# Self-Supervised Contrastive Learning with Adversarial Perturbations for Defending Word Substitution-based Attacks

arXiv: https://arxiv.org/abs/2107.07610 · HTML: https://mlsanity.com/paper/2107.07610/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 30 references that are in the map.
- First posted on arXiv 15 Jul 2021.

## Builds on (most cited first, 22 of 22)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [A Simple Framework for Contrastive Learning of Visual Representations](https://mlsanity.com/paper/2002.05709/) (2020)
- [Momentum Contrast for Unsupervised Visual Representation Learning](https://mlsanity.com/paper/1911.05722/) (2019)
- [Towards Deep Learning Models Resistant to Adversarial Attacks](https://mlsanity.com/paper/1706.06083/) (2017)
- [Character-level Convolutional Networks for Text Classification](https://mlsanity.com/paper/1509.01626/) (2015)
- [SimCSE: Simple Contrastive Learning of Sentence Embeddings](https://mlsanity.com/paper/2104.08821/) (2021)
- [Aligning Books and Movies: Towards Story-like Visual Explanations by Watching Movies and Reading Books](https://mlsanity.com/paper/1506.06724/) (2015)
- [Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment](https://mlsanity.com/paper/1907.11932/) (2019)
- [Adversarial Examples for Evaluating Reading Comprehension Systems](https://mlsanity.com/paper/1707.07328/) (2017)
- [Generating Natural Language Adversarial Examples](https://mlsanity.com/paper/1804.07998/) (2018)
- [Supervised Contrastive Learning for Pre-trained Language Model Fine-tuning](https://mlsanity.com/paper/2011.01403/) (2020)
- [FreeLB: Enhanced Adversarial Training for Natural Language Understanding](https://mlsanity.com/paper/1909.11764/) (2019)
- [DeCLUTR: Deep Contrastive Learning for Unsupervised Textual Representations](https://mlsanity.com/paper/2006.03659/) (2020)
- [Counter-fitting Word Vectors to Linguistic Constraints](https://mlsanity.com/paper/1603.00892/) (2016)
- [Adversarial Self-Supervised Contrastive Learning](https://mlsanity.com/paper/2006.07589/) (2020)
- [Adversarial Training for Large Neural Language Models](https://mlsanity.com/paper/2004.08994/) (2020)
- [Contrastive Learning for Many-to-many Multilingual Neural Machine Translation](https://mlsanity.com/paper/2105.09501/) (2021)
- [Self-Guided Contrastive Learning for BERT Sentence Representations](https://mlsanity.com/paper/2106.07345/) (2021)
- [InfoBERT: Improving Robustness of Language Models from An Information Theoretic Perspective](https://mlsanity.com/paper/2010.02329/) (2020)
- [Contrastive Learning with Adversarial Perturbations for Conditional Text Generation](https://mlsanity.com/paper/2012.07280/) (2020)
- [CLINE: Contrastive Learning with Semantic Negative Examples for Natural Language Understanding](https://mlsanity.com/paper/2107.00440/) (2021)
- [A Geometry-Inspired Attack for Generating Natural Language Adversarial Examples](https://mlsanity.com/paper/2010.01345/) (2020)

## Built on by (most cited first, 4 of 4)

- [Large Language Models and Causal Inference in Collaboration: A Survey](https://mlsanity.com/paper/2403.09606/) (2024)
- [Assessing Adversarial Robustness of Large Language Models: An Empirical Study](https://mlsanity.com/paper/2405.02764/) (2024)
- [Learning under Label Proportions for Text Classification](https://mlsanity.com/paper/2310.11707/) (2023)
- [Unsupervised Candidate Ranking for Lexical Substitution via Holistic Sentence Semantics](https://mlsanity.com/paper/2509.11513/) (2025)

## Abstract (arXiv)

In this paper, we present an approach to improve the robustness of BERT language models against word substitution-based adversarial attacks by leveraging adversarial perturbations for self-supervised contrastive learning. We create a word-level adversarial attack generating hard positives on-the-fly as adversarial examples during contrastive learning. In contrast to previous works, our method improves model robustness without using any labeled data. Experimental results show that our method improves robustness of BERT against four different word substitution-based adversarial attacks, and combining our method with adversarial training gives higher robustness than adversarial training alone. As our method improves the robustness of BERT purely with unlabeled data, it opens up the possibility of using large text datasets to train robust language models against word substitution-based adversarial attacks.