English

Benign Adversarial Attack: Tricking Models for Goodness

Artificial Intelligence 2022-07-06 v2 Computer Vision and Pattern Recognition

Abstract

In spite of the successful application in many fields, machine learning models today suffer from notorious problems like vulnerability to adversarial examples. Beyond falling into the cat-and-mouse game between adversarial attack and defense, this paper provides alternative perspective to consider adversarial example and explore whether we can exploit it in benign applications. We first attribute adversarial example to the human-model disparity on employing non-semantic features. While largely ignored in classical machine learning mechanisms, non-semantic feature enjoys three interesting characteristics as (1) exclusive to model, (2) critical to affect inference, and (3) utilizable as features. Inspired by this, we present brave new idea of benign adversarial attack to exploit adversarial examples for goodness in three directions: (1) adversarial Turing test, (2) rejecting malicious model application, and (3) adversarial data augmentation. Each direction is positioned with motivation elaboration, justification analysis and prototype applications to showcase its potential.

Keywords

Cite

@article{arxiv.2107.11986,
  title  = {Benign Adversarial Attack: Tricking Models for Goodness},
  author = {Jitao Sang and Xian Zhao and Jiaming Zhang and Zhiyu Lin},
  journal= {arXiv preprint arXiv:2107.11986},
  year   = {2022}
}

Comments

ACM MM2022 Brave New Idea

R2 v1 2026-06-24T04:30:53.620Z