Membership Inference on 12 datasets mean over 32 replicates
0.786Offline AUCBaVarIA-n
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BaVarIA-nK (number of shadow models)=2542026.03 | 0.786 | 0.034 | |
| BaVarIA-nK (number of shadow models)=642026.03 | 0.785 | 0.029 | |
| BaVarIA-tK (number of shadow models)=642026.03 | 0.784 | 0.031 | |
| BaVarIA-tK (number of shadow models)=2542026.03 | 0.784 | 0.036 | |
| BaVarIA-tK (number of shadow models)=82026.03 | 0.778 | 0.015 | |
| LiRAK (number of shadow models)=2542026.03 | 0.775 | 0.045 | |
| BaVarIA-nK (number of shadow models)=82026.03 | 0.773 | 0.013 | |
| LiRAK (number of shadow models)=642026.03 | 0.772 | 0.042 | |
| RMIAK (number of shadow models)=2542026.03 | 0.77 | 0.008 | |
| RMIAK (number of shadow models)=642026.03 | 0.769 | 0.008 | |
| RMIAK (number of shadow models)=82026.03 | 0.763 | 0.009 | |
| LiRAK (number of shadow models)=82026.03 | 0.755 | 0.033 |