Research Article
Complete Defense Framework to Protect Deep Neural Networks against Adversarial Examples
Table 6
AUC of ROC curve for minor alteration using various parameters on the training set.
| Operation | ri | Parameter | 30 | 40 | 50 | 60 | 70 | ROC-AUC | 0.9804 | 0.9801 | 0.9827 | 0.9839 | 0.9818 |
| Operation | re | Parameter | 5 | 10 | 15 | 20 | 25 | ROC-AUC | 0.9738 | 0.9775 | 0.9781 | 0.9779 | 0.9753 |
| Operation | rc | Parameter | 1 | 2 | 3 | 4 | 5 | ROC-AUC | 0.9719 | 0.9793 | 0.9782 | 0.9771 | 0.9766 |
| Operation | ra | Parameter | 1 | 2 | 3 | 4 | 5 | ROC-AUC | 0.9717 | 0.9796 | 0.9793 | 0.9789 | 0.9777 |
|
|