Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Syndicated Bandits: A Framework for Auto Tuning Hyper-parameters in Contextual Bandit Algorithms
QIN DING, Yue Kang, Yi-Wei Liu, Thomas Chun Man Lee, Cho-Jui Hsieh, James Sharpnack
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 theoretical exploration parameter instead of best grid-searched value | LinUCB · Simulation, Changing features | cumulative regret: 312.69 -> 582.59 | Hurt |
| use theoretical exploration parameter instead of best grid-searched value | LinTS · Simulation, Fixed features | cumulative regret: 336.44 -> 576.83 | Hurt |
| use theoretical exploration parameter instead of best grid-searched value | LinTS · Simulation, Changing features | cumulative regret: 352.79 -> 488.99 | Hurt |
| use theoretical exploration parameter instead of best grid-searched value | LinUCB · Simulation, Fixed features | cumulative regret: 357.21 -> 364.99 | No clear effect |
| change feature type from fixed to changing arm features (best alpha shifts 4.0 -> 1.5) | LinUCB · Simulation | cumulative regret: 357.21 -> 312.69 | Mixed |
| tune exploration parameter online with two-layer EXP3 structure (TL) vs prior tuning methods | LinUCB/LinTS/UCB-GLM · Simulation and MovieLens 100K | cumulative regret | Improved |
| tune exploration and regularization parameters separately with Syndicated Bandits vs TL-Combined | LinUCB/LinTS/UCB-GLM · Simulation and MovieLens 100K | cumulative regret | Improved |
| tune exploration and regularization parameters with Syndicated vs Corral-Combined | LinUCB/LinTS/UCB-GLM · Simulation and MovieLens 100K | cumulative regret | Improved |
| tune multiple hyper-parameters instead of exploration parameter only | Syndicated Bandits · Simulation and MovieLens 100K | cumulative regret | Improved |
Closest papers by meaning
- A Contextual Bandit Bake-off (JMLR 2021 · similarity 0.89)
- Contextual Bandits for Unbounded Context Distributions (ICML 2025 · similarity 0.88)
- Active Context Selection Improves Simple Regret in Contextual Bandits (NeurIPS 2026 · similarity 0.87)
- Towards Domain Adaptive Neural Contextual Bandits (ICLR 2025 · similarity 0.87)
- High Probability Bound for Cross-Learning Contextual Bandits with Unknown Context Distributions (ICML 2025 · similarity 0.87)
- Leveraging Offline Data in Linear Latent Contextual Bandits (ICML 2025 · similarity 0.86)
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.
