Time Series Forecasting using Recurrent Neural Networks modified by Bayesian Inference in the Learning Process

Typically, time series forecasting is done by using models based directly on the past observations from the same sequence. In these cases, when the model is learning from data, there is not an extra quantity of noiseless data available and computational resources are unlimited. In practice, it is ne...

Descripción completa

Detalles Bibliográficos
Autores Principales: Rivero C.R., Pucheta J., Otaño P., Orjuela-Cañon A.D., Patiño D., Franco L., Gorrostieta E., Puglisi J.L., Juarez G.
Formato: Objeto de conferencia (Conference Object)
Lenguaje:Inglés (English)
Publicado: Institute of Electrical and Electronics Engineers Inc. 2019
Materias:
Acceso en línea:https://repository.urosario.edu.co/handle/10336/22344
https://doi.org/10.1109/ColCACI.2019.8781984
id ir-10336-22344
recordtype dspace
spelling ir-10336-223442022-05-02T12:37:20Z Time Series Forecasting using Recurrent Neural Networks modified by Bayesian Inference in the Learning Process Rivero C.R. Pucheta J. Otaño P. Orjuela-Cañon A.D. Patiño D. Franco L. Gorrostieta E. Puglisi J.L. Juarez G. Bayesian networks Forecasting Inference engines Learning systems Time series Bayesian Bayesian neural networks Computational resources Kullback Leibler divergence Kullback-Leibler information Marginal likelihood Subjective uncertainty Time series forecasting Recurrent neural networks Bayesian approximation Kullback-Leibler Divergence Recurrent neural network Time Series Forecasting Typically, time series forecasting is done by using models based directly on the past observations from the same sequence. In these cases, when the model is learning from data, there is not an extra quantity of noiseless data available and computational resources are unlimited. In practice, it is necessary to deal with finite noisy datasets, which lead to uncertainty about what so appropriate the model is. For this, the employment of models based on Bayesian inference are preferable. Then, probabilities are treated as a way to represent the subjective uncertainty from rational agent, performing an approximated inference by maximizing a lower bound on the marginal likelihood. A modified algorithm using long-short memory recurrent neural networks for time series forecasting was presented. This new approach was chosen in order to be as close as possible to the original series in the sense of minimizing the associated Kullback-Leibler Information Criterion. A simulation study was conducted to evaluate and illustrate results, comparing this approach with Bayesian neural-networks-based algorithms for artificial chaotic time-series. © 2019 IEEE. 2019 2020-05-25T23:56:10Z info:eu-repo/semantics/conferenceObject info:eu-repo/semantics/publishedVersion https://repository.urosario.edu.co/handle/10336/22344 https://doi.org/10.1109/ColCACI.2019.8781984 eng info:eu-repo/semantics/openAccess application/pdf Institute of Electrical and Electronics Engineers Inc. instname:Universidad del Rosario
institution EdocUR - Universidad del Rosario
collection DSpace
language Inglés (English)
topic Bayesian networks
Forecasting
Inference engines
Learning systems
Time series
Bayesian
Bayesian neural networks
Computational resources
Kullback Leibler divergence
Kullback-Leibler information
Marginal likelihood
Subjective uncertainty
Time series forecasting
Recurrent neural networks
Bayesian approximation
Kullback-Leibler Divergence
Recurrent neural network
Time Series Forecasting
spellingShingle Bayesian networks
Forecasting
Inference engines
Learning systems
Time series
Bayesian
Bayesian neural networks
Computational resources
Kullback Leibler divergence
Kullback-Leibler information
Marginal likelihood
Subjective uncertainty
Time series forecasting
Recurrent neural networks
Bayesian approximation
Kullback-Leibler Divergence
Recurrent neural network
Time Series Forecasting
Rivero C.R.
Pucheta J.
Otaño P.
Orjuela-Cañon A.D.
Patiño D.
Franco L.
Gorrostieta E.
Puglisi J.L.
Juarez G.
Time Series Forecasting using Recurrent Neural Networks modified by Bayesian Inference in the Learning Process
description Typically, time series forecasting is done by using models based directly on the past observations from the same sequence. In these cases, when the model is learning from data, there is not an extra quantity of noiseless data available and computational resources are unlimited. In practice, it is necessary to deal with finite noisy datasets, which lead to uncertainty about what so appropriate the model is. For this, the employment of models based on Bayesian inference are preferable. Then, probabilities are treated as a way to represent the subjective uncertainty from rational agent, performing an approximated inference by maximizing a lower bound on the marginal likelihood. A modified algorithm using long-short memory recurrent neural networks for time series forecasting was presented. This new approach was chosen in order to be as close as possible to the original series in the sense of minimizing the associated Kullback-Leibler Information Criterion. A simulation study was conducted to evaluate and illustrate results, comparing this approach with Bayesian neural-networks-based algorithms for artificial chaotic time-series. © 2019 IEEE.
format Objeto de conferencia (Conference Object)
author Rivero C.R.
Pucheta J.
Otaño P.
Orjuela-Cañon A.D.
Patiño D.
Franco L.
Gorrostieta E.
Puglisi J.L.
Juarez G.
author_facet Rivero C.R.
Pucheta J.
Otaño P.
Orjuela-Cañon A.D.
Patiño D.
Franco L.
Gorrostieta E.
Puglisi J.L.
Juarez G.
author_sort Rivero C.R.
title Time Series Forecasting using Recurrent Neural Networks modified by Bayesian Inference in the Learning Process
title_short Time Series Forecasting using Recurrent Neural Networks modified by Bayesian Inference in the Learning Process
title_full Time Series Forecasting using Recurrent Neural Networks modified by Bayesian Inference in the Learning Process
title_fullStr Time Series Forecasting using Recurrent Neural Networks modified by Bayesian Inference in the Learning Process
title_full_unstemmed Time Series Forecasting using Recurrent Neural Networks modified by Bayesian Inference in the Learning Process
title_sort time series forecasting using recurrent neural networks modified by bayesian inference in the learning process
publisher Institute of Electrical and Electronics Engineers Inc.
publishDate 2019
url https://repository.urosario.edu.co/handle/10336/22344
https://doi.org/10.1109/ColCACI.2019.8781984
_version_ 1740172212652998656
score 12,131701