Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

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

About

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.

Yantao Lu, Yunhan Jia, Jianyu Wang, Bai Li, Weiheng Chai, Lawrence Carin, Senem Velipasalar• 2019

Related benchmarks

TaskDatasetResultRank
Image ClassificationSVHN (test)--
199
Video RecognitionUCF101--
64
Video RecognitionKinetics 400 (test)
ASR52.25
54
Visual ReasoningNLVR2--
49
Image CaptioningMSCOCO (test)
CIDEr95.52
29
Visual EntailmentSNLI-VE
Accuracy0.1371
24
RECRefCOCO+
ASR26.3
16
RECRefCOCOg
ASR26.39
16
Image ClassificationImageNet-1K
ASR10.43
4
Showing 9 of 9 rows

Other info

Follow for update