DeepDefense: A Steganalysis-Based Backdoor Detecting and Mitigating Protocol in Deep Neural Networks for AI Security

<div>The effect performance on different detection rate <span class="nowrap"><svg height="9.39034pt" id="M103" style="vertical-align:-3.42943pt" version="1.1" viewbox="-0.0498162 -5.96091 7.13289 9.39034" width="7.13289pt" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"><g transform="matrix(.013,0,0,-0.013,0,0)"><path d="M516 277C516 378 460 448 376 448C335 448 288 434 240 401C158 345 127 277 107 159C53 -161 38 -221 23 -247L28 -261C43 -260 95 -250 114 -226C122 -216 129 -199 146 -75C152 -29 157 0 164 31C182 9 206 -12 237 -12C267 -12 325 8 376 46C451 102 516 179 516 277ZM432 259C432 134 366 40 259 40C236 40 204 48 171 76L195 203C219 327 253 400 326 400C398 400 432 336 432 259Z"></path></g></svg>.</span></div>

Security and Communication Networks

fig5

Figure 5

Figure 5: DeepDefense: A Steganalysis-Based Backdoor Detecting and Mitigating Protocol in Deep Neural Networks for AI Security 