Complete Defense Framework to Protect Deep Neural Networks against Adversarial Examples

<table class="table-group" id="tab11"><tr><td><table class="table"><tr><td class="thead-hr" colspan="5"><hr/></td></tr><tr class="thead"><td class="align_left">Attack</td><td class="align_center" colspan="2">FGSM</td><td class="align_center" colspan="2">BIM</td></tr><tr><td class="thead-hr" colspan="5"><hr/></td></tr><tr><td class="align_left">Targeted network</td><td class="align_center">Inception-v3</td><td class="align_center">Complete defense</td><td class="align_center">Inception-v3</td><td class="align_center">Complete defense</td></tr><tr><td class="align_left">Success rate</td><td class="align_center">73.9</td><td class="align_center"><b>28.9</b></td><td class="align_center">100</td><td class="align_center"><b>13.7</b></td></tr><tr class="table-tr"><td colspan="5"><hr class="tbody-hr"/></td></tr></table></td></tr></table>

<div>The success rates of FGSM and BIM attacking the Inception-v3 and complete defense framework (%).</div>

Mathematical Problems in Engineering

tab11

Table 11

Table 11: Complete Defense Framework to Protect Deep Neural Networks against Adversarial Examples 