Lune

EMNLP Findings2022

Weight Perturbation as Defense against Adversarial Word Substitutions

Jianhan Xu, Linyang Li, Jiping Zhang, Xiaoqing Zheng, Kai-Wei Chang, Cho-Jui Hsieh, Xuanjing Huang

Abstract

The existence and pervasiveness of textual adversarial examples have raised serious concerns to security-critical applications. Many methods have been developed to defend against adversarial attacks for neural natural language processing (NLP) models. Adversarial training is one of the most successful defense methods by adding some random or intentional perturbations to the original input texts and making the models robust to the perturbed examples. In this study, we explore the feasibility of improving the adversarial robustness of NLP models by performing perturbations in the parameter space rather than the input feature space. The weight perturbation helps to find a better solution (i.e., the values of weights) that minimizes the adversarial loss among other feasible solutions. We found that the weight perturbation can significantly improve the robustness of NLP models when it is combined with the perturbation in the input embedding space, yielding the highest accuracy on both clean and adversarial examples across different datasets.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

1 top-tier paper has cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Towards LLM Unlearning Resilient to Relearning Attacks: A Sharpness-Aware Minimization Perspective and Beyond cites Sharpness-aware Minimization for Efficiently Improving Generalization
  • BERT-ATTACK: Adversarial Attack Against BERT Using BERT cites TextBugger: Generating Adversarial Text Against Real-world Applications
  • Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment cites TextBugger: Generating Adversarial Text Against Real-world Applications
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers1

  1. Towards LLM Unlearning Resilient to Relearning Attacks: A Sharpness-Aware Minimization Perspective and BeyondChongyu Fan, Jinghan Jia, Yihua Zhang, Anil Ramakrishna et al.ICML 2025

Builds on12

  1. Sharpness-aware Minimization for Efficiently Improving GeneralizationPierre Foret, Ariel Kleiner, Hossein Mobahi, Behnam NeyshaburICLR 2021 · 1,861 citations
  2. Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and EntailmentDi Jin, Zhijing Jin, Joey Tianyi Zhou, Peter SzolovitsAAAI 2020 · 1,333 citations
  3. Adversarial Weight Perturbation Helps Robust GeneralizationDongxian Wu, Shu-Tao Xia, Yisen WangNeurIPS 2020 · 917 citations
  4. TextBugger: Generating Adversarial Text Against Real-world ApplicationsJinfeng Li, Shouling Ji, Tianyu Du, Bo Li et al.NDSS 2019 · 876 citations
  5. BERT-ATTACK: Adversarial Attack Against BERT Using BERTLinyang Li, Ruotian Ma, Qipeng Guo, Xiangyang Xue et al.EMNLP 2020 · 529 citations

Related papers

  1. Impact of Adversarial Training on Robustness and Generalizability of Language ModelsEnes Altinisik, Hassan Sajjad, Husrev T. Sencar, Safa Messaoud et al.ACL Findings 2023 · 10 citations
  2. TextGrad: Advancing Robustness Evaluation in NLP by Gradient-Driven OptimizationBairu Hou, Jinghan Jia, Yihua Zhang, Guanhua Zhang et al.ICLR 2023 · 1 citation
  3. Word Level Robustness Enhancement: Fight Perturbation with PerturbationPei Huang, Yuting Yang, Fuqi Jia, Minghao Liu et al.AAAI 2022 · 14 citations
  4. Flooding-X: Improving BERT's Resistance to Adversarial Attacks via Loss-Restricted Fine-TuningQin Liu, Rui Zheng, Bao Rong, Jingyi Liu et al.ACL 2022 · 35 citations
  5. Towards Adversarially Robust Text Classifiers by Learning to Reweight Clean ExamplesJianhan Xu, Cenyuan Zhang, Xiaoqing Zheng, Linyang Li et al.ACL Findings 2022 · 4 citations
Dusk over the sea between two cliffs drawn in fine vertical lines