SLIDE 1
Defense Against Adversarial Images using Web-Scale Nearest-Neighbor Search
Abhimanyu Dubey, Laurens van der Maaten, I. Zeki Yalniz, Yixuan Li and Dhruv Mahajan
Defense Against Adversarial Images using Web-Scale Nearest-Neighbor - - PowerPoint PPT Presentation
Defense Against Adversarial Images using Web-Scale Nearest-Neighbor - - PowerPoint PPT Presentation
Defense Against Adversarial Images using Web-Scale Nearest-Neighbor Search Abhimanyu Dubey, Laurens van der Maaten, I. Zeki Yalniz, Yixuan Li and Dhruv Mahajan Adversarial Images adversarial swan pelican perturbation
SLIDE 2
SLIDE 3
Nearest—Neighbors Defense
- Adversarial perturbations move the input away from the image “manifold”.
- KNN Defense: Project the image back on to the manifold.
clean image adversarial image
Approximation of manifold by tens of billions of images
nearest neighbors
SLIDE 4
Nearest—Neighbors Defense
web-scale database
“swan” “pelican”
adversarial input
KNN
SLIDE 5
Effect of Scale
Log-linear relationship.
PGD Attack
SLIDE 6
Comparison with other defenses
Attack Type: PGD Model: ResNet-50
SLIDE 7
KNN Based Attack
Separate database available to attacker. Overlap: %-age of images shared between attacker’s and defense database
Data obfuscation is a viable defense strategy.
SLIDE 8
Thank You!
Defense Against Adversarial Images using Web-Scale Nearest-Neighbor Search Arxiv: https://arxiv.org/pdf/1903.01612.pdf Poster Session 3.1, Poster #87, 10:00 AM Thursday.