The Latent Guardian: Defending Collaborative Perception via Feature-Level Consistency Verification

Zhuangzhuang Zhang, Mingxin Li, Libing Wu, Wei-Bin Lee, Jianping Wang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:156696-156712, 2026.

Abstract

Collaborative perception (CP) significantly extends the sensing range of connected and autonomous vehicles (CAVs). However, its reliance on data fusion among multiple CAVs makes it inherently vulnerable to adversarial attacks from malicious participants. Existing defenses primarily rely on output-level consensus, assuming that malicious messages manifest as statistical outliers, while suffering from poor adaptability to environmental noise. This makes them vulnerable to stealthy adversarial attacks and prone to high false positive rates. To address this challenge, we shift the defense paradigm from superficial output-level consensus to deeper consistency within the internal feature space. Guided by this principle, we propose Cerberus, a novel defense framework against adversarial attacks in CP systems by leveraging multi-dimensional consistency in the feature space. By quantifying conflicts in topological structure, semantic direction, and energy distribution within feature maps, Cerberus effectively detects adversarial perturbations and provides dynamic protection against adversarial attacks. Experimental results demonstrate that Cerberus significantly outperforms state-of-the-art methods, effectively limiting the attack success rate to as low as 0.05% while restoring the AP to 0.88.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-zhang26cs, title = {The Latent Guardian: Defending Collaborative Perception via Feature-Level Consistency Verification}, author = {Zhang, Zhuangzhuang and Li, Mingxin and Wu, Libing and Lee, Wei-Bin and Wang, Jianping}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {156696--156712}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/zhang26cs/zhang26cs.pdf}, url = {https://proceedings.mlr.press/v306/zhang26cs.html}, abstract = {Collaborative perception (CP) significantly extends the sensing range of connected and autonomous vehicles (CAVs). However, its reliance on data fusion among multiple CAVs makes it inherently vulnerable to adversarial attacks from malicious participants. Existing defenses primarily rely on output-level consensus, assuming that malicious messages manifest as statistical outliers, while suffering from poor adaptability to environmental noise. This makes them vulnerable to stealthy adversarial attacks and prone to high false positive rates. To address this challenge, we shift the defense paradigm from superficial output-level consensus to deeper consistency within the internal feature space. Guided by this principle, we propose Cerberus, a novel defense framework against adversarial attacks in CP systems by leveraging multi-dimensional consistency in the feature space. By quantifying conflicts in topological structure, semantic direction, and energy distribution within feature maps, Cerberus effectively detects adversarial perturbations and provides dynamic protection against adversarial attacks. Experimental results demonstrate that Cerberus significantly outperforms state-of-the-art methods, effectively limiting the attack success rate to as low as 0.05% while restoring the AP to 0.88.} }
Endnote
%0 Conference Paper %T The Latent Guardian: Defending Collaborative Perception via Feature-Level Consistency Verification %A Zhuangzhuang Zhang %A Mingxin Li %A Libing Wu %A Wei-Bin Lee %A Jianping Wang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-zhang26cs %I PMLR %P 156696--156712 %U https://proceedings.mlr.press/v306/zhang26cs.html %V 306 %X Collaborative perception (CP) significantly extends the sensing range of connected and autonomous vehicles (CAVs). However, its reliance on data fusion among multiple CAVs makes it inherently vulnerable to adversarial attacks from malicious participants. Existing defenses primarily rely on output-level consensus, assuming that malicious messages manifest as statistical outliers, while suffering from poor adaptability to environmental noise. This makes them vulnerable to stealthy adversarial attacks and prone to high false positive rates. To address this challenge, we shift the defense paradigm from superficial output-level consensus to deeper consistency within the internal feature space. Guided by this principle, we propose Cerberus, a novel defense framework against adversarial attacks in CP systems by leveraging multi-dimensional consistency in the feature space. By quantifying conflicts in topological structure, semantic direction, and energy distribution within feature maps, Cerberus effectively detects adversarial perturbations and provides dynamic protection against adversarial attacks. Experimental results demonstrate that Cerberus significantly outperforms state-of-the-art methods, effectively limiting the attack success rate to as low as 0.05% while restoring the AP to 0.88.
APA
Zhang, Z., Li, M., Wu, L., Lee, W. & Wang, J.. (2026). The Latent Guardian: Defending Collaborative Perception via Feature-Level Consistency Verification. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:156696-156712 Available from https://proceedings.mlr.press/v306/zhang26cs.html.

Related Material