Differentially Private Tabular Synthetic Data Generation on SCM simulation Tree prior (test)
81.41AccuracyUB
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| UBPrior=Tree, epsilon=1.02026.06 | 81.41 | 90.17 | 12.6 | 16.4 | 20.2 | 98.05 | 97.74 | |
| Tab-PEPrior=Tree, epsilon=1.02026.06 | 68.78 | 75.24 | 13.2 | 17 | 20.8 | 98.63 | 79.61 | |
| PrivMRFPrior=Tree, epsilon=1.02026.06 | 65.68 | 71.26 | 13.9 | 17.8 | 21.6 | 92.11 | 98.34 | |
| AIMPrior=Tree, epsilon=1.02026.06 | 65.4 | 71.19 | 12.8 | 16.7 | 20.6 | 93.47 | 98.33 | |
| RAP++Prior=Tree, epsilon=1.02026.06 | 64.77 | 69.77 | 19 | 23.4 | 27.8 | 93.21 | 23.94 | |
| PrivGSDPrior=Tree, epsilon=1.02026.06 | 61.99 | 66.22 | 14.1 | 18.1 | 22.1 | 84.79 | 91.74 | |
| GEMPrior=Tree, epsilon=1.02026.06 | 56.66 | 60.47 | 18.7 | 24.1 | 29.5 | 59.65 | 97.92 | |
| PrivSynPrior=Tree, epsilon=1.02026.06 | 51.04 | — | 18.2 | 23.5 | 28.8 | 65.61 | 98.08 |