits (NeurIPS 2024 · similarity 0.91)
- Improved Regret Bounds for Linear Bandits with Heavy-Tailed Rewards (NeurIPS 2025 · similarity 0.90)
- Linear Bandits with Partially Observable Features (ICML 2025 · similarity 0.90)
- An Improved Algorithm for Adversarial Linear Contextual Bandits via Reduction (NeurIPS 2025 · similarity 0.89)
- Geometry-Aware Approaches for Balancing Performance and Theoretical Guarantees in Linear Bandits (ICLR 2025 · 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.