Metraipolis › archive › dc_ca990176576d

Test-Time Scaling for Safe Text-Guided Image Generation via Intermediate Clean Estimates — arXiv:2608.03284

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.03284
License (as archived): CC0-1.0
Archived: 2026-10-31T12:18:44.447Z
SHA-256 of the archived text: 910c1c00bc9a40071dbab19c33511c2d6916d77dc1d27a72188018b3cd6c0be6
Stable URL of this page: https://metraipolis.com/documento/dc_ca990176576d

How to cite

Test-Time Scaling for Safe Text-Guided Image Generation via Intermediate Clean Estimates — arXiv:2608.03284 — arXiv (en), https://arxiv.org/abs/2608.03284 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ca990176576d (Metraipolis archive dc_ca990176576d, sha256 910c1c00bc9a40071dbab19c33511c2d6916d77dc1d27a72188018b3cd6c0be6)

BibTeX

@misc{dcca990176576d,
  title = {Test-Time Scaling for Safe Text-Guided Image Generation via Intermediate Clean Estimates — arXiv:2608.03284},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.03284},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ca990176576d; sha256 910c1c00bc9a40071dbab19c33511c2d6916d77dc1d27a72188018b3cd6c0be6},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.03284. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Ensuring safety and policy compliance in text-to-image diffusion models remains a critical challenge, as benign or adversarial prompts can often elicit prohibited content, e.g. nudity and protected intellectual property. While training-based unlearning methods are effective, they are computationally expensive and prone to catastrophic interference with general capabilities. Conversely, existing test-time defenses are primarily prompt-centric, relying on modifying textual descriptions only, and overlook the visual signals for detection. In this paper, we propose to leverage the intermediate clean image estimated during the generation process and employ a sparse margin objective to detect prohibited concepts. When a violation is detected, we immediately intervene by optimizing a structured low-rank residual in the text-conditioning space via truncated backpropagation. This design allows weight-preserving detection, keeps non-violating inference latency nearly unchanged as the maximum budget increases, and offers flexibility in safety performance via test-time scaling. Extensive experiments on Stable Diffusion v1.4 and v3.5 across nudity removal, IP protection, and style erasure demonstrate superior performance across suppression, fidelity and preservation compared to prior weight-preserving baselines, providing a scalable and flexible solution for safe generative deployment.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.AI).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.03284 — 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 2026-10-31T12:05:53.028Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).