TY - GEN
T1 - ZK-GanDef
T2 - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, DSN 2019
AU - Liu, Guanxiong
AU - Khalil, Issa
AU - Khreishah, Abdallah
PY - 2019/6
Y1 - 2019/6
N2 - Neural Network classifiers have been used successfully in a wide range of applications. However, their underlying assumption of attack free environment has been defied by adversarial examples. Researchers tried to develop defenses; however, existing approaches are still far from providing effective solutions to this evolving problem. In this paper, we design a generative adversarial net (GAN) based zero knowledge adversarial training defense, dubbed ZK-GanDef, which does not consume adversarial examples during training. Therefore, ZK-GanDef is not only efficient in training but also adaptive to new adversarial examples. This advantage comes at the cost of small degradation in test accuracy compared to full knowledge approaches. Our experiments show that ZK-GanDef enhances test accuracy on adversarial examples by up-To 49.17% compared to zero knowledge approaches. More importantly, its test accuracy is close to that of the state-of-The-Art full knowledge approaches (maximum degradation of 8.46%), while taking much less training time.
AB - Neural Network classifiers have been used successfully in a wide range of applications. However, their underlying assumption of attack free environment has been defied by adversarial examples. Researchers tried to develop defenses; however, existing approaches are still far from providing effective solutions to this evolving problem. In this paper, we design a generative adversarial net (GAN) based zero knowledge adversarial training defense, dubbed ZK-GanDef, which does not consume adversarial examples during training. Therefore, ZK-GanDef is not only efficient in training but also adaptive to new adversarial examples. This advantage comes at the cost of small degradation in test accuracy compared to full knowledge approaches. Our experiments show that ZK-GanDef enhances test accuracy on adversarial examples by up-To 49.17% compared to zero knowledge approaches. More importantly, its test accuracy is close to that of the state-of-The-Art full knowledge approaches (maximum degradation of 8.46%), while taking much less training time.
KW - Adversarial Training Defense
KW - Generative Adversarial Nets
KW - full knowledge training
KW - zero knowledge training
UR - http://www.scopus.com/inward/record.url?scp=85072117412&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85072117412&partnerID=8YFLogxK
U2 - 10.1109/DSN.2019.00021
DO - 10.1109/DSN.2019.00021
M3 - Conference contribution
T3 - Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, DSN 2019
SP - 64
EP - 75
BT - Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, DSN 2019
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 24 June 2019 through 27 June 2019
ER -