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 2502.04670cs.LGcs.AI
收藏

CCS:通过初始噪声扰动实现扩散模型的可控与约束采样

CCS: Controllable and Constrained Sampling with Diffusion Models via Initial Noise Perturbation

University of Michigan(密歇根大学);Rutgers University(罗格斯大学)

Bowen Song, Zecheng Zhang, Zhaoxu Luo, Jason Hu, Wei Yuan, Jing Jia, Zhengxu Tang, Guanyang Wang, Liyue Shen

AI总结:

本文发现扩散模型输出变化与初始噪声扰动尺度呈线性关系,并据此提出CCS可控约束采样方法,在保持样本质量与多样性的同时实现更精确的统计特性控制。

摘要:

扩散模型已成为生成任务中的强大工具,能够在多种领域产生高质量输出。然而,生成数据如何响应扩散模型中的初始噪声扰动仍未得到充分探索,这阻碍了对采样过程可控性的理解。在本工作中,我们首先观察到一个有趣的现象:在扩散ODE采样过程中,生成输出的变化与初始噪声扰动的尺度之间呈高度线性关系。随后,我们提供了理论和实证研究来证明这一输入-输出(噪声-生成数据)关系的线性特性。受这些新见解的启发,我们提出了一种新颖的可控与约束采样方法(CCS)以及一种新的控制器算法,用于扩散模型在保持良好样本质量的同时,采样具有期望统计特性的数据。我们进行了大量实验,将所提出的采样方法与其他方法在采样可控性和采样数据质量方面进行比较。结果表明,我们的CCS方法在保持优异样本质量和多样性的同时,实现了更精确的可控采样。

英文摘要:

Diffusion models have emerged as powerful tools for generative tasks, producing high-quality outputs across diverse domains. However, how the generated data responds to the initial noise perturbation in diffusion models remains under-explored, which hinders understanding the controllability of the sampling process. In this work, we first observe an interesting phenomenon: the relationship between the change of generation outputs and the scale of initial noise perturbation is highly linear through the diffusion ODE sampling. Then we provide both theoretical and empirical study to justify this linearity property of this input-output (noise-generation data) relationship. Inspired by these new insights, we propose a novel Controllable and Constrained Sampling method (CCS) together with a new controller algorithm for diffusion models to sample with desired statistical properties while preserving good sample quality. We perform extensive experiments to compare our proposed sampling approach with other methods on both sampling controllability and sampled data quality. Results show that our CCS method achieves more precisely controlled sampling while maintaining superior sample quality and diversity.

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。

↑