Tim R. Davidson, Luca Falorsi, Nicola De Cao, Thomas Kipf, Jakub M. Tomczak
TL;DR本研究提出了利用 von Mises-Fisher 分布替代 Gaussian 分布模型的 VAE,能更好地捕捉数据中的超球面潜在结构,且在低维数据类型中优于传统 VAE。
Abstract
The variational auto-encoder (VAE) is one of the most used unsupervised
machine learning models. But although the default choice of a Gaussian
distribution for both the prior and posterior represents a mathematically
convenient distribution often leading to competitive results, we show