AIBullisharXiv – CS AI · 5h ago6/10
🧠
SAEmnesia: Erasing Concepts in Diffusion Models with Supervised Sparse Autoencoders
Researchers introduce SAEmnesia, a supervised sparse autoencoder framework that enables efficient concept unlearning in diffusion models by binding concepts to individual neurons. The method reduces computational overhead by 96.67% compared to existing approaches and achieves 9.22% improvement on benchmark tests, with demonstrated robustness against adversarial attacks.