A Survey on Generative Model Unlearning: Fundamentals, Taxonomy, Evaluation, and Future Direction
read the original abstract
With the rapid advancement of generative models, associated privacy concerns have attracted growing attention. To address this, researchers have begun adapting machine unlearning techniques from traditional classification models to generative settings. Although notable progress has been made in this area, a unified framework for systematically organizing and integrating existing work is still lacking. The substantial differences among current studies in terms of unlearning objectives and evaluation protocols hinder the objective and fair comparison of various approaches. While some studies focus on specific types of generative models, they often overlook the commonalities and systematic characteristics inherent in Generative Model Unlearning (GenMU). To bridge this gap, we provide a comprehensive review of current research on GenMU and propose a unified analytical framework for categorizing unlearning objectives, methodological strategies, and evaluation metrics. In addition, we explore the connections between GenMU and related techniques, including model editing, reinforcement learning from human feedback, and controllable generation. We further highlight the potential practical value of unlearning techniques in real-world applications. Finally, we identify key challenges and outline future research directions aimed at laying a solid foundation for further advancements in this field. We consistently maintain the related open-source materials at https://github.com/caxLee/Generative-model-unlearning-survey.
This paper has not been read by Pith yet.
Forward citations
Cited by 2 Pith papers
-
Look But Don't Touch with Sparse Autoencoders for Unlearning in Diffusion Models
SAEs detect concepts well in diffusion models but fail as direct intervention points for unlearning; a detection-guided patch replacement method yields significantly cleaner erasure results.
-
Look But Don't Touch with Sparse Autoencoders for Unlearning in Diffusion Models
SAEs detect semantic concepts in diffusion models but direct latent intervention induces out-of-distribution activations and artifacts, while detection-based patch replacement enables cleaner object erasure.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.