Strong Regret Minimization on Dueling Bandits
2Theoretical Guarantee (Order)TG-ITE-STRONG-KNOWNHORIZON
Evaluation Results
| Method | Links | |
|---|---|---|
| TG-ITE-STRONG-KNOWNHORIZONIdentify-Then-Exploit Policy=TREEASCENT then self-play, Horizon (Known/Anytime)=known H2026.06 | 2 | |
| TG-ITE-STRONG-ANYTIMEIdentify-Then-Exploit Policy=repeated TREEASCENT + self-play blocks, Horizon (Known/Anytime)=anytime2026.06 | 2 |