We propose an efficient method to generate white-box adversarial examples to trick a character-level neural classifier. We find that only a few manipulations are needed to greatly decrease the accuracy. Our method relies on an atomic flip operation, which swaps one token for another, based on the gradients of the one-hot input vectors. Due to efficiency of our method, we can perform adversarial training which makes the model more robust to attacks at test time. With the use of a few semantics-preserving constraints, we demonstrate that HotFlip can be adapted to attack a word-level classifier as well.
@article{arxiv.1712.06751,
title = {HotFlip: White-Box Adversarial Examples for Text Classification},
author = {Javid Ebrahimi and Anyi Rao and Daniel Lowd and Dejing Dou},
journal= {arXiv preprint arXiv:1712.06751},
year = {2018}
}