Metraipolis › archive › dc_0ed4faf69a74

FreqDoor: A Hidden Trojan in the Frequency Domain for Backdoor Attacks on Vision-Language Models — arXiv:2609.07048

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.07048
License (as archived)
CC0-1.0
Archived
2026-10-16T15:11:35.755Z
SHA-256 of the archived text
fb781c396b2061f753e32de04699298e6dbac8ef044fcc0b2e9ff0d90b547b6d
Stable URL of this page
https://metraipolis.com/documento/dc_0ed4faf69a74

How to cite

FreqDoor: A Hidden Trojan in the Frequency Domain for Backdoor Attacks on Vision-Language Models — arXiv:2609.07048 — arXiv (en), https://arxiv.org/abs/2609.07048 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0ed4faf69a74 (Metraipolis archive dc_0ed4faf69a74, sha256 fb781c396b2061f753e32de04699298e6dbac8ef044fcc0b2e9ff0d90b547b6d)
BibTeX
@misc{dc0ed4faf69a74,
  title = {FreqDoor: A Hidden Trojan in the Frequency Domain for Backdoor Attacks on Vision-Language Models — arXiv:2609.07048},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.07048},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0ed4faf69a74; sha256 fb781c396b2061f753e32de04699298e6dbac8ef044fcc0b2e9ff0d90b547b6d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.07048. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-language models (VLMs) have recently shown excellent progress in open-ended image-to-text generation. However, their multimodal nature makes them persistently vulnerable to backdoor attacks. Existing backdoor triggers for VLMs are either spatial, textual, or bimodal, which may yield localized or recognizable trigger patterns. In this work, we explore a different attack surface and propose \ textsc {FreqDoor}, a training-time backdoor attack that implants triggers in the frequency domain. \ textsc {FreqDoor} mixes amplitude-spectrum components from a trigger-source image selectively while preserving the phase of a clean image to generate a spatially distributed and visually imperceptible trigger without modifying the textual input. We evaluate the attack on BLIP-2, InstructBLIP, and LLaVA for image captioning and visual question answering. On Flickr8k, \ textsc {FreqDoor} achieves attack success rates of $99.6\%$, $99.8\%$, and $98.4\%$ on the three models, respectively, while preserving the semantic quality of the generated captions. On VQAv2, the corresponding attack success rates are $99.6\%$, $92.4\%$, and $79.6\%$.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.07048 — 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 2026-10-16T14:49:57.170Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).