BriefGPT.xyz
Apr, 2022
SIReN-VAE: 基于流和分摊推断的贝叶斯网络优化
SIReN-VAE: Leveraging Flows and Amortized Inference for Bayesian Networks
HTML
PDF
Jacobie Mouton, Steve Kroon
TL;DR
本文探讨了如何将贝叶斯网络中的任意依赖结构引入变分自编码器 (VAEs) 中,通过引入图形残差流来实现。我们在多个合成数据集上比较了模型的性能,并展示其在数据稀缺情况下的潜力。
Abstract
Initial work on
variational autoencoders
assumed independent
latent variables
with simple distributions. Subsequent work has explored incorporating more complex distributions and
→