arXiv cs.AI· Bata Vasic, Bane Vasic·· 10 天前AI 评分22
能量条件噪声调度与白化：面向频谱扩散模型的方法
Energy-Conditioned Noise Schedule and Whitening for Spectral Diffusion
AI 导读
该研究为变换域扩散模型提出能量自适应噪声调度与白化策略，将全局频谱白化与能量条件噪声分配结合，按各变换系数的能量及图像相关的能量路径共同调制注入噪声。该方法保持高斯转移与闭式边缘分布，兼容标准 DDPM 与 DDIM 流程且无需改动扩散架构。在 CIFAR-10 上，其将紧凑型 DCTdiff U-Net 变体的 FID 从 142.48 降至 100.45。
正文 · 原文
Abstract:This paper introduces an energy-adaptive noise scheduling and whitening strategy for transform-domain diffusion models. Existing spectral diffusion methods account for the non-uniform statistics of transform coefficients through coefficient scaling, normalization, or frequency prioritization, while the forward diffusion noise schedule remains largely independent of the underlying spectral-energy distribution. We investigate whether the temporal evolution of the forward diffusion process should also follow the spectral organization of natural images. The proposed formulation combines global spectral whitening with energy-conditioned noise allocation that jointly modulates the injected noise according to the energy of individual transform coefficients and an image-dependent energy path over diffusion time. The resulting forward process preserves Gaussian transitions with closed-form marginals and remains compatible with standard DDPM and DDIM procedures without modifying the diffusion architecture. Experiments on CIFAR-10 demonstrate the contribution of the proposed energy-conditioned noise schedule and spectral whitening, reducing Fréchet Inception Distance from 142.48 for a compact DCTdiff U-Net variant to 100.45.
| Comments: | 7 pages, 5 figures, 2 tables, Manuscript submitted for publication in Elsevier Pattern Recognition Letters |
| Subjects: | Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Emerging Technologies (cs.ET) |
| Cite as: | arXiv:2610.07206 [cs.AI] |
| (or arXiv:2610.07206v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07206 arXiv-issued DOI via DataCite (pending registration) |
|
| Related DOI: | https://doi.org/10.2139/ssrn.7557215
DOI(s) linking to related resources |
Submission history
From: Bata Vasic Prof. [view email]
[v1]
Mon, 5 Oct 2026 18:19:45 UTC (869 KB)
来源：arXiv cs.AI · arxiv.org