DANCin SEQ2SEQ: Fooling Text Classifiers with Adversarial Text Example Generation

TL;DR

This paper introduces DANCin SEQ2SEQ, a GAN-inspired reinforcement learning method for generating adversarial text examples that can fool black-box classifiers, providing insights into model vulnerabilities.

Contribution

The work presents a novel GAN-inspired algorithm for adversarial text generation applicable to black-box classifiers, recasting the problem as reinforcement learning.

Findings

Preliminary results show promising adversarial example generation.

The method works in black-box attack scenarios.

Semantic meaningfulness of generated adversarial texts is demonstrated.

Abstract

Machine learning models are powerful but fallible. Generating adversarial examples - inputs deliberately crafted to cause model misclassification or other errors - can yield important insight into model assumptions and vulnerabilities. Despite significant recent work on adversarial example generation targeting image classifiers, relatively little work exists exploring adversarial example generation for text classifiers; additionally, many existing adversarial example generation algorithms require full access to target model parameters, rendering them impractical for many real-world attacks. In this work, we introduce DANCin SEQ2SEQ, a GAN-inspired algorithm for adversarial text example generation targeting largely black-box text classifiers. We recast adversarial text example generation as a reinforcement learning problem, and demonstrate that our algorithm offers preliminary but promising steps towards generating semantically meaningful adversarial text examples in a real-world attack scenario.