论文标题
洋葱：对文本后门攻击的简单有效的防御
ONION: A Simple and Effective Defense Against Textual Backdoor Attacks
论文作者
论文摘要
后门攻击是对深神经网络（DNNS）的紧急训练时间威胁。他们可以操纵DNN的产出并具有高度的阴险。在自然语言处理领域，已经提出了一些攻击方法，并在多个流行模型上获得了很高的攻击成功率。然而，很少有关于防御文本后门攻击的研究。在本文中，我们提出了一种名为Onion的简单有效的文本后门防御，该防御基于异常单词检测，据我们所知，它是第一种可以处理所有文本后门攻击情况的方法。实验证明了我们模型在捍卫Bilstm和BERT防止五种不同的后门攻击方面的有效性。本文的所有代码和数据都可以在https://github.com/thunlp/onion上获得。
Backdoor attacks are a kind of emergent training-time threat to deep neural networks (DNNs). They can manipulate the output of DNNs and possess high insidiousness. In the field of natural language processing, some attack methods have been proposed and achieve very high attack success rates on multiple popular models. Nevertheless, there are few studies on defending against textual backdoor attacks. In this paper, we propose a simple and effective textual backdoor defense named ONION, which is based on outlier word detection and, to the best of our knowledge, is the first method that can handle all the textual backdoor attack situations. Experiments demonstrate the effectiveness of our model in defending BiLSTM and BERT against five different backdoor attacks. All the code and data of this paper can be obtained at https://github.com/thunlp/ONION.
