Auxiliary Deep Generative Models
Lars Maaløe, Casper Kaae Sønderby, Søren Kaae Sønderby, Ole Winther
Abstract
Deep generative models parameterized by neural networks have recently achieved state-of-the-art performance in unsupervised and semi-supervised learning. We extend deep generative models with auxiliary variables which improves the variational approximation. The auxiliary variables leave the generative model unchanged but make the variational distribution more expressive. Inspired by the structure of the auxiliary variable we also propose a model with two stochastic layers and skip connections. Our findings suggest that more expressive and properly specified deep generative models converge faster with better results. We show state-of-the-art performance within semi-supervised learning on MNIST, SVHN and NORB datasets.
BibTeX
@InProceedings{pmlr-v48-maaloe16,
title = {Auxiliary Deep Generative Models},
author = {Maaløe, Lars and Sønderby, Casper Kaae and Sønderby, Søren Kaae and Winther, Ole},
booktitle = {Proceedings of The 33rd International Conference on Machine Learning},
pages = {1445--1453},
year = {2016},
editor = {Balcan, Maria Florina and Weinberger, Kilian Q.},
volume = {48},
series = {Proceedings of Machine Learning Research},
address = {New York, New York, USA},
month = {20--22 Jun},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v48/maaloe16.pdf},
url = {https://proceedings.mlr.press/v48/maaloe16.html},
abstract = {Deep generative models parameterized by neural networks have recently achieved state-of-the-art performance in unsupervised and semi-supervised learning. We extend deep generative models with auxiliary variables which improves the variational approximation. The auxiliary variables leave the generative model unchanged but make the variational distribution more expressive. Inspired by the structure of the auxiliary variable we also propose a model with two stochastic layers and skip connections. Our findings suggest that more expressive and properly specified deep generative models converge faster with better results. We show state-of-the-art performance within semi-supervised learning on MNIST, SVHN and NORB datasets.}
}