Text Classification on Medical Transcriptions (test)
35.3PrecisionDP-based DA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DP-based DATraining samples=6000, epsilon=42024.02 | 35.3 | 37.7 | 34 | 37.6 | |
| DP-based DATraining samples=3750, epsilon=42024.02 | 34 | 37.3 | 33.7 | 37.2 | |
| DA w/ PublicTraining samples=6000, DP Protection=None2024.02 | 31.3 | 34.7 | 28.7 | 34.8 | |
| DA w/ GhostTraining samples=3750, epsilon=42024.02 | 30.3 | 34.7 | 29.7 | 34.4 | |
| DA w/ DP-SGDTraining samples=3750, epsilon=42024.02 | 29 | 35 | 27.7 | 35.2 | |
| DA w/ DP-SGDTraining samples=6000, epsilon=42024.02 | 28.3 | 35 | 28 | 35.2 | |
| DA w/ GhostTraining samples=6000, epsilon=42024.02 | 28.3 | 35 | 29.7 | 35 | |
| DA w/ PublicTraining samples=3750, DP Protection=None2024.02 | 28 | 34.7 | 28 | 34.8 | |
| PrivateTraining samples=3750, DP Protection=None2024.02 | 22.4 | 36.7 | 24 | 36.7 | |
| GhostTraining samples=3750, epsilon=42024.02 | 14.3 | 28.9 | 16.6 | 29.1 | |
| DP-SGDTraining samples=3750, epsilon=42024.02 | 13 | 31 | 17 | 31.4 |