ONION: A Simple and Effective Defense Against Textual Backdoor Attacks
By Fanchao Qi, Yangyi Chen, Mukai Li, Zhiyuan Liu, Maosong Sun. Conference on Empirical Methods in Natural Language Processing, 2020
414 citations recorded by the source corpus.
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.
Subfields: Artificial Intelligence, Computer Networks and Communications
View Paper: https://www.semanticscholar.org/paper/3a1f8829e641b46f661775f64a7f27b933a46103
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Zhiyuan Liu
- Maosong Sun
- Zhengyan Zhang
- Yangyi Chen
- Goutham Ramakrishnan
- Aws Albarghouthi
- Mukai Li
- Yasheng Wang
In Computer Networks and Communications
Citations in the corpus
- Backdoors in Neural Models of Source Code (2020, background): “More recently, backdoors attacks have been studied for NLP tasks [7, 24, 25], demonstrating the vulnerability of models for text classification, question answering, etc.”
- A Survey on Large Language Model (LLM) Security and Privacy: The Good, the Bad, and the Ugly (2023, background): “Jin et al. [121] conducted a comprehensive study on LLMs’ understanding of binary code semantics [122] across different architectures and optimization levels, providing key insights for future research in this area.”
- Security and Privacy Challenges of Large Language Models: A Survey (2024, neutral): “As a result, the model was shut down just after launching public access [209].”
- Hidden Killer: Invisible Textual Backdoor Attacks with Syntactic Trigger (2021, background): “In consequence, the trigger-embedded poisoned samples would be easily detected and removed (Chen and Dai, 2020; Qi et al., 2020), which leads to the failure of backdoor attacks.”