Corpus ID: 225041208

Rewriting Meaningful Sentences via Conditional BERT Sampling and an application on fooling text classifiers

@article{Xu2020RewritingMS,
  title={Rewriting Meaningful Sentences via Conditional BERT Sampling and an application on fooling text classifiers},
  author={Lei Xu and Ivan Ramirez and K. Veeramachaneni},
  journal={ArXiv},
  year={2020},
  volume={abs/2010.11869}
}
Most adversarial attack methods that are designed to deceive a text classifier change the text classifier's prediction by modifying a few words or characters. Few try to attack classifiers by rewriting a whole sentence, due to the difficulties inherent in sentence-level rephrasing as well as the problem of setting the criteria for legitimate rewriting. In this paper, we explore the problem of creating adversarial examples with sentence-level rewriting. We design a new sampling method, named… Expand

References

SHOWING 1-10 OF 56 REFERENCES
Is BERT Really Robust? Natural Language Attack on Text Classification and Entailment
  • 82
  • Highly Influential
  • PDF
Certified Robustness to Adversarial Word Substitutions
  • 71
  • Highly Influential
  • PDF
Deep Text Classification Can be Fooled
  • 181
  • PDF
Generating Adversarial Text Samples
  • 14
DANCin SEQ2SEQ: Fooling Text Classifiers with Adversarial Text Example Generation
  • 16
  • PDF
Robust Neural Machine Translation with Doubly Adversarial Inputs
  • 76
  • PDF
Language Models are Unsupervised Multitask Learners
  • 2,929
  • Highly Influential
  • PDF
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
  • 17,055
  • Highly Influential
  • PDF
XLNet: Generalized Autoregressive Pretraining for Language Understanding
  • 2,370
  • PDF
...
1
2
3
4
5
...