Research map / Bandits and online learning / Multi-armed bandits
Covariance-adaptive best arm identification
El Mehdi Saad, Gilles Blanchard, Nicolas Verzelen
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 simultaneous observations and pairwise comparisons that estimate reward-difference variance | Pairwise-BAI · Bounded rewards, fixed-confidence best-arm identification | query complexity | Improved |
| Use simultaneous observations and pairwise comparisons that estimate reward-difference variance | Pairwise-BAI · Gaussian rewards, fixed-confidence best-arm identification | query complexity | Improved |
| Continue querying arms after they are eliminated from the candidate set (ablation) | Pairwise-BAI · Bounded rewards, fixed-confidence best-arm identification | query complexity guarantee | Improved |
| Stop querying eliminated arms rather than continuing to sample them (ablation) | Pairwise-BAI · Gaussian rewards, fixed-confidence best-arm identification | query complexity guarantee | Hurt |
| Continue sampling eliminated arms until round 2t instead of round 82t (ablation) | Pairwise-BAI+ · Synthetic Gaussian experiments | average sample complexity | Hurt |
| Remove the instruction to continue querying eliminated arms (ablation) | Pairwise-BAI · Synthetic Gaussian experiments | average sample complexity | Improved |
| Compare candidate arms with convex combinations of other arms rather than only pairwise | Convex-BAI · Bounded rewards, fixed-confidence best-arm identification | query complexity | Improved |
| Increase the number of highly correlated suboptimal arms | Pairwise-BAI and Pairwise-BAI+ · Synthetic Gaussian clustered-arm experiment | average sample complexity | Improved |
Closest papers by meaning
- Balancing Performance and Costs in Best Arm Identification (NeurIPS 2025 · similarity 0.92)
- Constrained Best Arm Identification (NeurIPS 2025 · similarity 0.92)
- Optimal Multi-Fidelity Best-Arm Identification (NeurIPS 2024 · similarity 0.91)
- The Role of Contextual Information in Best Arm Identification (JMLR 2026 · similarity 0.91)
- Best Arm Identification in Generalized Linear Bandits via Hybrid Feedback (NeurIPS 2026 · similarity 0.90)
- Fixed Confidence Best Arm Identification in the Bayesian Setting (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.
