Robust estimation of confidence interval in neural networks applied to time series

Rodrigo Salas, Romina Torres, Héctor Allende, Claudio Moraga

Resultado de la investigación: Article

2 Citas (Scopus)

Resumen

Artificial neural networks (ANN) have been widely used in regression or predictions problems and it is usually desirable that some form of confidence bound is placed on the predicted value. A number of methods have been proposed for estimating the uncertainty associated with a value predicted by a feedforward neural network (FANN), but these methods are computationally intensive or only valid under certain assumptions, which are rarely satisfied in practice. We present the theoretical results about the construction of confidence intervals in the prediction of nonlinear time series modeled by FANN, this method is based on M-estimators that are a robust learning algorithm for parameter estimation when the data set is contaminated. The confidence interval that we propose is constructed from the study of the Influence Function of the estimator. We demonstrate our technique on computer generated Time Series data.

Idioma originalEnglish
Páginas (desde-hasta)441-448
Número de páginas8
PublicaciónLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volumen2687
EstadoPublished - 2003

Huella dactilar

Feedforward neural networks
Robust Estimation
Confidence interval
Time series
Feedforward Neural Networks
Neural Networks
Confidence Intervals
Neural networks
Parameter estimation
Learning algorithms
Confidence Bounds
Nonlinear Time Series
Influence Function
M-estimator
Prediction
Robust Algorithm
Time Series Data
Uncertainty
Artificial Neural Network
Parameter Estimation

ASJC Scopus subject areas

  • Computer Science(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Theoretical Computer Science

Citar esto

@article{31c405c79ed14174b2bdcaeaa173b84f,
title = "Robust estimation of confidence interval in neural networks applied to time series",
abstract = "Artificial neural networks (ANN) have been widely used in regression or predictions problems and it is usually desirable that some form of confidence bound is placed on the predicted value. A number of methods have been proposed for estimating the uncertainty associated with a value predicted by a feedforward neural network (FANN), but these methods are computationally intensive or only valid under certain assumptions, which are rarely satisfied in practice. We present the theoretical results about the construction of confidence intervals in the prediction of nonlinear time series modeled by FANN, this method is based on M-estimators that are a robust learning algorithm for parameter estimation when the data set is contaminated. The confidence interval that we propose is constructed from the study of the Influence Function of the estimator. We demonstrate our technique on computer generated Time Series data.",
keywords = "Confidence Interval, Feedforward Artificial Neural Networks, Robust Learning algorithm, Time Series",
author = "Rodrigo Salas and Romina Torres and H{\'e}ctor Allende and Claudio Moraga",
year = "2003",
language = "English",
volume = "2687",
pages = "441--448",
journal = "Lecture Notes in Computer Science",
issn = "0302-9743",
publisher = "Springer Verlag",

}

TY - JOUR

T1 - Robust estimation of confidence interval in neural networks applied to time series

AU - Salas, Rodrigo

AU - Torres, Romina

AU - Allende, Héctor

AU - Moraga, Claudio

PY - 2003

Y1 - 2003

N2 - Artificial neural networks (ANN) have been widely used in regression or predictions problems and it is usually desirable that some form of confidence bound is placed on the predicted value. A number of methods have been proposed for estimating the uncertainty associated with a value predicted by a feedforward neural network (FANN), but these methods are computationally intensive or only valid under certain assumptions, which are rarely satisfied in practice. We present the theoretical results about the construction of confidence intervals in the prediction of nonlinear time series modeled by FANN, this method is based on M-estimators that are a robust learning algorithm for parameter estimation when the data set is contaminated. The confidence interval that we propose is constructed from the study of the Influence Function of the estimator. We demonstrate our technique on computer generated Time Series data.

AB - Artificial neural networks (ANN) have been widely used in regression or predictions problems and it is usually desirable that some form of confidence bound is placed on the predicted value. A number of methods have been proposed for estimating the uncertainty associated with a value predicted by a feedforward neural network (FANN), but these methods are computationally intensive or only valid under certain assumptions, which are rarely satisfied in practice. We present the theoretical results about the construction of confidence intervals in the prediction of nonlinear time series modeled by FANN, this method is based on M-estimators that are a robust learning algorithm for parameter estimation when the data set is contaminated. The confidence interval that we propose is constructed from the study of the Influence Function of the estimator. We demonstrate our technique on computer generated Time Series data.

KW - Confidence Interval

KW - Feedforward Artificial Neural Networks

KW - Robust Learning algorithm

KW - Time Series

UR - http://www.scopus.com/inward/record.url?scp=21144438970&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:21144438970

VL - 2687

SP - 441

EP - 448

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

ER -