View a PDF of the paper titled Variational Trajectory Optimization of Anisotropic Diffusion Schedules, by Pengxi Liu and 2 other authors
Abstract:We introduce a variational framework for diffusion models with anisotropic noise schedules parameterized by a matrix-valued path $M_t(\theta)$ that allocates noise across subspaces. Central to our framework is a trajectory-level objective that jointly trains the score network and \emph{learns} $M_t(\theta)$, which encompasses general parameterization classes of matrix-valued noise schedules. We further derive an estimator for $\partial_\theta \nabla \log p_t$ that enables efficient optimization of the $M_t(\theta)$ schedule. For inference, we develop an efficiently-implementable reverse-ODE solver that is an anisotropic generalization of the second-order Heun discretization algorithm. Across CIFAR-10, AFHQv2, FFHQ, and ImageNet-64, our method consistently improves FID with largest gains at low NFE. Code is available at this https URL.
Submission history
From: Pengxi Liu [view email]
[v1]
Mon, 23 Feb 2026 04:56:41 UTC (8,597 KB)
[v2]
Sun, 15 Nov 2026 02:51:15 UTC (8,599 KB)
![[2602.19512] Variational Trajectory Optimization of Anisotropic Diffusion Schedules [2602.19512] Variational Trajectory Optimization of Anisotropic Diffusion Schedules](https://techwavedispatch.com/wp-content/uploads/2026/11/260219512-Variational-Trajectory-Optimization-of-Anisotropic-Diffusion-Schedules-1024x597.png)