Poisson Flow Consistency Training
Anthony Zhang, Mahmut Gokmen, Dennis Hein, Rongjun Ge, Wenjun Xia, Ge Wang, Jin Chen
TL;DR
This work addresses the limitation of Poisson Flow Consistency Models requiring a pre-trained PFGM++ by introducing Poisson Flow Consistency Training (PFCT), a method to train PFCMs in isolation. PFCT combines a perturbation kernel with a sinusoidal discretization schedule and a Beta noise distribution, enhancing stability and sample quality, and demonstrates effective LDCT denoising on the Low-Dose CT Image Denoising and Projection Dataset. Although PFCT does not surpass all diffusion-based or distillation methods across every metric, it achieves clear improvements over the low-dose baseline in LPIPS, SSIM, and PSNR, establishing a practical, flexible training pathway for PFCMs in various data modalities. The approach broadens the applicability of PFCMs in medical imaging and beyond, paving the way for future cross-domain applications and targeted ablations to optimize performance.
Abstract
The Poisson Flow Consistency Model (PFCM) is a consistency-style model based on the robust Poisson Flow Generative Model++ (PFGM++) which has achieved success in unconditional image generation and CT image denoising. Yet the PFCM can only be trained in distillation which limits the potential of the PFCM in many data modalities. The objective of this research was to create a method to train the PFCM in isolation called Poisson Flow Consistency Training (PFCT). The perturbation kernel was leveraged to remove the pretrained PFGM++, and the sinusoidal discretization schedule and Beta noise distribution were introduced in order to facilitate adaptability and improve sample quality. The model was applied to the task of low dose computed tomography image denoising and improved the low dose image in terms of LPIPS and SSIM. It also displayed similar denoising effectiveness as models like the Consistency Model. PFCT is established as a valid method of training the PFCM from its effectiveness in denoising CT images, showing potential with competitive results to other generative models. Further study is needed in the precise optimization of PFCT and in its applicability to other generative modeling tasks. The framework of PFCT creates more flexibility for the ways in which a PFCM can be created and can be applied to the field of generative modeling.