Cargando…

Estimating Uncertainty Intervals from Collaborating Networks

Effective decision making requires understanding the uncertainty inherent in a prediction. In regression, this uncertainty can be estimated by a variety of methods; however, many of these methods are laborious to tune, generate overconfident uncertainty intervals, or lack sharpness (give imprecise i...

Descripción completa

Detalles Bibliográficos
Autores principales: Zhou, Tianhui, Li, Yitong, Wu, Yuan, Carlson, David
Formato: Online Artículo Texto
Lenguaje:English
Publicado: 2021
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9231643/
https://www.ncbi.nlm.nih.gov/pubmed/35754923
_version_ 1784735391715164160
author Zhou, Tianhui
Li, Yitong
Wu, Yuan
Carlson, David
author_facet Zhou, Tianhui
Li, Yitong
Wu, Yuan
Carlson, David
author_sort Zhou, Tianhui
collection PubMed
description Effective decision making requires understanding the uncertainty inherent in a prediction. In regression, this uncertainty can be estimated by a variety of methods; however, many of these methods are laborious to tune, generate overconfident uncertainty intervals, or lack sharpness (give imprecise intervals). We address these challenges by proposing a novel method to capture predictive distributions in regression by defining two neural networks with two distinct loss functions. Specifically, one network approximates the cumulative distribution function, and the second network approximates its inverse. We refer to this method as Collaborating Networks (CN). Theoretical analysis demonstrates that a fixed point of the optimization is at the idealized solution, and that the method is asymptotically consistent to the ground truth distribution. Empirically, learning is straightforward and robust. We benchmark CN against several common approaches on two synthetic and six real-world datasets, including forecasting A1c values in diabetic patients from electronic health records, where uncertainty is critical. In the synthetic data, the proposed approach essentially matches ground truth. In the real-world datasets, CN improves results on many performance metrics, including log-likelihood estimates, mean absolute errors, coverage estimates, and prediction interval widths.
format Online
Article
Text
id pubmed-9231643
institution National Center for Biotechnology Information
language English
publishDate 2021
record_format MEDLINE/PubMed
spelling pubmed-92316432022-06-24 Estimating Uncertainty Intervals from Collaborating Networks Zhou, Tianhui Li, Yitong Wu, Yuan Carlson, David J Mach Learn Res Article Effective decision making requires understanding the uncertainty inherent in a prediction. In regression, this uncertainty can be estimated by a variety of methods; however, many of these methods are laborious to tune, generate overconfident uncertainty intervals, or lack sharpness (give imprecise intervals). We address these challenges by proposing a novel method to capture predictive distributions in regression by defining two neural networks with two distinct loss functions. Specifically, one network approximates the cumulative distribution function, and the second network approximates its inverse. We refer to this method as Collaborating Networks (CN). Theoretical analysis demonstrates that a fixed point of the optimization is at the idealized solution, and that the method is asymptotically consistent to the ground truth distribution. Empirically, learning is straightforward and robust. We benchmark CN against several common approaches on two synthetic and six real-world datasets, including forecasting A1c values in diabetic patients from electronic health records, where uncertainty is critical. In the synthetic data, the proposed approach essentially matches ground truth. In the real-world datasets, CN improves results on many performance metrics, including log-likelihood estimates, mean absolute errors, coverage estimates, and prediction interval widths. 2021 /pmc/articles/PMC9231643/ /pubmed/35754923 Text en https://creativecommons.org/licenses/by/4.0/License : CC-BY 4.0, see https://creativecommons.org/licenses/by/4.0/.
spellingShingle Article
Zhou, Tianhui
Li, Yitong
Wu, Yuan
Carlson, David
Estimating Uncertainty Intervals from Collaborating Networks
title Estimating Uncertainty Intervals from Collaborating Networks
title_full Estimating Uncertainty Intervals from Collaborating Networks
title_fullStr Estimating Uncertainty Intervals from Collaborating Networks
title_full_unstemmed Estimating Uncertainty Intervals from Collaborating Networks
title_short Estimating Uncertainty Intervals from Collaborating Networks
title_sort estimating uncertainty intervals from collaborating networks
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9231643/
https://www.ncbi.nlm.nih.gov/pubmed/35754923
work_keys_str_mv AT zhoutianhui estimatinguncertaintyintervalsfromcollaboratingnetworks
AT liyitong estimatinguncertaintyintervalsfromcollaboratingnetworks
AT wuyuan estimatinguncertaintyintervalsfromcollaboratingnetworks
AT carlsondavid estimatinguncertaintyintervalsfromcollaboratingnetworks