Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Efficient First-Order Contextual Bandits: Prediction, Allocation, and Triangular Discrimination
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 the logarithmic-loss oracle instead of the square-loss oracle (ablation) | SquareCB · Contextual bandit bake-off, multiclass datasets | progressive validation loss | Helped |
| Use reweighted inverse gap weighting instead of original inverse gap weighting (ablation) | SquareCB with logistic-loss oracle · Contextual bandit bake-off, per-dataset hyperparameter tuning | progressive validation loss: 14/17 | Mixed |
| Use reweighted inverse gap weighting instead of original inverse gap weighting (ablation) | SquareCB with logistic-loss oracle · Contextual bandit bake-off, datasets with small optimal loss | progressive validation loss: 14/17; L^{\star}\leq{}0.2 | Helped |
| Use reweighted inverse gap weighting instead of original inverse gap weighting (ablation) | SquareCB with logistic-loss oracle · Contextual bandit bake-off, datasets where optimal loss is not small | progressive validation loss | Hurt |
| Replace least-squares regression with logarithmic-loss regression | Plug-in cost-sensitive classifier · Realizable cost-sensitive classification | excess risk: 16\sqrt{\frac{L^{\star} \cdot A{}\rbr{\log|\Fcal| + \log(A{}/\delta)}}{n}} + 68\frac{A{}\rbr{\log |\Fcal| + \log(A{}/\de | Helped |
| Use least-squares regression for plug-in classification (ablation) | Plug-in cost-sensitive classifier · Realizable cost-sensitive classification | excess risk: L^{\star}\leq \tfrac{2^{7}}{n}; L(\pihatls) - L^{\star} \geq{}2^{-5}\sqrt{\frac{1}{n}} | Hurt |
Closest papers by meaning
- Beyond UCB: Optimal and Efficient Contextual Bandits with Regression Oracles (ICML 2020 · similarity 0.92)
- High Probability Bound for Cross-Learning Contextual Bandits with Unknown Context Distributions (ICML 2025 · similarity 0.90)
- Efficient Contextual Bandits with Uninformed Feedback Graphs (ICML 2024 · similarity 0.90)
- Causal Bandits: The Pareto Optimal Frontier of Adaptivity, a Reduction to Linear Bandits, and Limitations around Unknown Marginals (ICML 2024 · similarity 0.90)
- Regret Bounds for Adversarial Contextual Bandits with General Function Approximation and Delayed Feedback (NeurIPS 2025 · similarity 0.90)
- Sequential Probability Assignment with Contexts: Minimax Regret, Contextual Shtarkov Sums, and Contextual Normalized Maximum Likelihood (NeurIPS 2024 · 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.