Compositional Concepts Erasing: Disentangling Object-Style via Domain Manifold Unlearning

arXiv:2610.14514v1 Announce Type: new Abstract: Diffusion models pose significant risks regarding the generation of illicit content and copyright violations. This has catalyzed the development of machine unlearning to erase harmful concepts or copyrighted styles from a model's generative repertoire without compromising its general performance. However, previous unlearning approaches on diffusion models have largely focused on either object-level or style-level removal and failed to address the challenge of object-style combination unlearning. At the same time, multiple unlearning requests often lead to catastrophic forgetting. To address these challenges, we propose Domain Manifold Unlearning (DMU), which uses an EMA-updated feature proxy extracted from intermediate diffusion features to represent the retained target-object direction. As a lightweight estimating representation of an object, DMU enables selective unlearning of domain-specific components in a specific object while preserving other semantic representations, thereby significantly enhancing robustness in multiple combination unlearning scenarios. Extensive experiments show that DMU significantly mitigates catastrophic forgetting, outperforming state-of-the-art methods of batch unlearning and sequential unlearning.

Related

• Disentangled Sparse Representations for Concept-Separated Diffusion Unlearning
• Dynamic Eraser for Guided Concept Erasure in Diffusion Models
• Compositional Concept Erasure in Text-to-Image Diffusion Models via Hierarchically Grounded Semantic Surgery
• Rethinking Robust Adversarial Concept Erasure in Diffusion Models
• RASteer: Retain-Aware Activation Steering for Concept Erasure in Diffusion Models

Source: arXiv cs.CV | 2026-10-14