Journal
REMOTE SENSING
Volume 14, Issue 22, Pages -Publisher
MDPI
DOI: 10.3390/rs14225833
Keywords
adversarial examples; targeted adversarial attacks; scene classification; remote sensing; deep learning
Categories
Funding
- Ministry of Education, Republic of Singapore [RG61/22]
Ask authors/readers for more resources
Researchers are focusing on the vulnerabilities of deep learning models for remote sensing and propose two variants of targeted universal adversarial examples. Extensive experiments demonstrate the strong attackability of these targeted adversarial variants, inspiring research on defenses against adversarial examples in remote sensing.
Researchers are focusing on the vulnerabilities of deep learning models for remote sensing; various attack methods have been proposed, including universal adversarial examples. Existing universal adversarial examples, however, are only designed to fool deep learning models rather than target specific goals, i.e., targeted attacks. To this end, we propose two variants of universal adversarial examples called targeted universal adversarial examples and source-targeted universal adversarial examples. Extensive experiments on three popular datasets showed strong attackability of the two targeted adversarial variants. We hope such strong attacks can inspire and motivate research on the defenses against adversarial examples in remote sensing.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available