EUVIMEDEUVIMED entdecken
Crossref

CluCERT: Certifying LLM Robustness via Clustering-Guided Denoising Smoothing

Zixia Wang, Gaojie Jin, Jia Hu, Ronghui Mu

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (44) · 37998-38006

DOI: 10.1609/aaai.v40i44.41137 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Zixia Wang und 3 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Recent advancements in Large Language Models (LLMs) have led to their widespread adoption in daily applications. Despite their impressive capabilities, they remain vulnerable to adversarial attacks, as even minor meaning-preserving changes such as synonym substitutions can lead to incorrect predictions. As a result, certifying the robustness of LLMs against such adversarial prompts is of vital importance. Existing approaches focused on word deletion or simple denoising strategies to achieve robustness certification. However, these methods face two critical limitations: (1) they yield loose robustness bounds due to the lack of semantic validation for perturbed outputs and (2) they suffer from high computational costs due to repeated sampling. To address these limitations, we propose CluCERT, a novel framework for certifying LLM robustness via clustering-guided denoising smoothing. Specifically, to achieve tighter certified bounds, we introduce a semantic clustering filter that reduces noisy samples and retains meaningful perturbations, supported by theoretical analysis. Furthermore, we enhance computational efficiency through two mechanisms: a refine module that extracts core semantics, and a fast synonym substitution strategy that accelerates the denoising process. Finally, we conduct extensive experiments on various downstream tasks and jailbreak defense scenarios. Experimental results demonstrate that our method outperforms existing certified approaches in both robustness bounds and computational efficiency.

Zitieren

APA

Wang, Z., Jin, G., Hu, J., & Mu, R. (2026). CluCERT: Certifying LLM Robustness via Clustering-Guided Denoising Smoothing. Proceedings of the AAAI Conference on Artificial Intelligence, 40(44), 37998-38006. https://doi.org/10.1609/aaai.v40i44.41137

Vancouver

Wang Z, Jin G, Hu J, Mu R. CluCERT: Certifying LLM Robustness via Clustering-Guided Denoising Smoothing. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(44):37998-38006. doi:10.1609/aaai.v40i44.41137

BibTeX
@article{Wang2026,
  author = {Wang, Zixia and Jin, Gaojie and Hu, Jia and Mu, Ronghui},
  title = {CluCERT: Certifying LLM Robustness via Clustering-Guided Denoising Smoothing},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {44},
  pages = {37998--38006},
  doi = {10.1609/aaai.v40i44.41137}
}
Kostenlos starten →