研究人员开发了一种名为StAD的新方法，以提高扩散和流模型生成器中似然计算的速度和准确性。该技术绕过了计算概率流ODE的雅可比矩阵的需要，而是使用Langevin-Stein算子直接学习散度。StAD在各种密度估计任务上已证明其性能与Hutchinson和Hutch++等现有方法相比具有竞争力，显示出更高的方差和速度。 AI
影响 加速扩散和流模型中的似然计算，有益于贝叶斯分析和密度估计任务。
排序理由 该集群包含一篇详细介绍生成模型新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2605.16486v1 Announce Type: new Abstract: Diffusion and flow-based models are ubiquitously used for generative modelling and density estimation. They admit a deterministic probability flow ordinary differential equation (PF-ODE), analogous to continuous normalizing flows (C…
Diffusion and flow-based models are ubiquitously used for generative modelling and density estimation. They admit a deterministic probability flow ordinary differential equation (PF-ODE), analogous to continuous normalizing flows (CNFs), which describes the transport of the proba…