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

2025-11-10

AI Summary

3D point cloud recognition models suffer from vulnerability to adversarial perturbations and insufficient robustness due to entangled feature representations. To address this, this paper introduces the neural collapse mechanism—previously studied in 2D vision—to 3D point cloud learning for the first time, proposing a decoupled representation learning framework. It integrates a Simplex ETF classifier, Representation Balance Learning (RBL), and a Dynamic Feature Direction Loss (FDL) to achieve discriminative feature alignment under class imbalance and robust geometric structure modeling. On ModelNet40, the method boosts DGCNN’s adversarial accuracy from 27.2% to 80.9%, yielding an absolute improvement of 53.7%, significantly outperforming existing defenses. Key contributions are: (1) pioneering neural-collapse-driven decoupled representation learning for 3D point clouds; and (2) establishing an end-to-end robust training paradigm explicitly tailored to geometric similarity preservation and class-skewed distributions.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Natural Language Processing: Safety and Robustness

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

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%.

Problem

• Enhancing 3D point cloud recognition robustness against adversarial attacks
• Addressing class imbalance and geometric similarities in 3D datasets
• Developing disentangled feature spaces through adaptive neural collapse mechanisms

Innovation

• Adaptive Neural Collapse for robust feature learning
• ETF-aligned classification with balanced representation training
• Dynamic feature direction loss to handle geometric complexities