Research map / Bandits and online learning / Multi-armed bandits
Mean-based Best Arm Identification in Stochastic Bandits under Reward Contamination
Arpan Mukherjee, Ali Tajer, Pin-Yu Chen, Payel Das
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 G-CBAI (gap-based, trimmed mean) instead of median-based successive elimination and random selection | G-CBAI · Synthetic Gaussian bandit K=4, eps=0.1 | average sample complexity E[tau] | Improved |
| Use G-CBAI instead of existing baselines on real data | G-CBAI · NYCC and PKIS2, eps=0.1 | average sample complexity E[tau] | Improved |
| Replace empirical median estimator with trimmed mean in successive elimination on a Gaussian bandit | SE-CBAI · Synthetic Gaussian bandit K=4, eps=0.1 | average sample complexity E[tau] | Hurt |
| Use the theoretical confidence interval in G-CBAI instead of a tighter log(log t/delta) interval (ablation) | G-CBAI · Synthetic and real data | average sample complexity E[tau] | Hurt |
| Use the theoretically supported confidence interval for both algorithms (ablation) | SE-CBAI vs G-CBAI · Experiments | average sample complexity E[tau] | Improved |
| Increase confidence parameter delta (looser confidence) | random selection vs successive elimination · Experiments | gap in E[tau] | Mixed |
| Use trimmed mean estimator instead of other popular estimators (ablation) | CBAI algorithms · Appendix comparisons across bandit environments | average sample complexity E[tau] | Improved |
| Increase contamination level epsilon | proposed algorithms · Synthetic Gaussian bandit | average sample complexity E[tau] | Hurt |
Closest papers by meaning
- Balancing Performance and Costs in Best Arm Identification (NeurIPS 2025 · similarity 0.90)
- Best Arm Identification in Generalized Linear Bandits via Hybrid Feedback (NeurIPS 2026 · similarity 0.90)
- Best Arm Identification for Stochastic Rising Bandits (ICML 2024 · similarity 0.90)
- Optimal Multi-Fidelity Best-Arm Identification (NeurIPS 2024 · similarity 0.90)
- Almost Minimax Optimal Best Arm Identification in Piecewise Stationary Linear Bandits (NeurIPS 2024 · similarity 0.90)
- Stochastic Bandits Robust to Adversarial Attacks (ICLR 2025 · similarity 0.89)
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.
