Break it, Imitate it, Fix it: Robustness by Generating Human-Like Attacks

Published: 29 Jan 2024, Last Modified: 29 Jan 2024Accepted by TMLREveryoneRevisionsBibTeX
Authors that are also TMLR Expert Reviewers: ~Ahmad_Beirami1
Abstract: Real-world natural language processing systems need to be robust to human adversaries. Collecting examples of human adversaries for training is an effective but expensive solution. On the other hand, training on synthetic attacks with small perturbations---such as word-substitution---does not actually improve robustness to human adversaries. In this paper, we propose an adversarial training framework that uses limited human adversarial examples to generate more useful adversarial examples at scale. We demonstrate the advantages of this system on the ANLI and hate speech detection benchmark datasets---both collected via an iterative, adversarial human-and-model-in-the-loop procedure. Compared to training only on observed human attacks, also training on our synthetic adversarial examples improves model robustness to future rounds. In ANLI, we see accuracy gains on the current set of attacks (44.1\%$\,\to\,$50.1\%) and on two future unseen rounds of human generated attacks (32.5\%$\,\to\,$43.4\%, and 29.4\%$\,\to\,$40.2\%). In hate speech detection, we see AUC gains on current attacks (0.76 $\to$ 0.84) and a future round (0.77 $\to$ 0.79). Attacks from methods that do not learn the distribution of existing human adversaries, meanwhile, degrade robustness.
Certifications: Expert Certification
Submission Length: Regular submission (no more than 12 pages of main content)
Previous TMLR Submission Url:
Changes Since Last Submission: Font changed to reflect TMLR template
Assigned Action Editor: ~Pin-Yu_Chen1
License: Creative Commons Attribution 4.0 International (CC BY 4.0)
Submission Number: 1784