Improving the Robustness of Transformer-based Large Language Models with Dynamic Attention

Abstract

Transformer-based models, such as BERT and GPT, have been widely adopted in natural language processing (NLP) due to their exceptional performance. However, recent studies show their vulnerability to textual adversarial attacks where the model's output can be misled by intentionally manipulating the text inputs. Despite various methods that have been proposed to enhance the model's robustness and mitigate this vulnerability, many require heavy consumption resources (e.g., adversarial training) or only provide limited protection (e.g., defensive dropout). In this paper, we propose a novel method called dynamic attention, tailored for the transformer architecture, to enhance the inherent robustness of the model itself against various adversarial attacks. Our method requires no downstream task knowledge and does not incur additional costs. The proposed dynamic attention consists of two modules: (I) attention rectification, which masks or weakens the attention value of the chosen tokens, and (ii) dynamic modeling, which dynamically builds the set of candidate tokens. Extensive experiments demonstrate that dynamic attention significantly mitigates the impact of adversarial attacks, improving up to 33\% better performance than previous methods against widely-used adversarial attacks. The model-level design of dynamic attention enables it to be easily combined with other defense methods (e.g., adversarial training) to further enhance the model's robustness. Furthermore, we demonstrate that dynamic attention preserves the state-of-the-art robustness space of the original model compared to other dynamic modeling methods.

References in corpus (17)

• Explaining and Harnessing Adversarial Examples
• Language Models are Few-Shot Learners
• Certified Adversarial Robustness via Randomized Smoothing
• Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing
• Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting
• TextBugger: Generating Adversarial Text Against Real-world Applications
• Prefix-Tuning: Optimizing Continuous Prompts for Generation
• Adversarial Attacks on Deep Learning Models in Natural Language Processing: A Survey
• Adversarial Graph Augmentation to Improve Graph Contrastive Learning
• Backdoor Pre-trained Models Can Transfer to All
• Recent Advances in Adversarial Training for Adversarial Robustness
• Improving the Adversarial Robustness of NLP Models by Information Bottleneck
• Gradient Obfuscation Gives a False Sense of Security in Federated Learning
• Why adversarial training can hurt robust accuracy
• Detecting Adversarial Examples from Sensitivity Inconsistency of Spatial-Transform Domain
• Towards Improving Adversarial Training of NLP Models
• Defense of Word-level Adversarial Attacks via Random Substitution Encoding