TY - GEN

T1 - Probabilistic Forecasting Using Monte Carlo Dropout Neural Networks

AU - Serpell, Cristián

AU - ARAYA ZAMORANO, IGNACIO DANIEL

AU - Valle, Carlos

AU - Allende, Héctor

N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2019.
Copyright:
Copyright 2019 Elsevier B.V., All rights reserved.

PY - 2019

Y1 - 2019

N2 - Using artificial neural networks for forecasting tasks is a popular approach that has proven to be very accurate. When used to estimate prediction intervals, a normal distribution is usually assumed as the data noise uncertainty term, as in MVE networks, while model parameters uncertainty is often ignored. Because of this, prediction intervals estimated by them are narrow in uncertain regions where train data is scarce. To tackle this problem we apply Monte Carlo dropout, which is a model uncertainty representation technique, to the network parameters of a Long Short-Term Memory MVE network, allowing us to construct better prediction intervals in probabilistic forecasting tasks. We compare our proposal with the pure MVE method in four wind speed and one consumer load real forecasting datasets, showing that our method improves results in terms of the Winkler loss in both one step ahead and multi-step ahead probabilistic forecasting.

AB - Using artificial neural networks for forecasting tasks is a popular approach that has proven to be very accurate. When used to estimate prediction intervals, a normal distribution is usually assumed as the data noise uncertainty term, as in MVE networks, while model parameters uncertainty is often ignored. Because of this, prediction intervals estimated by them are narrow in uncertain regions where train data is scarce. To tackle this problem we apply Monte Carlo dropout, which is a model uncertainty representation technique, to the network parameters of a Long Short-Term Memory MVE network, allowing us to construct better prediction intervals in probabilistic forecasting tasks. We compare our proposal with the pure MVE method in four wind speed and one consumer load real forecasting datasets, showing that our method improves results in terms of the Winkler loss in both one step ahead and multi-step ahead probabilistic forecasting.

KW - Consumer load forecasting

KW - Mean Variance Estimation

KW - Monte Carlo dropout

KW - Prediction interval

KW - Probabilistic forecasting

KW - Wind speed forecasting

UR - http://www.scopus.com/inward/record.url?scp=85075682649&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-33904-3_36

DO - 10.1007/978-3-030-33904-3_36

M3 - Conference contribution

AN - SCOPUS:85075682649

SN - 9783030339036

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 387

EP - 397

BT - Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications - 24th Iberoamerican Congress, CIARP 2019, Proceedings

A2 - Nyström, Ingela

A2 - Hernández Heredia, Yanio

A2 - Milián Núñez, Vladimir

PB - Springer

Y2 - 28 October 2019 through 31 October 2019

ER -