Gradient Shielding: Towards Understanding Vulnerability of Deep Neural Networks
- 22 May 2020
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network Science and Engineering
- Vol. 8 (2) , 921-932
Abstract
Deep neural networks (DNNs) have been widely adopted but they are vulnerable to intentionally crafted adversarial examples. Various attack methods against DNNs have been proposed, yet there still lacks theoretical explanation of adversarial examples. In this paper, we aim to understand adversarial examples from the attacking process and we assume adding perturbations to the key/sensitive regions of the image could fool image classification DNNs. We propose gradient shielding to verify the assumption which ignores insensitive information during generating adversarial examples. Specifically, we propose interactive gradient shielding (IGS) method which selects sensitive regions and then applies gradient-based attack. To remove region selection, we propose adaptive gradient shielding (AGS) method which ignores insensitive gradients automatically. We conduct extensive experiments to evaluate the performance and the results also corroborate our perspective. With this method, we won the first place in IJCAI-AAAC 2019 Non-targeted Adversarial Attack competition.Keywords
Funding Information
- National Natural Science Foundation of China (61902082, U1636215)
- National Key Research and Development Program of China (2019YFB1706003)
- Guangdong Key R&D Program of China (2019B010136003)
This publication has 30 references indexed in Scilit:
- Towards Evaluating the Robustness of Neural Networks2017
- Practical Black-Box Attacks against Machine Learning2017
- Cloud-Based Malware Detection Game for Mobile Devices with OffloadingIEEE Transactions on Mobile Computing, 2017
- DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks2016
- Accessorize to a Crime2016
- Rethinking the Inception Architecture for Computer Vision2016
- Adversarial Diversity and Hard Positive Generation2016
- Distillation as a Defense to Adversarial Perturbations Against Deep Neural Networks2016
- The Limitations of Deep Learning in Adversarial Settings2016
- Sequential Short-Text Classification with Recurrent and Convolutional Neural Networks2016