Abstract:Diffusion models generate structure by progressively transforming noise into data, yet the mechanisms underlying this transition remain poorly understood. In this work, we show that pattern formation in trained diffusion models can be explained as an out-of-equilibrium phase transition driven by instabilities in the denoising dynamics. We develop a theoretical framework linking data symmetries and architectural constraints, such as locality and translation equivariance, to the emergence of collective spatial modes. In this view, structure arises when low-frequency modes become unstable, triggering a rapid growth of spatial correlations that organizes noise into coherent patterns. We validate this theory through a combination of analytical models and experiments. In a controlled patch-based model, we observe a sharp increase in correlation length and a simultaneous softening of low-frequency modes at a well-defined critical time, accurately predicted by theory. Similar signatures are found in trained convolutional diffusion models on Fashion-MNIST and in large-scale ImageNet models, where pattern formation coincides with a peak in estimated correlation length and a pronounced weakening of spatial modes. Finally, intervention experiments show that applying guidance precisely at this critical stage significantly improves class alignment compared to applying it at random times, demonstrating that this regime is not only descriptive but functionally important.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2603.20092 [cs.LG] |
| (or arXiv:2603.20092v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2603.20092 arXiv-issued DOI via DataCite |
From: Luca Ambrogioni [view email]
[v1]
Fri, 20 Mar 2026 16:19:10 UTC (4,329 KB)
[v2]
Fri, 3 Apr 2026 18:46:43 UTC (4,873 KB)
[v3]
Tue, 21 Apr 2026 10:39:19 UTC (4,838 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。