Neural network validation: an illustration from the monitoring of multi-phase flows

C.M. Bishop

Neural network validation: an illustration from the monitoring of multi-phase flows

School of Informatics

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

One of the key factors limiting the use of neural networks in many industrial applications has been the difficulty of demonstrating that a trained network will continue to generate reliable outputs once it is in routine use. An important potential source of errors arises from novel input data, that is input data which differs significantly from the data used to train the network. In this paper, the authors investigate the relation between the degree of novelty of input data and the corresponding reliability of the outputs from the network. They describe a quantitative procedure for assessing novelty, and demonstrate its performance using an application involving the monitoring of oil flow in multiphase pipelines

Original language	English
Title of host publication	Artificial Neural Networks, 1993., Third International Conference on
Publisher	IET
Pages	41-45
Number of pages	5
Publication status	Published - 1 May 1993

Keywords

engineering computing
neural nets
oil technology
performance evaluation
reliability
errors
input data
multiphase oil flow monitoring
multiphase pipelines
neural networks
output reliability

Access to Document

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=263260

Fingerprint Dive into the research topics of 'Neural network validation: an illustration from the monitoring of multi-phase flows'. Together they form a unique fingerprint.

Cite this

@inproceedings{5ca28db716694981b2f4edf78027599b,

title = "Neural network validation: an illustration from the monitoring of multi-phase flows",

abstract = "One of the key factors limiting the use of neural networks in many industrial applications has been the difficulty of demonstrating that a trained network will continue to generate reliable outputs once it is in routine use. An important potential source of errors arises from novel input data, that is input data which differs significantly from the data used to train the network. In this paper, the authors investigate the relation between the degree of novelty of input data and the corresponding reliability of the outputs from the network. They describe a quantitative procedure for assessing novelty, and demonstrate its performance using an application involving the monitoring of oil flow in multiphase pipelines",

keywords = "engineering computing, neural nets, oil technology, performance evaluation, reliability, errors, input data, multiphase oil flow monitoring, multiphase pipelines, neural networks, output reliability",

author = "C.M. Bishop",

year = "1993",

month = may,

day = "1",

language = "English",

pages = "41--45",

booktitle = "Artificial Neural Networks, 1993., Third International Conference on",

publisher = "IET",

}

TY - GEN

T1 - Neural network validation: an illustration from the monitoring of multi-phase flows

AU - Bishop, C.M.

PY - 1993/5/1

Y1 - 1993/5/1

N2 - One of the key factors limiting the use of neural networks in many industrial applications has been the difficulty of demonstrating that a trained network will continue to generate reliable outputs once it is in routine use. An important potential source of errors arises from novel input data, that is input data which differs significantly from the data used to train the network. In this paper, the authors investigate the relation between the degree of novelty of input data and the corresponding reliability of the outputs from the network. They describe a quantitative procedure for assessing novelty, and demonstrate its performance using an application involving the monitoring of oil flow in multiphase pipelines

AB - One of the key factors limiting the use of neural networks in many industrial applications has been the difficulty of demonstrating that a trained network will continue to generate reliable outputs once it is in routine use. An important potential source of errors arises from novel input data, that is input data which differs significantly from the data used to train the network. In this paper, the authors investigate the relation between the degree of novelty of input data and the corresponding reliability of the outputs from the network. They describe a quantitative procedure for assessing novelty, and demonstrate its performance using an application involving the monitoring of oil flow in multiphase pipelines

KW - engineering computing

KW - neural nets

KW - oil technology

KW - performance evaluation

KW - reliability

KW - errors

KW - input data

KW - multiphase oil flow monitoring

KW - multiphase pipelines

KW - neural networks

KW - output reliability

M3 - Conference contribution

SP - 41

EP - 45

BT - Artificial Neural Networks, 1993., Third International Conference on

PB - IET

ER -