EUVIMEDEUVIMED entdecken
Crossref

3D-ANC: Adaptive Neural Collapse for Robust 3D Point Cloud Recognition

Yuanmin Huang, Wenxuan Li, Mi Zhang, Xiaohan Zhang, Xiaoyu You, Min Yang

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (7) · 5194-5202

DOI: 10.1609/aaai.v40i7.37434 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Yuanmin Huang und 5 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep neural networks have recently achieved notable progress in 3D point cloud recognition, yet their vulnerability to adversarial perturbations poses critical security challenges in practical deployments. Conventional defense mechanisms struggle to address the evolving landscape of multifaceted attack patterns. Through systematic analysis of existing defenses, we identify that their unsatisfactory performance primarily originates from an entangled feature space, where adversarial attacks can be performed easily. To this end, we present 3D-ANC, a novel approach that capitalizes on the Neural Collapse (NC) mechanism to orchestrate discriminative feature learning. In particular, NC depicts where last-layer features and classifier weights jointly evolve into a simplex equiangular tight frame (ETF) arrangement, establishing maximally separable class prototypes. However, leveraging this advantage in 3D recognition confronts two substantial challenges: (1) prevalent class imbalance in point cloud datasets, and (2) complex geometric similarities between object categories. To tackle these obstacles, our solution combines an ETF-aligned classification module with an adaptive training framework consisting of representation-balanced learning (RBL) and dynamic feature direction loss (FDL). 3D-ANC seamlessly empowers existing models to develop disentangled feature spaces despite the complexity in 3D data distribution. Comprehensive evaluations state that 3D-ANC significantly improves the robustness of models with various structures on two datasets. For instance, DGCNN's classification accuracy is elevated from 27.2% to 80.9% on ModelNet40 -- a 53.7% absolute gain that surpasses leading baselines by 34.0%.

Zitieren

APA

Huang, Y., Li, W., Zhang, M., Zhang, X., You, X., & Yang, M. (2026). 3D-ANC: Adaptive Neural Collapse for Robust 3D Point Cloud Recognition. Proceedings of the AAAI Conference on Artificial Intelligence, 40(7), 5194-5202. https://doi.org/10.1609/aaai.v40i7.37434

Vancouver

Huang Y, Li W, Zhang M, Zhang X, You X, Yang M. 3D-ANC: Adaptive Neural Collapse for Robust 3D Point Cloud Recognition. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(7):5194-5202. doi:10.1609/aaai.v40i7.37434

BibTeX
@article{Huang2026,
  author = {Huang, Yuanmin and Li, Wenxuan and Zhang, Mi and Zhang, Xiaohan and You, Xiaoyu and Yang, Min},
  title = {3D-ANC: Adaptive Neural Collapse for Robust 3D Point Cloud Recognition},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {7},
  pages = {5194--5202},
  doi = {10.1609/aaai.v40i7.37434}
}
Kostenlos starten →