Best Arm Identification on Linear instances K=50, d=20 Appendix G (randomly generated)
14.71SamplesLinGapE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LinGapEElimination Strategy=Elim. stopping + sampling2022.05 | 14.71 | 0.17 | -23 | |
| LinGameElimination Strategy=Elim. stopping + sampling2022.05 | 14.87 | 0.19 | -24 | |
| FWSElimination Strategy=Elim. stopping + sampling2022.05 | 14.99 | 0.56 | -32 | |
| FWSElimination Strategy=Elim. stopping2022.05 | 15.24 | 0.81 | — | |
| FWSElimination Strategy=No elimination (LLR)2022.05 | 15.26 | 0.83 | — | |
| LinGapEElimination Strategy=Elim. stopping2022.05 | 17.29 | 0.19 | — | |
| LinGapEElimination Strategy=No elimination (LLR)2022.05 | 17.31 | 0.22 | — | |
| LinGameElimination Strategy=Elim. stopping2022.05 | 23.05 | 0.21 | — | |
| Lazy TaSElimination Strategy=Elim. stopping + sampling2022.05 | 23.51 | 0.24 | -25 | |
| LinGameElimination Strategy=No elimination (LLR)2022.05 | 23.77 | 0.25 | — | |
| OracleElimination Strategy=Elim. stopping2022.05 | 28.65 | 0.03 | — | |
| OracleElimination Strategy=No elimination (LLR)2022.05 | 29.1 | 0.06 | — | |
| LinGapEElimination Strategy=Elim. stopping + sampling2022.05 | 29.89 | 0.18 | -22 | |
| LinGameElimination Strategy=Elim. stopping + sampling2022.05 | 32.49 | 0.18 | -22 | |
| FWSElimination Strategy=Elim. stopping + sampling2022.05 | 32.62 | 0.45 | -38 | |
| LinGapEElimination Strategy=Elim. stopping2022.05 | 33.11 | 0.2 | — | |
| LinGapEElimination Strategy=No elimination (LLR)2022.05 | 33.19 | 0.23 | — | |
| Lazy TaSElimination Strategy=Elim. stopping2022.05 | 33.98 | 0.3 | — | |
| Lazy TaSElimination Strategy=No elimination (LLR)2022.05 | 35.11 | 0.32 | — | |
| FWSElimination Strategy=Elim. stopping2022.05 | 42.25 | 0.7 | — | |
| FWSElimination Strategy=No elimination (LLR)2022.05 | 42.26 | 0.73 | — | |
| LinGameElimination Strategy=Elim. stopping2022.05 | 43.67 | 0.21 | — | |
| LinGameElimination Strategy=No elimination (LLR)2022.05 | 45.34 | 0.23 | — | |
| OracleElimination Strategy=Elim. stopping2022.05 | 55.36 | 0.02 | — | |
| OracleElimination Strategy=No elimination (LLR)2022.05 | 56.36 | 0.05 | — | |
| Lazy TaSElimination Strategy=Elim. stopping + sampling2022.05 | 64.48 | 0.12 | -20 | |
| Lazy TaSElimination Strategy=Elim. stopping2022.05 | 74.08 | 0.13 | — | |
| Lazy TaSElimination Strategy=No elimination (LLR)2022.05 | 76.33 | 0.15 | — | |
| XY-AdaptiveElimination Strategy=natively elimination-based2022.05 | 87.08 | 0.44 | — | |
| RAGEElimination Strategy=natively elimination-based2022.05 | 106.87 | 0.02 | — |