Adaptive Moments are Surprisingly Effective for Plug-and-Play Diffusion Sampling
This addresses noise issues in diffusion models for image generation and restoration, though it appears incremental as it adapts an existing optimization technique.
The paper tackles the problem of noise in guided diffusion sampling by using adaptive moment estimation to stabilize likelihood scores, achieving state-of-the-art results on image restoration and class-conditional generation tasks with concrete performance improvements.
Guided diffusion sampling relies on approximating often intractable likelihood scores, which introduces significant noise into the sampling dynamics. We propose using adaptive moment estimation to stabilize these noisy likelihood scores during sampling. Despite its simplicity, our approach achieves state-of-the-art results on image restoration and class-conditional generation tasks, outperforming more complicated methods, which are often computationally more expensive. We provide empirical analysis of our method on both synthetic and real data, demonstrating that mitigating gradient noise through adaptive moments offers an effective way to improve alignment.