BriefGPT.xyz
Jun, 2018
非可导模型的重参数化梯度
Reparameterization Gradient for Non-differentiable Models
HTML
PDF
Wonyeol Lee, Hangyeol Yu, Hongseok Yang
TL;DR
提出了一种新的针对非可微密度模型的随机变分推断算法,通过对可微区域应用标准的重新参数化技巧、对边界区域应用流形采样,估计并得出梯度的高效率降低了方差并保持偏差的不变。
Abstract
We present a new algorithm for
stochastic variational inference
that targets at models with
non-differentiable densities
. One of the key challenges in
→