解耦总方差与信噪比可改进扩散模型

View PDF HTML (experimental)

Abstract:Diffusion models have been widely adopted as a powerful generative framework across numerous scientific domains, yet their inherently iterative sampling process remains a significant computational bottleneck. While this is typically mitigated by reducing the number of reverse diffusion steps, sample quality becomes highly sensitive to the chosen noise schedule, i.e. the manner in which noise is introduced and the signal is reduced at each step. Although prior work has improved on the original variance-preserving and variance-exploding schedules, these approaches passively adjust the total variance, without direct independent control over it. In this work, we propose a novel total-variance/signal-to-noise-ratio disentangled (TV/SNR) framework, where TV and SNR can be controlled independently. Our approach reveals that schedules with exponentially exploding TV can be consistently improved by adopting a constant TV schedule while preserving the underlying SNR schedule. Furthermore, we demonstrate that generalizing the SNR schedule of optimal transport flow matching significantly enhances generation efficiency. Crucially, these gains remain consistent across various reverse diffusion solvers and reduce the search space to two independent and interpretable parameters, enabling fast and robust sampling for complex scientific applications, such as molecular structure generation.

Subjects:	Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as:	arXiv:2502.08598 [cs.LG]
	(or arXiv:2502.08598v3 [cs.LG] for this version)
	https://doi.org/10.48550/arXiv.2502.08598 arXiv-issued DOI via DataCite

Submission history

From: Khaled Kahouli [view email]
[v1] Wed, 12 Feb 2025 17:35:43 UTC (13,426 KB)
[v2] Mon, 2 Jun 2025 10:48:00 UTC (17,580 KB)
[v3] Fri, 9 Oct 2026 14:19:51 UTC (20,835 KB)