BriefGPT.xyz
Jun, 2019
热力学变分目标
The Thermodynamic Variational Objective
HTML
PDF
Vaden Masrani, Tuan Anh Le, Frank Wood
TL;DR
介绍使用热力学变分目标（TVO）来学习连续和离散的深层生成模型；提供了一个计算上高效的梯度估计器；并使用TVo来学习深层生成模型和推理网络，取得了最前沿的模型和推理网络学习效果。
Abstract
We introduce the
thermodynamic variational objective
(TVO) for learning in both continuous and discrete
deep generative models
. The TVO arises from a key connection between
→