The Variational InfoMax AutoEncoder
Machine Learning
2020-11-10 v2 Machine Learning
Abstract
The Variational AutoEncoder (VAE) learns simultaneously an inference and a generative model, but only one of these models can be learned at optimum, this behaviour is associated to the ELBO learning objective, that is optimised by a non-informative generator. In order to solve such an issue, we provide a learning objective, learning a maximal informative generator while maintaining bounded the network capacity: the Variational InfoMax (VIM). The contribution of the VIM derivation is twofold: an objective learning both an optimal inference and generative model and the explicit definition of the network capacity, an estimation of the network robustness.
Keywords
Cite
@article{arxiv.1905.10549,
title = {The Variational InfoMax AutoEncoder},
author = {Vincenzo Crescimanna and Bruce Graham},
journal= {arXiv preprint arXiv:1905.10549},
year = {2020}
}