arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2303.01748cs.LGcs.CVstat.ML
收藏

扩散生成模型的完整配方

A Complete Recipe for Diffusion Generative Models

University of California, Irvine(加州大学欧文分校)

Kushagra Pandey, Stephan Mandt

AI总结:

提出扩散生成模型前向过程的完整配方,并据此引入相空间朗之万扩散(PSLD),在增广空间进行分数建模,实现更优样本质量与速度-质量权衡,FID达2.10。

摘要:

基于分数的生成模型(SGMs)已在各种任务中展现出卓越的合成效果。然而,当前前向扩散过程的设计空间在很大程度上尚未被开发,且通常依赖于物理启发或简化假设。利用可扩展贝叶斯后验采样器开发中的见解,我们提出了一个用于构建SGMs前向过程的完整配方,确保收敛到期望的目标分布。我们的方法揭示,若干现有的SGMs可被视为我们框架的具体表现形式。基于此方法,我们引入了相空间朗之万扩散(PSLD),它依赖于在由类似于物理相空间的辅助变量增强的增广空间中进行基于分数的建模。实验结果表明,与多种竞争方法相比,PSLD在已建立的图像合成基准上具有更优的样本质量和改进的速度-质量权衡。值得注意的是,PSLD达到了与最先进的SGMs相当的样本质量(在无条件CIFAR-10生成上FID为2.10)。最后,我们展示了PSLD在使用预训练分数网络进行条件合成中的适用性,为未来的进展提供了一个有吸引力的SGM骨干替代方案。代码和模型检查点可在\url{https://github.com/mandt-lab/PSLD}访问。

英文摘要:

Score-based Generative Models (SGMs) have demonstrated exceptional synthesis outcomes across various tasks. However, the current design landscape of the forward diffusion process remains largely untapped and often relies on physical heuristics or simplifying assumptions. Utilizing insights from the development of scalable Bayesian posterior samplers, we present a complete recipe for formulating forward processes in SGMs, ensuring convergence to the desired target distribution. Our approach reveals that several existing SGMs can be seen as specific manifestations of our framework. Building upon this method, we introduce Phase Space Langevin Diffusion (PSLD), which relies on score-based modeling within an augmented space enriched by auxiliary variables akin to physical phase space. Empirical results exhibit the superior sample quality and improved speed-quality trade-off of PSLD compared to various competing approaches on established image synthesis benchmarks. Remarkably, PSLD achieves sample quality akin to state-of-the-art SGMs (FID: 2.10 for unconditional CIFAR-10 generation). Lastly, we demonstrate the applicability of PSLD in conditional synthesis using pre-trained score networks, offering an appealing alternative as an SGM backbone for future advancements. Code and model checkpoints can be accessed at \url{https://github.com/mandt-lab/PSLD}.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑