Saved in:
Bibliographic Details
Main Authors: Thakral, Kartik, Glaser, Tamar, Hassner, Tal, Vatsa, Mayank, Singh, Richa
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.19783
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917968010543104
author Thakral, Kartik
Glaser, Tamar
Hassner, Tal
Vatsa, Mayank
Singh, Richa
author_facet Thakral, Kartik
Glaser, Tamar
Hassner, Tal
Vatsa, Mayank
Singh, Richa
contents Existing unlearning algorithms in text-to-image generative models often fail to preserve the knowledge of semantically related concepts when removing specific target concepts: a challenge known as adjacency. To address this, we propose FADE (Fine grained Attenuation for Diffusion Erasure), introducing adjacency aware unlearning in diffusion models. FADE comprises two components: (1) the Concept Neighborhood, which identifies an adjacency set of related concepts, and (2) Mesh Modules, employing a structured combination of Expungement, Adjacency, and Guidance loss components. These enable precise erasure of target concepts while preserving fidelity across related and unrelated concepts. Evaluated on datasets like Stanford Dogs, Oxford Flowers, CUB, I2P, Imagenette, and ImageNet1k, FADE effectively removes target concepts with minimal impact on correlated concepts, achieving atleast a 12% improvement in retention performance over state-of-the-art methods.
format Preprint
id arxiv_https___arxiv_org_abs_2503_19783
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Fine-Grained Erasure in Text-to-Image Diffusion-based Foundation Models
Thakral, Kartik
Glaser, Tamar
Hassner, Tal
Vatsa, Mayank
Singh, Richa
Computer Vision and Pattern Recognition
Existing unlearning algorithms in text-to-image generative models often fail to preserve the knowledge of semantically related concepts when removing specific target concepts: a challenge known as adjacency. To address this, we propose FADE (Fine grained Attenuation for Diffusion Erasure), introducing adjacency aware unlearning in diffusion models. FADE comprises two components: (1) the Concept Neighborhood, which identifies an adjacency set of related concepts, and (2) Mesh Modules, employing a structured combination of Expungement, Adjacency, and Guidance loss components. These enable precise erasure of target concepts while preserving fidelity across related and unrelated concepts. Evaluated on datasets like Stanford Dogs, Oxford Flowers, CUB, I2P, Imagenette, and ImageNet1k, FADE effectively removes target concepts with minimal impact on correlated concepts, achieving atleast a 12% improvement in retention performance over state-of-the-art methods.
title Fine-Grained Erasure in Text-to-Image Diffusion-based Foundation Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2503.19783