AUC maximization on diabetes (test)
81.3AUCLocalized SGD for Pairwise Learning
Evaluation results
| Method | Links | |
|---|---|---|
| Localized SGD for Pairwise Learningepsilon=0, training size=256, mode=Non-Private2021.11 | 81.3 | |
| DPEGDepsilon=1.5, training size=256, delta=1/n2021.11 | 79.7 | |
| DPEGDepsilon=1, training size=256, delta=1/n2021.11 | 79.6 | |
| DPEGDepsilon=2, training size=256, delta=1/n2021.11 | 79.2 | |
| Differentially Private Localized SGD for Pairwise Learningepsilon=2, training size=256, delta=1/n2021.11 | 78.9 | |
| Differentially Private Localized SGD for Pairwise Learningepsilon=1.5, training size=256, delta=1/n2021.11 | 78.4 | |
| Differentially Private Localized SGD for Pairwise Learningepsilon=1, training size=256, delta=1/n2021.11 | 78.3 | |
| Differentially Private Localized SGD for Pairwise Learningepsilon=0.8, training size=256, delta=1/n2021.11 | 77.1 | |
| DPEGDepsilon=0.8, training size=256, delta=1/n2021.11 | 76.8 | |
| Differentially Private Localized SGD for Pairwise Learningepsilon=0.5, training size=256, delta=1/n2021.11 | 75.1 | |
| DPEGDepsilon=0.5, training size=256, delta=1/n2021.11 | 72.7 | |
| Differentially Private Localized SGD for Pairwise Learningepsilon=0.2, training size=256, delta=1/n2021.11 | 69 | |
| DPEGDepsilon=0.2, training size=256, delta=1/n2021.11 | 62.4 |