mlsanityData to 7 Oct 2026

NCL: Textual Backdoor Defense Using Noise-augmented Contrastive Learning

arXiv 2303.01742 ↗First posted 3 Mar 2023cs.CR cs.CLcited by 7

On the map: LLM safety and security → Backdoor and data poisoning attacks

Abstract (arXiv)

At present, backdoor attacks attract attention as they do great harm to deep learning models. The adversary poisons the training data making the model being injected with a backdoor after being trained unconsciously by victims using the poisoned dataset. In the field of text, however, existing works do not provide sufficient defense against backdoor attacks. In this paper, we propose a Noise-augmented Contrastive Learning (NCL) framework to defend against textual backdoor attacks when training models with untrustworthy data. With the aim of mitigating the mapping between triggers and the target label, we add appropriate noise perturbing possible backdoor triggers, augment the training dataset, and then pull homology samples in the feature space utilizing contrastive learning objective. Experiments demonstrate the effectiveness of our method in defending three types of textual backdoor attacks, outperforming the prior works.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Placed in Backdoor and data poisoning attacks, one of 12 directions in LLM safety and security. Builds on 18 of its 23 references that are in the map. First posted on arXiv 3 Mar 2023.

Builds on

18 of 23 references.

Closest ancestors · 4

Same direction, newest first

  1. 2021Mind the Style of Text! Adversarial and Backdoor Attacks Based on Text Style Transfer14 Oct
  2. 2021BFClass: A Backdoor-free Text Classification Framework22 Sep
  3. 2021Hidden Killer: Invisible Textual Backdoor Attacks with Syntactic Trigger26 May
  4. 2021Hidden Backdoors in Human-Centric Language Models1 May

Other references · 14

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,266
  3. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,243
  4. 2015Character-level Convolutional Networks for Text Classificationcited by 2,293
  5. 2021SimCSE: Simple Contrastive Learning of Sentence Embeddingscited by 2,168
  6. 2020Weight Poisoning Attacks on Pre-trained Modelscited by 318
  7. 2018Adversarial Example Generation with Syntactically Controlled Paraphrase Networkscited by 282
  8. 2020Supervised Contrastive Learning for Pre-trained Language Model Fine-tuningcited by 275
  9. 2020ONION: A Simple and Effective Defense Against Textual Backdoor Attackscited by 231
  10. 2019A backdoor attack against LSTM-based text classification systemscited by 214
  11. 2020BadNL: Backdoor Attacks against NLP Models with Semantic-preserving Improvementscited by 191
  12. 2020Reformulating Unsupervised Style Transfer as Paraphrase Generationcited by 181
2 more
  1. 2020Graph Backdoorcited by 112
  2. 2020Mitigating backdoor attacks in LSTM-based Text Classification Systems by Backdoor Keyword Identificationcited by 80

Built on by

Cited by year

Citing papers per year: 2023: 2; 2024: 2; 2025: 3.

7 papers.

  1. 2023Text-to-Image Diffusion Models can be Easily Backdoored through Multimodal Data Poisoningcited by 91
  2. 2024A Survey of Recent Backdoor Attacks and Defenses in Large Language Modelscited by 36
  3. 2023Data and Model Poisoning Backdoor Attacks on Wireless Federated Learning, and the Defense Mechanisms: A Comprehensive Surveycited by 17
  4. 2025Efficient Input-level Backdoor Defense on Text-to-Image Synthesis via Neuron Activation Variationcited by 8
  5. 2024Acquiring Clean Language Models from Backdoor Poisoned Datasets by Downscaling Frequency Spacecited by 6
  6. 2025Paladin: Defending LLM-enabled Phishing Emails with a New Trigger-Tag Paradigmcited by 4
  7. 2025Bridging Distribution Shift and AI Safety: Conceptual and Methodological Synergiescited by 2