BriefGPT.xyz
Dec, 2023
摊销重参数化：隐状态随机微分方程的高效可扩展变分推断
Amortized Reparametrization: Efficient and Scalable Variational Inference for Latent SDEs
HTML
PDF
Kevin Course, Prasanth B. Nair
TL;DR
通过一种新的分摊策略以及线性SDEs下的期望重新参数化的方法，我们在训练时的模型评估次数减少了一个数量级，从而实现了类似于基于伴随灵敏度的方法的相似性能。
Abstract
We consider the problem of inferring
latent stochastic differential equations
(SDEs) with a time and memory cost that scales independently with the amount of data, the total length of the time series, and the stiffness of the
→