Metraipolis › archive › dc_3d04097162cf

Your Unlearning Gives You Away: Identifying Erased Concepts in Diffusion Models — arXiv:2610.05601

Source: arXiv
Original: https://arxiv.org/abs/2610.05601
License (as archived): CC0-1.0
Archived: 2026-10-06T05:27:03.581Z
SHA-256 of the archived text: 90873fdc349985f9fa1183046b86e706b281f50929eea4b2b2fd0eea2a899c6b
Stable URL of this page: https://metraipolis.com/documento/dc_3d04097162cf

How to cite

Your Unlearning Gives You Away: Identifying Erased Concepts in Diffusion Models — arXiv:2610.05601 — arXiv, https://arxiv.org/abs/2610.05601 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3d04097162cf (Metraipolis archive dc_3d04097162cf, sha256 90873fdc349985f9fa1183046b86e706b281f50929eea4b2b2fd0eea2a899c6b)

BibTeX

@misc{dc3d04097162cf,
  title = {Your Unlearning Gives You Away: Identifying Erased Concepts in Diffusion Models — arXiv:2610.05601},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.05601},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3d04097162cf; sha256 90873fdc349985f9fa1183046b86e706b281f50929eea4b2b2fd0eea2a899c6b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.05601. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Existing attacks on unlearned diffusion models assume that the erased concepts are known in advance and focus on recovering them. In practice, however, model providers may not disclose which concepts have been removed, and even with access to the original base model, an adversary may still lack a clear target to attack. In this paper, we aim to answer the following critical but overlooked questions: which concepts have been erased from the model, and how many have been erased in total? To this end, we present Tracer, a framework that rapidly and accurately identifies erased concepts and estimates their number. Tracer efficiently identifies erased concepts without generating and classifying images. By combining lightweight spectral analysis of weight footprints, it enables efficient search over large candidate vocabularies. To distinguish multiple erased concepts, we introduce a footprint coverage objective that guides sequential discovery. Tracer estimates the number of erased concepts by detecting a sharp decline in candidate confidence as the selected concepts account for the erasure footprint, without requiring labeled examples for calibration. The framework requires only lightweight linear algebra and limited forward probes, with no prior knowledge of the unlearning algorithm. Experiments across text-to-image and text-to-video backbones and diverse unlearning methods demonstrate that Tracer identifies erased concepts and estimates their number in seconds, achieving 150 to 137,000 times and 133 to 20,000 times speedups over MIA and brute-force search on image and video models, respectively, with substantially higher identification accuracy.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.05601 — 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 2026-10-06T05:24:06.053Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).