Metraipolis › archive › dc_7e0770a94f63

Is Forgetting Unified? Exploring Concept Unlearning in Unified Multimodal Models — arXiv:2611.04467

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.04467
License (as archived)
CC0-1.0
Archived
2026-11-04T12:56:26.604Z
SHA-256 of the archived text
32510317908841efb10245ec786176daa1a40cdf4c8631143bfa0290f49fd6db
Stable URL of this page
https://metraipolis.com/documento/dc_7e0770a94f63

How to cite

Is Forgetting Unified? Exploring Concept Unlearning in Unified Multimodal Models — arXiv:2611.04467 — arXiv (en), https://arxiv.org/abs/2611.04467 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7e0770a94f63 (Metraipolis archive dc_7e0770a94f63, sha256 32510317908841efb10245ec786176daa1a40cdf4c8631143bfa0290f49fd6db)
BibTeX
@misc{dc7e0770a94f63,
  title = {Is Forgetting Unified? Exploring Concept Unlearning in Unified Multimodal Models — arXiv:2611.04467},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.04467},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7e0770a94f63; sha256 32510317908841efb10245ec786176daa1a40cdf4c8631143bfa0290f49fd6db},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.04467. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Unified multimodal models (UMMs) integrate image generation and visual understanding, but are existing single-pathway unlearning methods sufficient to suppress target concepts across both? In this work, we study cross-pathway concept unlearning. Our goal is to prevent a model from both generating images of a target concept and identifying it in input images, while preserving other capabilities. To this end, we introduce UUBench, a diagnostic benchmark covering 30 target concepts and 60 semantically related neighboring concepts. Experiments on representative UMMs show that the evaluated single-pathway methods achieve limited joint forgetting: generation-side editors strongly suppress target generation but yield limited forgetting in visual understanding, while understanding-side baselines struggle to achieve substantial forgetting even in their own pathway. Their direct composition also yields limited joint forgetting. We therefore propose UniForget, achieving joint forgetting rates of 80.0% on BAGEL-7B-MoT and 92.7% on Janus-Pro-7B, outperforming representative baselines. Further analysis reveals trade-offs between joint forgetting and capability preservation.

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/2611.04467 — 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 2026-11-04T12:33:47.151Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).