arXiv：cs.AI· Gihoon Kim, Taesup Kim·· 4 小时前AI 评分34
用频谱对齐校正引导扩散轨迹
Correcting Guided Diffusion Trajectories with Spectral Alignment
AI 导读
研究者提出 Spectral Correction Guidance，一种免训练方法，通过在采样过程中校正中间状态频谱相对解析参考谱的偏差来改进引导扩散采样。该方法可跨扩散骨干网络与条件生成任务使用，无需修改底层模型，在文生图任务上偏好类指标稳定优于基线引导方法，在 ImageNet 上生成质量优于 CFG，且在多种引导尺度和更少去噪步数下依然有效。
正文
Abstract:The practical success of conditional image generation hinges on fine-grained differences in condition alignment and visual fidelity. Classifier-free guidance (CFG) is central to this success, but its lack of an explicit criterion makes it difficult to assess whether the guided trajectory is progressing as intended. To address this gap, we show that spectral alignment provides a principled criterion for understanding guidance behavior and improving guided diffusion sampling through adaptive correction. Our analysis identifies the spectra of intermediate states as an indicator of consistency with the expected spectral evolution of the forward process. Based on this observation, we introduce Spectral Correction Guidance, a method that corrects deviations from an analytic reference spectrum during sampling. The proposed method is training-free and applicable across diffusion backbones and conditional generation tasks without modifying the underlying model. Experiments demonstrate consistent gains in preference-based metrics over baseline guidance methods in text-to-image generation and improved generation quality over CFG on ImageNet. These improvements persist across a range of guidance scales and with fewer denoising steps. Our analyses and ablations provide insight into guidance behavior and how the proposed method affects generation quality.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.02753 [cs.CV] |
| (or arXiv:2610.02753v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02753 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Gihoon Kim [view email]
[v1]
Fri, 2 Oct 2026 03:30:02 UTC (46,480 KB)
来源：arXiv：cs.AI · arxiv.org