ICED: Concept-level Machine Unlearning via Interpretable Concept Decomposition

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Lin, Shen, Lin, Jing, Dong, Junhao, Koniusz, Piotr, Xu, Li
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866918502875529216
author Lin, Shen
Lin, Jing
Dong, Junhao
Koniusz, Piotr
Xu, Li
author_facet Lin, Shen
Lin, Jing
Dong, Junhao
Koniusz, Piotr
Xu, Li
contents Machine unlearning in Vision-Language Models (VLMs) is typically performed at the image or instance level, making it difficult to precisely remove target knowledge without affecting unrelated semantics. This issue is especially pronounced since a single image often contains multiple entangled concepts, including both target concepts to be forgotten and contextual information that should be preserved. In this paper, we propose an interpretable concept-level unlearning framework for VLMs, which constructs a compact task-specific concept vocabulary from the forgetting set using a multimodal large language model. In addition to modality alignment, visual representations are decomposed into sparse, nonnegative combinations of semantic concepts, providing an explicit interface for fine-grained knowledge manipulation. Based on this decomposition, our method formulates unlearning as concept-level optimization, where target concepts are selectively suppressed while intra-instance non-target semantics and global cross-modal knowledge are preserved. Extensive experiments across both in-domain and out-of-domain forgetting settings demonstrate that our method enables more comprehensive target forgetting, better preserves non-target knowledge within the same image, and maintains competitive model utility compared with existing VLM unlearning methods.
format Preprint
id arxiv_https___arxiv_org_abs_2605_14309
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle ICED: Concept-level Machine Unlearning via Interpretable Concept Decomposition
Lin, Shen
Lin, Jing
Dong, Junhao
Koniusz, Piotr
Xu, Li
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Machine unlearning in Vision-Language Models (VLMs) is typically performed at the image or instance level, making it difficult to precisely remove target knowledge without affecting unrelated semantics. This issue is especially pronounced since a single image often contains multiple entangled concepts, including both target concepts to be forgotten and contextual information that should be preserved. In this paper, we propose an interpretable concept-level unlearning framework for VLMs, which constructs a compact task-specific concept vocabulary from the forgetting set using a multimodal large language model. In addition to modality alignment, visual representations are decomposed into sparse, nonnegative combinations of semantic concepts, providing an explicit interface for fine-grained knowledge manipulation. Based on this decomposition, our method formulates unlearning as concept-level optimization, where target concepts are selectively suppressed while intra-instance non-target semantics and global cross-modal knowledge are preserved. Extensive experiments across both in-domain and out-of-domain forgetting settings demonstrate that our method enables more comprehensive target forgetting, better preserves non-target knowledge within the same image, and maintains competitive model utility compared with existing VLM unlearning methods.
title ICED: Concept-level Machine Unlearning via Interpretable Concept Decomposition
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2605.14309