Cargando…

Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder

We present a coupled variational autoencoder (VAE) method, which improves the accuracy and robustness of the model representation of handwritten numeral images. The improvement is measured in both increasing the likelihood of the reconstructed images and in reducing divergence between the posterior...

Descripción completa

Detalles Bibliográficos
Autores principales: Cao, Shichen, Li, Jingjing, Nelson, Kenric P., Kon, Mark A.
Formato: Online Artículo Texto
Lenguaje:English
Publicado: MDPI 2022
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8947750/
https://www.ncbi.nlm.nih.gov/pubmed/35327933
http://dx.doi.org/10.3390/e24030423
_version_ 1784674514006704128
author Cao, Shichen
Li, Jingjing
Nelson, Kenric P.
Kon, Mark A.
author_facet Cao, Shichen
Li, Jingjing
Nelson, Kenric P.
Kon, Mark A.
author_sort Cao, Shichen
collection PubMed
description We present a coupled variational autoencoder (VAE) method, which improves the accuracy and robustness of the model representation of handwritten numeral images. The improvement is measured in both increasing the likelihood of the reconstructed images and in reducing divergence between the posterior and a prior latent distribution. The new method weighs outlier samples with a higher penalty by generalizing the original evidence lower bound function using a coupled entropy function based on the principles of nonlinear statistical coupling. We evaluated the performance of the coupled VAE model using the Modified National Institute of Standards and Technology (MNIST) dataset and its corrupted modification C-MNIST. Histograms of the likelihood that the reconstruction matches the original image show that the coupled VAE improves the reconstruction and this improvement is more substantial when seeded with corrupted images. All five corruptions evaluated showed improvement. For instance, with the Gaussian corruption seed the accuracy improves by [Formula: see text] (from [Formula: see text] to [Formula: see text]) and robustness improves by [Formula: see text] (from [Formula: see text] to [Formula: see text]). Furthermore, the divergence between the posterior and prior distribution of the latent distribution is reduced. Thus, in contrast to the [Formula: see text]-VAE design, the coupled VAE algorithm improves model representation, rather than trading off the performance of the reconstruction and latent distribution divergence.
format Online
Article
Text
id pubmed-8947750
institution National Center for Biotechnology Information
language English
publishDate 2022
publisher MDPI
record_format MEDLINE/PubMed
spelling pubmed-89477502022-03-25 Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder Cao, Shichen Li, Jingjing Nelson, Kenric P. Kon, Mark A. Entropy (Basel) Article We present a coupled variational autoencoder (VAE) method, which improves the accuracy and robustness of the model representation of handwritten numeral images. The improvement is measured in both increasing the likelihood of the reconstructed images and in reducing divergence between the posterior and a prior latent distribution. The new method weighs outlier samples with a higher penalty by generalizing the original evidence lower bound function using a coupled entropy function based on the principles of nonlinear statistical coupling. We evaluated the performance of the coupled VAE model using the Modified National Institute of Standards and Technology (MNIST) dataset and its corrupted modification C-MNIST. Histograms of the likelihood that the reconstruction matches the original image show that the coupled VAE improves the reconstruction and this improvement is more substantial when seeded with corrupted images. All five corruptions evaluated showed improvement. For instance, with the Gaussian corruption seed the accuracy improves by [Formula: see text] (from [Formula: see text] to [Formula: see text]) and robustness improves by [Formula: see text] (from [Formula: see text] to [Formula: see text]). Furthermore, the divergence between the posterior and prior distribution of the latent distribution is reduced. Thus, in contrast to the [Formula: see text]-VAE design, the coupled VAE algorithm improves model representation, rather than trading off the performance of the reconstruction and latent distribution divergence. MDPI 2022-03-18 /pmc/articles/PMC8947750/ /pubmed/35327933 http://dx.doi.org/10.3390/e24030423 Text en © 2022 by the authors. https://creativecommons.org/licenses/by/4.0/Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
spellingShingle Article
Cao, Shichen
Li, Jingjing
Nelson, Kenric P.
Kon, Mark A.
Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder
title Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder
title_full Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder
title_fullStr Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder
title_full_unstemmed Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder
title_short Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder
title_sort coupled vae: improved accuracy and robustness of a variational autoencoder
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8947750/
https://www.ncbi.nlm.nih.gov/pubmed/35327933
http://dx.doi.org/10.3390/e24030423
work_keys_str_mv AT caoshichen coupledvaeimprovedaccuracyandrobustnessofavariationalautoencoder
AT lijingjing coupledvaeimprovedaccuracyandrobustnessofavariationalautoencoder
AT nelsonkenricp coupledvaeimprovedaccuracyandrobustnessofavariationalautoencoder
AT konmarka coupledvaeimprovedaccuracyandrobustnessofavariationalautoencoder