Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Balanced Linear Contextual 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 |
|---|---|---|---|
| Add inverse-propensity balancing to LinTS | LinTS · Synthetic well-specified | simulations finding optimal assignment: 84% -> 92% | Helped |
| Add inverse-propensity balancing to LinUCB | LinUCB · Synthetic well-specified | simulations finding optimal assignment: 51% -> 79% | Helped |
| Add inverse-propensity balancing to LinTS | LinTS · Synthetic misspecified | simulations finding optimal assignment: 39% -> 58% | Helped |
| Add inverse-propensity balancing to LinUCB | LinUCB · Synthetic misspecified | simulations finding optimal assignment: 29% -> 30% | No clear effect |
| Use Thompson sampling rather than deterministic UCB assignment | Balanced linear contextual bandits · Synthetic well-specified | simulations finding optimal assignment: 79% -> 92% | Helped |
| Use Thompson sampling rather than deterministic UCB assignment | Balanced linear contextual bandits · Synthetic misspecified | simulations finding optimal assignment: 30% -> 58% | Helped |
| Use balancing rather than direct reward-model estimation | Linear contextual bandits on 300 classification datasets · 300 OpenML multiclass classification datasets | normalized cumulative regret | Helped |
| Use Thompson sampling rather than UCB assignment | Linear contextual bandits on 300 classification datasets · 300 OpenML multiclass classification datasets | normalized cumulative regret | Helped |
| Use balancing with Thompson sampling rather than a policy-based method | Contextual bandits on 300 classification datasets · 300 OpenML multiclass classification datasets | normalized cumulative regret | Helped |
| Use balanced rather than direct estimation | Linear contextual bandits · Linear contextual bandits | regret bound | No clear effect |
Closest papers by meaning
- A Reduction from Linear Contextual Bandits Lower Bounds to Estimations Lower Bounds (ICML 2022 · similarity 0.91)
- Model Selection for Contextual Bandits (NeurIPS 2019 · similarity 0.91)
- Mixed-Effects Contextual Bandits (AAAI 2024 · similarity 0.91)
- Linear Contextual Bandits With Interference (ICML 2025 · similarity 0.90)
- Active Context Selection Improves Simple Regret in Contextual Bandits (NeurIPS 2026 · similarity 0.89)
- Catoni Contextual Bandits are Robust to Heavy-tailed Rewards (ICML 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.