English

Greedy Attack and Gumbel Attack: Generating Adversarial Examples for Discrete Data

Machine Learning 2018-06-01 v1 Artificial Intelligence Computation and Language Cryptography and Security Machine Learning

Abstract

We present a probabilistic framework for studying adversarial attacks on discrete data. Based on this framework, we derive a perturbation-based method, Greedy Attack, and a scalable learning-based method, Gumbel Attack, that illustrate various tradeoffs in the design of attacks. We demonstrate the effectiveness of these methods using both quantitative metrics and human evaluation on various state-of-the-art models for text classification, including a word-based CNN, a character-based CNN and an LSTM. As as example of our results, we show that the accuracy of character-based convolutional networks drops to the level of random selection by modifying only five characters through Greedy Attack.

Keywords

Cite

@article{arxiv.1805.12316,
  title  = {Greedy Attack and Gumbel Attack: Generating Adversarial Examples for Discrete Data},
  author = {Puyudi Yang and Jianbo Chen and Cho-Jui Hsieh and Jane-Ling Wang and Michael I. Jordan},
  journal= {arXiv preprint arXiv:1805.12316},
  year   = {2018}
}

Comments

The first two authors contributed equally

R2 v1 2026-06-23T02:14:17.252Z