Enhancing Cross-task Black-Box Transferability of Adversarial Examples with Dispersion Reduction

Neural networks are known to be vulnerable to carefully crafted adversarial examples, and these malicious samples often transfer, i.e., they remain adversarial even against other models. Although great efforts have been delved into the transferability across models, surprisingly, less attention has...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2019-11
Main Authors: Lu, Yantao, Jia, Yunhan, Wang, Jianyu, Bai, Li, Chai, Weiheng, Lawrence, Carin, Velipasalar, Senem
Format: Article
Language:English
Subjects:
Cloud computing
Computer vision
Dispersion
Feature maps
Image classification
Image detection
Image segmentation
Neural networks
Object recognition
Performance degradation
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Neural networks are known to be vulnerable to carefully crafted adversarial examples, and these malicious samples often transfer, i.e., they remain adversarial even against other models. Although great efforts have been delved into the transferability across models, surprisingly, less attention has been paid to the cross-task transferability, which represents the real-world cybercriminal's situation, where an ensemble of different defense/detection mechanisms need to be evaded all at once. In this paper, we investigate the transferability of adversarial examples across a wide range of real-world computer vision tasks, including image classification, object detection, semantic segmentation, explicit content detection, and text detection. Our proposed attack minimizes the ``dispersion'' of the internal feature map, which overcomes existing attacks' limitation of requiring task-specific loss functions and/or probing a target model. We conduct evaluation on open source detection and segmentation models as well as four different computer vision tasks provided by Google Cloud Vision (GCV) APIs, to show how our approach outperforms existing attacks by degrading performance of multiple CV tasks by a large margin with only modest perturbations linf=16.
ISSN:2331-8422