Ranking Preservation on NLP (Natural Language Processing) domain pool k=2 to 20
27.11AUC (MAE)D-optimality
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| D-optimalitySelection Strategy=D-optimality, Dataset Representation=Simple Features, Subset Size Range=k=2-202026.06 | 27.11 | 16.51 | 17.87 | — | 4.86 | |
| RandomSelection Strategy=Random, Subset Size Range=k=2-202026.06 | 26.5 | 16.75 | 17.8 | — | 11.51 | |
| A-optimalitySelection Strategy=A-optimality, Dataset Representation=Simple Features, Subset Size Range=k=2-202026.06 | 26.27 | 16.61 | 17.86 | — | 5.47 | |
| Cosine Farthest-FirstSelection Strategy=Cosine Farthest-First, Dataset Representation=BERT-features, Subset Size Range=k=2-202026.06 | 22.41 | 17.2 | 17.89 | — | 7.88 | |
| Euclidean Farthest-FirstSelection Strategy=Euclidean Farthest-First, Dataset Representation=BERT-features, Subset Size Range=k=2-202026.06 | 21.49 | 17.17 | 17.89 | — | 8.6 | |
| K-MeansSelection Strategy=K-Means, Dataset Representation=BERT-features, Subset Size Range=k=2-202026.06 | 21.05 | 16.93 | 17.91 | — | 12.38 |