Attack to Fool and Explain Deep Networks

Research output: Contribution to journalArticle

Abstract

Deep visual models are susceptible to adversarial perturbations to inputs. Although these signals are carefully crafted, they still appear noise-like patterns to humans. This observation has led to the argument that deep visual representation is misaligned with human perception. We counter-argue by providing evidence of human-meaningful patterns in adversarial perturbations. We first propose an attack that fools a network to confuse a whole category of objects (source class) with a target label. Our attack also limits the unintended fooling by samples from non-sources classes, thereby circumscribing human-defined semantic notions for network fooling. We show that the proposed attack not only leads to the emergence of regular geometric patterns in the perturbations, but also reveals insightful information about the decision boundaries of deep models. Exploring this phenomenon further, we alter the ‘adversarial’ objective of our attack to use it as a tool to ‘explain’ deep visual representation. We show that by careful channeling and projection of the perturbations computed by our method, we can visualize a model's understanding of human-defined semantic notions. Finally, we exploit the explanability properties of our perturbations to perform image generation, inpainting and interactive image manipulation by attacking adversarially robust ‘classifiers’.
Original languageEnglish
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
DOIs
Publication statusE-pub ahead of print - 26 May 2021

Fingerprint

Dive into the research topics of 'Attack to Fool and Explain Deep Networks'. Together they form a unique fingerprint.

Cite this