EUVIMEDEUVIMED entdecken
Crossref

Breaking the Adversarial Robustness-Performance Trade-off in Text Classification via Manifold Purification

Chenhao Dang, Jing Ma

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (36) · 30476-30484

DOI: 10.1609/aaai.v40i36.40301 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Chenhao Dang und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

A persistent challenge in text classification (TC) is that enhancing model robustness against adversarial attacks typically degrades performance on clean data. We argue that this challenge can be resolved by modeling the distribution of clean samples in the encoder’s embedding manifold. To this end, we propose the Manifold-Correcting Causal Flow (MC²F), a two-module system that operates directly on sentence embeddings. A Stratified Riemannian Continuous Normalizing Flow (SR-CNF) learns the density of the clean data manifold. It identifies out-of-distribution embeddings, which are then corrected by a Geodesic Purification Solver. This solver projects adversarial points back onto the learned manifold via the shortest path, restoring a clean, semantically coherent representation. We conducted extensive evaluations on text classification (TC) across three datasets and multiple adversarial attacks. The results demonstrate that our method, MC²F, not only establishes a new state-of-the-art in adversarial robustness but also fully preserves performance on clean data, even yielding modest gains in Accuracy.

Zitieren

APA

Dang, C., & Ma, J. (2026). Breaking the Adversarial Robustness-Performance Trade-off in Text Classification via Manifold Purification. Proceedings of the AAAI Conference on Artificial Intelligence, 40(36), 30476-30484. https://doi.org/10.1609/aaai.v40i36.40301

Vancouver

Dang C, Ma J. Breaking the Adversarial Robustness-Performance Trade-off in Text Classification via Manifold Purification. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(36):30476-30484. doi:10.1609/aaai.v40i36.40301

BibTeX
@article{Dang2026,
  author = {Dang, Chenhao and Ma, Jing},
  title = {Breaking the Adversarial Robustness-Performance Trade-off in Text Classification via Manifold Purification},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {36},
  pages = {30476--30484},
  doi = {10.1609/aaai.v40i36.40301}
}
Kostenlos starten →