Abstract
Despite their well-documented learning capabilities in clean environments, deep convolutional neural networks (CNNs) are extremely fragile in adversarial settings, where carefully crafted perturbations created by an attacker can easily disrupt the task at hand. Numerous methods have been proposed for designing effective attacks, while the design of effective defense schemes is still an open area. This work leverages randomization-based defense schemes to introduce a sampling mechanism for strong and efficient defense. To this end, sampling is proposed to take place over the matricized mid-layer data in the neural network, and the sampling probabilities are systematically obtained via variance minimization. The proposed defense only requires adding sampling blocks to the network in the inference phase without extra overhead in the training. In addition, it can be utilized on any pre-trained network without altering the weights. Numerical tests corroborate the improved defense against various attack schemes in comparison with state-of-the-art randomized defenses.
Original language | English (US) |
---|---|
Title of host publication | 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings |
Publisher | Institute of Electrical and Electronics Engineers Inc. |
Pages | 3277-3281 |
Number of pages | 5 |
ISBN (Electronic) | 9781479981311 |
DOIs | |
State | Published - May 2019 |
Event | 44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Brighton, United Kingdom Duration: May 12 2019 → May 17 2019 |
Publication series
Name | ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings |
---|---|
Volume | 2019-May |
ISSN (Print) | 1520-6149 |
Conference
Conference | 44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 |
---|---|
Country/Territory | United Kingdom |
City | Brighton |
Period | 5/12/19 → 5/17/19 |
Bibliographical note
Funding Information:Majority of this work was done during a summer research internship at Technicolor AI Lab in Palo Alto, CA - USA. This research was supported in part by NSF grant 1500713, 151405\6, 1505970, and 1711471. Author emails: sheik081@umn.edu, swayambhoo.jain@technicolor.com, georgios@umn.edu
Publisher Copyright:
© 2019 IEEE.
Keywords
- Deep learning
- adversarial examples
- convolutional neural networks
- image classification
- randomized defenses