Holistic Unlearning Benchmark: A Multi-Faceted Evaluation for Text-to-Image Diffusion Model Unlearning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Moon, Saemi, Lee, Minjong, Park, Sangdon, Kim, Dongwoo
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911256421597184
author Moon, Saemi
Lee, Minjong
Park, Sangdon
Kim, Dongwoo
author_facet Moon, Saemi
Lee, Minjong
Park, Sangdon
Kim, Dongwoo
contents As text-to-image diffusion models gain widespread commercial applications, there are increasing concerns about unethical or harmful use, including the unauthorized generation of copyrighted or sensitive content. Concept unlearning has emerged as a promising solution to these challenges by removing undesired and harmful information from the pre-trained model. However, the previous evaluations primarily focus on whether target concepts are removed while preserving image quality, neglecting the broader impacts such as unintended side effects. In this work, we propose Holistic Unlearning Benchmark (HUB), a comprehensive framework for evaluating unlearning methods across six key dimensions: faithfulness, alignment, pinpoint-ness, multilingual robustness, attack robustness, and efficiency. Our benchmark covers 33 target concepts, including 16,000 prompts per concept, spanning four categories: Celebrity, Style, Intellectual Property, and NSFW. Our investigation reveals that no single method excels across all evaluation criteria. By releasing our evaluation code and dataset, we hope to inspire further research in this area, leading to more reliable and effective unlearning methods.
format Preprint
id arxiv_https___arxiv_org_abs_2410_05664
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Holistic Unlearning Benchmark: A Multi-Faceted Evaluation for Text-to-Image Diffusion Model Unlearning
Moon, Saemi
Lee, Minjong
Park, Sangdon
Kim, Dongwoo
Computer Vision and Pattern Recognition
Machine Learning
As text-to-image diffusion models gain widespread commercial applications, there are increasing concerns about unethical or harmful use, including the unauthorized generation of copyrighted or sensitive content. Concept unlearning has emerged as a promising solution to these challenges by removing undesired and harmful information from the pre-trained model. However, the previous evaluations primarily focus on whether target concepts are removed while preserving image quality, neglecting the broader impacts such as unintended side effects. In this work, we propose Holistic Unlearning Benchmark (HUB), a comprehensive framework for evaluating unlearning methods across six key dimensions: faithfulness, alignment, pinpoint-ness, multilingual robustness, attack robustness, and efficiency. Our benchmark covers 33 target concepts, including 16,000 prompts per concept, spanning four categories: Celebrity, Style, Intellectual Property, and NSFW. Our investigation reveals that no single method excels across all evaluation criteria. By releasing our evaluation code and dataset, we hope to inspire further research in this area, leading to more reliable and effective unlearning methods.
title Holistic Unlearning Benchmark: A Multi-Faceted Evaluation for Text-to-Image Diffusion Model Unlearning
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2410.05664