TY - GEN
T1 - Improving DNN robustness to adversarial attacks using jacobian regularization
AU - Jakubovitz, Daniel
AU - Giryes, Raja
N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2018.
PY - 2018
Y1 - 2018
N2 - Deep neural networks have lately shown tremendous performance in various applications including vision and speech processing tasks. However, alongside their ability to perform these tasks with such high accuracy, it has been shown that they are highly susceptible to adversarial attacks: a small change in the input would cause the network to err with high confidence. This phenomenon exposes an inherent fault in these networks and their ability to generalize well. For this reason, providing robustness to adversarial attacks is an important challenge in networks training, which has led to extensive research. In this work, we suggest a theoretically inspired novel approach to improve the networks’ robustness. Our method applies regularization using the Frobenius norm of the Jacobian of the network, which is applied as post-processing, after regular training has finished. We demonstrate empirically that it leads to enhanced robustness results with a minimal change in the original network’s accuracy.
AB - Deep neural networks have lately shown tremendous performance in various applications including vision and speech processing tasks. However, alongside their ability to perform these tasks with such high accuracy, it has been shown that they are highly susceptible to adversarial attacks: a small change in the input would cause the network to err with high confidence. This phenomenon exposes an inherent fault in these networks and their ability to generalize well. For this reason, providing robustness to adversarial attacks is an important challenge in networks training, which has led to extensive research. In this work, we suggest a theoretically inspired novel approach to improve the networks’ robustness. Our method applies regularization using the Frobenius norm of the Jacobian of the network, which is applied as post-processing, after regular training has finished. We demonstrate empirically that it leads to enhanced robustness results with a minimal change in the original network’s accuracy.
KW - Adversarial examples
KW - Classification robustness
KW - Data perturbation
KW - Deep learning
KW - Jacobian regularization
KW - Neural networks
UR - http://www.scopus.com/inward/record.url?scp=85055081553&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-01258-8_32
DO - 10.1007/978-3-030-01258-8_32
M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???
AN - SCOPUS:85055081553
SN - 9783030012571
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 525
EP - 541
BT - Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings
A2 - Hebert, Martial
A2 - Ferrari, Vittorio
A2 - Sminchisescu, Cristian
A2 - Weiss, Yair
PB - Springer Verlag
T2 - 15th European Conference on Computer Vision, ECCV 2018
Y2 - 8 September 2018 through 14 September 2018
ER -