From: Feature partitioning for robust tree ensembles and their certification in adversarial scenarios
Dataset | \(|{\mathcal {F}}|\) | #top \({\mathcal {F}}\) | maj. class | \(|{\mathcal {D}}|\) |
---|---|---|---|---|
Wine | 13 | 7 | 73.0% | 178 |
Breast_Cancer | 30 | 15 | 62.7% | 569 |
Spam_Base | 57 | 26 | 60.6% | 4600 |
MNIST 0/1 | 784 | 54 | 53.3% | 14780 |
MNIST 5/6 | 784 | 173 | 52.1% | 13189 |
MNIST 1/7 | 784 | 79 | 51.9% | 15170 |