Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Contextual Bandits with Large Action Spaces: Made Practical
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 SpannerIGW instead of SquareCB | oneshotwiki-14031 with bilinear regression · oneshotwiki-14031 | average progressive reward (scaled by 1000): [ 7.57, 8.59 ] -> [ 8.84, 9.68 ] | Helped |
| Use SpannerIGW instead of SquareCB | oneshotwiki-14031 with deep regression · oneshotwiki-14031 | average progressive reward (scaled by 1000): [ 10.4, 11.3 ] -> [ 11.2, 12.2 ] | Mixed |
| Use deep rather than bilinear regression models | all methods on oneshotwiki-14031 · oneshotwiki-14031 | average progressive reward (scaled by 1000): ε-Greedy: [ 5.00, 6.27 ] -> [ 7.15, 8.52 ]; SpannerGreedy: [ 6.29, 7.08 ] -> [ 6.67, 8.30 ]; SquareCB: [ 7.57, 8.59 ] -> | Helped |
| Duplicate an action | SpannerIGW · oneshotwiki-311 | average progressive reward (scaled by 100): [ 12.6, 13.0 ] across 0, 16, 256, and 1024 duplicates | No clear effect |
| Duplicate an action | SquareCB · oneshotwiki-311 | average progressive reward (scaled by 100): [ 12.2, 12.6 ] -> [ 8.3, 8.6 ] | Hurt |
| Use SpannerGreedy with k=5 and 3 exploration slots instead of k=1 | amazon-3m · amazon-3m | final reward: [0.1041, 0.1046] -> [0.438, 0.440] | Helped |
| Use embedding-based SpannerGreedy rather than the tree-based method | amazon-3m with k=5, r=3 · amazon-3m | reward: roughly 0.19; [0.438, 0.440] | Helped |
| Use embedding-based SpannerGreedy rather than the tree-based method | amazon-3m with k=5, r=3 · amazon-3m | inference time per example: 7.85ms -> 160ms | Hurt |
Closest papers by meaning
- Contextual Bandits with Smooth Regret: Efficient Learning in Continuous Action Spaces (ICML 2022 · similarity 0.95)
- Efficient Contextual Bandits with Continuous Actions (NeurIPS 2020 · similarity 0.92)
- Model Selection for Contextual Bandits (NeurIPS 2019 · similarity 0.92)
- Practical and Optimal Algorithm for Linear Contextual Bandits with Rare Parameter Updates (ICML 2026 · similarity 0.91)
- Active Context Selection Improves Simple Regret in Contextual Bandits (NeurIPS 2026 · similarity 0.91)
- Diffusion Models Meet Contextual Bandits (NeurIPS 2025 · similarity 0.90)
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.