HM-VAEs: a Deep Generative Model for Real-valued Data with Heterogeneous Marginals

HM-VAEs: a Deep Generative Model for Real-valued Data with Heterogeneous Marginals

Abstract

In this paper, we propose a very simple but effective VAE model (HM-VAE) that can handle real-valued data with heterogeneous marginals, meaning that they have drastically distinct marginal distributions, statistical properties as well as semantics. Preliminary results show that the HM-VAE can learn distributions with heterogeneous marginal distributions, whereas vanilla VAEs fails.

Grafik Top
Authors
  • Ma, Chao
  • Tschiatschek, Sebastian
  • Li, Yingzhen
  • Turner, Richard
  • Hernandez-Lobato, Jose Miguel
  • Zhang, Cheng
Grafik Top
Shortfacts
Category
Paper in Conference Proceedings or in Workshop Proceedings (Poster)
Event Title
2nd Symposium on Advances in Approximate Bayesian Inference
Divisions
Data Mining
Subjects
Kuenstliche Intelligenz
Angewandte Informatik
Event Location
Vancouver
Event Type
Workshop
Event Dates
December 8, 2019
Series Name
Proceedings of Machine Learning Research
Publisher
PMLR
Page Range
pp. 1-8
Date
2020
Official URL
http://proceedings.mlr.press/v118/ma20a.html
Export
Grafik Top