Metraipolis › archive › dc_5fbc3a028c28

GraphRectify: Graph-Based Transfer of Adversarial Example Detectors Across Neural Networks — arXiv:2610.10423

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.10423
License (as archived)
CC0-1.0
Archived
2026-10-08T12:48:07.408Z
SHA-256 of the archived text
80a88383efd27d93c1d898cc91debe9fc2e170d872621f9b71b0fda29284ea7b
Stable URL of this page
https://metraipolis.com/documento/dc_5fbc3a028c28

How to cite

GraphRectify: Graph-Based Transfer of Adversarial Example Detectors Across Neural Networks — arXiv:2610.10423 — arXiv (en), https://arxiv.org/abs/2610.10423 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5fbc3a028c28 (Metraipolis archive dc_5fbc3a028c28, sha256 80a88383efd27d93c1d898cc91debe9fc2e170d872621f9b71b0fda29284ea7b)
BibTeX
@misc{dc5fbc3a028c28,
  title = {GraphRectify: Graph-Based Transfer of Adversarial Example Detectors Across Neural Networks — arXiv:2610.10423},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.10423},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5fbc3a028c28; sha256 80a88383efd27d93c1d898cc91debe9fc2e170d872621f9b71b0fda29284ea7b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.10423. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Adversarial example detectors are often tied to the classifier backbone they were trained on, limiting reuse when the protected model is replaced or upgraded. Directly transferring such detectors across backbones is challenging because different networks generally produce incompatible internal representations. We propose GraphRectify, a graph-based framework for transferring adversarial image detectors across classifier backbones. GraphRectify learns a structured representation of intermediate classifier features and adapts representations from a new backbone to the detector learned on the original model, enabling detector reuse. We evaluate GraphRectify across multiple datasets, backbone architectures, and adversarial attacks, including detector-aware adaptive attacks that jointly target the classifier and detector. Across the complete evaluation matrix, GraphRectify achieves higher aggregate ROC-AUC than training a detector from scratch on the new backbone and the evaluated transfer ablations. The gains are particularly strong for transfers between different backbone families and when sufficient data are available. In contrast, training from scratch remains competitive in the most data-limited settings. These results show that adversarial detection knowledge can transfer effectively across heterogeneous classifier architectures rather than being relearned whenever the protected backbone changes.

Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2610.10423 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-08T12:18:15.496Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).