BriefGPT.xyz
Jan, 2018
变分自编码器中的推理次优性
Inference Suboptimality in Variational Autoencoders
HTML
PDF
Chris Cremer, Xuechen Li, David Duvenaud
TL;DR
研究了变分自编码器中的近似推理，发现推理网络的不完美会导致与真实后验的偏差，并且发现用于提高近似值的参数对于推理的泛化起到重要作用。
Abstract
amortized inference
has led to efficient
approximate inference
for large datasets. The quality of posterior inference is largely determined by two factors: a) the ability of the variational distribution to model
→