arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2008.05536cs.CLcs.LG
收藏

基于文本分类的模型鲁棒性:语义保持的对抗攻击

Model Robustness with Text Classification: Semantic-preserving adversarial attacks

Wells Fargo(富国银行)

Rahul Singh, Tarun Joshi, Vijayan N. Nair, Agus Sudjianto

摘要:

我们提出了创建对抗攻击的算法,以评估模型在文本分类问题中的鲁棒性。它们可以用来创建白盒攻击和黑盒攻击,同时保留原始文本的语义和语法。这些攻击在白盒设置中导致大量的翻转,并且可以在黑盒设置中使用相同的规则。在黑盒设置中,创建的攻击能够逆转基于Transformer的架构的决策。

英文摘要:

We propose algorithms to create adversarial attacks to assess model robustness in text classification problems. They can be used to create white box attacks and black box attacks while at the same time preserving the semantics and syntax of the original text. The attacks cause significant number of flips in white-box setting and same rule based can be used in black-box setting. In a black-box setting, the attacks created are able to reverse decisions of transformer based architectures.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑