Metraipolis › archive › dc_298576c026bf
Understanding Backdoor Attacks in Scene Graph Generation via Global Context Pathway — arXiv:2611.04379
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.04379
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-04T12:29:54.309Z
- SHA-256 of the archived text
691d1cb8170f05ca3f64455021c31b18c5145a11b39a2c2248d28be013773dcd- Stable URL of this page
https://metraipolis.com/documento/dc_298576c026bf
How to cite
Understanding Backdoor Attacks in Scene Graph Generation via Global Context Pathway — arXiv:2611.04379 — arXiv (en), https://arxiv.org/abs/2611.04379 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_298576c026bf (Metraipolis archive dc_298576c026bf, sha256 691d1cb8170f05ca3f64455021c31b18c5145a11b39a2c2248d28be013773dcd)
BibTeX
@misc{dc298576c026bf,
title = {Understanding Backdoor Attacks in Scene Graph Generation via Global Context Pathway — arXiv:2611.04379},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.04379},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_298576c026bf; sha256 691d1cb8170f05ca3f64455021c31b18c5145a11b39a2c2248d28be013773dcd},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.04379. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Scene Graph Generation (SGG) produces structured scene representations, yet its training-time backdoor risk remains unexplored. Unlike classification, SGG exposes a pair-level attack surface that enables selective corruption of relations involving a designated object category. However, because SGG largely relies on localized evidence, globally placed triggers may face difficulty in manipulating such predictions. Recent SGG architectures increasingly incorporate global context to improve scene modeling, which may in turn facilitate such attacks. In this paper, we systematically study this backdoor risk and investigate the role of the global context pathway in backdoor behavior. Concretely, we first instantiate this attack with an effective method, and propose a three-axis evaluation protocol to characterize backdoor behavior. Experiments on Visual Genome with representative open-vocabulary and closed-set SGG models demonstrate the effectiveness of the attack. Mechanistic analyses reveal that the explicit global context pathway serves as a major carrier of the backdoor effect across the evaluated SGG models. In particular, replacing the poisoned global context representation with its clean counterpart during inference sharply suppresses the attack, yielding 100% and 97% reductions in attack success for the OvSGTR family and RelationFormer, respectively. Moreover, training-time interventions in the OvSGTR family show that suppressing trigger signals in this pathway either reduces attack success or substantially degrades utility. Together, these findings suggest that explicit global context designs may pose a backdoor risk in SGG architectures. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2611.04379 — 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 2026-11-04T12:26:47.274Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).