Information Theoretic-Learning Auto-Encoder

Eder Santana; Matthew Emigh; Jose C Principe

Information Theoretic-Learning Auto-Encoder

Machine Learning 2016-03-23 v1

Authors: Eder Santana , Matthew Emigh , Jose C Principe

Abstract

We propose Information Theoretic-Learning (ITL) divergence measures for variational regularization of neural networks. We also explore ITL-regularized autoencoders as an alternative to variational autoencoding bayes, adversarial autoencoders and generative adversarial networks for randomly generating sample data without explicitly defining a partition function. This paper also formalizes, generative moment matching networks under the ITL framework.

Keywords

extreme learning machine deep learning variational autoencoder

Cite

@article{arxiv.1603.06653,
  title  = {Information Theoretic-Learning Auto-Encoder},
  author = {Eder Santana and Matthew Emigh and Jose C Principe},
  journal= {arXiv preprint arXiv:1603.06653},
  year   = {2016}
}

Comments

8 pages, 4 figures

Related papers

View all related →