An Optimization Framework for Denoising Generation
Noise-unconditioned denoising learns a single score, so generation can be viewed as gradient ascent on a fixed log-density landscape. This optimization interpretation leaves two questions: do trajectories approach meaningful data, and do different initializations retain diverse outcomes? We establish both phenomena for empirical and population multi-scale Gaussian-mixture objectives. Under explicit posterior and landscape conditions, ascent converges almost surely within $O(mσ_{\min}^2/τ)$ of an $m$-dimensional reference manifold with reach at least $τ$. We derive a source--sink identity for the same fixed field and use it to prove finite-resolution regional coverage. In an affine construction, we further identify a positive-bandwidth regime where even empirical-score optimization can avoid training-sample memorization while preserving diversity at the training resolution. The same fixed objective naturally supports penalty-guided controllable generation. Experiments support our predicted mechanisms, show improved robustness under large-step guidance, and demonstrate downstream gains consistent with finite-data generative generalization.
Authors
- Stephen James Wright (ORCID: https://orcid.org/0000-0001-6252-8271)
- Xinyu Zhou
- Jiawei Zhang
Publication Details
- Journal
- arXiv (Cornell University)
- Published
- 2026-11-17
- DOI
- https://doi.org/10.48550/arxiv.2611.22202
- Primary Topic
- Image and Signal Denoising Methods
- Type
- preprint