Best Arm Identification on Synthetic Multi-Armed Bandit (N=1,024, T=10,240) T = N log2 N (test)
0.0195Avg Simple RegretSH
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SHKnowledge of T=Yes (Fixed), r0=1.72026.02 | 0.0195 | 0 | 0.029 | |
| B3Knowledge of T=No (Anytime)2026.02 | 0.0268 | 0 | 0.0502 | |
| SHKnowledge of T=Yes (Fixed), r0=2.02026.02 | 0.0288 | 0 | 0.0507 | |
| UCB-EKnowledge of T=Yes (Fixed)2026.02 | 0.0403 | 0.0222 | 0.0588 | |
| BSHKnowledge of T=No (Anytime)2026.02 | 0.1531 | 0.0661 | 0.2475 | |
| BUCBKnowledge of T=No (Anytime)2026.02 | 0.2558 | 0.2374 | 0.3039 |