Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Model Selection for 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 |
|---|---|---|---|
| Adapt the active policy class using the residual-gap test and uniform exploration | ModCB · Realizable nested linear contextual bandits | regret: \tilde{O}\prn{T^{2/3}d^{1/3}_{m^\star}} | Helped |
| Tune the exploration schedule with κ=1/4 rather than κ=1/3 | ModCB · Realizable nested linear contextual bandits | regret: \tilde{O}(T^{3/4} + \sqrt{Td_{m^\star}}) | Mixed |
| Preprocess the feature maps to at most log(T) classes | ModCB · Realizable nested linear contextual bandits | regret dependence on m*: from polynomial dependence on m^{\star} to a single \log(T) factor | Helped |
| Use the residual-gap estimator instead of a plug-in estimator | EstimateResidual · Square-loss gap estimation | estimation error rate: \tilde{O}(\nicefrac{d}{n}) -> \tilde{O}(\nicefrac{\sqrt{d}}{n}) | Helped |
| Use ModCB with ILTCB rather than LinUCB on the ambient feature dimension | Synthetic contextual bandit · K=2, d=1000, d_{m*}=10 | cumulative regret | Helped |
| Use ModCB with ILTCB rather than the oracle ILTCB that knows d_{m*} | Synthetic contextual bandit · K=2, d=1000, d_{m*}=10 | cumulative regret | Helped |
Closest papers by meaning
- Universal and data-adaptive algorithms for model selection in linear contextual bandits (ICML 2022 · similarity 0.96)
- Model Selection in Contextual Stochastic Bandit Problems (NeurIPS 2020 · similarity 0.95)
- An Asymptotically Optimal Primal-Dual Incremental Algorithm for Contextual Linear Bandits (NeurIPS 2020 · similarity 0.93)
- Semiparametric Contextual Bandits (ICML 2018 · similarity 0.92)
- Active Context Selection Improves Simple Regret in Contextual Bandits (NeurIPS 2026 · similarity 0.91)
- Generalized Linear Bandits with Limited Adaptivity (NeurIPS 2024 · similarity 0.91)
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.