Evaluating the effectiveness of Bayesian feature selection

Ian T. Nabney, David J. Evans, Yann Brule, Caroline Gordon

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

A practical Bayesian approach for inference in neural network models has been available for ten years, and yet it is not used frequently in medical applications. In this chapter we show how both regularisation and feature selection can bring significant benefits in diagnostic tasks through two case studies: heart arrhythmia classification based on ECG data and the prognosis of lupus. In the first of these, the number of variables was reduced by two thirds without significantly affecting performance, while in the second, only the Bayesian models had an acceptable accuracy. In both tasks, neural networks outperformed other pattern recognition approaches.
Original languageEnglish
Title of host publicationApplications of probabilistic modelling in medical informatics and bioinformatics
EditorsR. Dybowski, D. Husmeier, S. J. Roberts
PublisherSpringer
Pages371-389
Number of pages19
ISBN (Print)9781846281198
DOIs
Publication statusPublished - 2005

Publication series

NameAdvanced information and knowledge processing
PublisherSpringer Verlag

Fingerprint

Feature extraction
Neural networks
Medical applications
Electrocardiography
Pattern recognition

Keywords

  • Bayesian approach
  • inference
  • neural network models
  • diagnostic tasks
  • heart arrhythmia
  • prognosis of lumps

Cite this

Nabney, I. T., Evans, D. J., Brule, Y., & Gordon, C. (2005). Evaluating the effectiveness of Bayesian feature selection. In R. Dybowski, D. Husmeier, & S. J. Roberts (Eds.), Applications of probabilistic modelling in medical informatics and bioinformatics (pp. 371-389). (Advanced information and knowledge processing). Springer. https://doi.org/10.1007/1-84628-119-9_12
Nabney, Ian T. ; Evans, David J. ; Brule, Yann ; Gordon, Caroline. / Evaluating the effectiveness of Bayesian feature selection. Applications of probabilistic modelling in medical informatics and bioinformatics. editor / R. Dybowski ; D. Husmeier ; S. J. Roberts. Springer, 2005. pp. 371-389 (Advanced information and knowledge processing).
@inbook{4e9387fe9f224606a9ae7cb67b0678a2,
title = "Evaluating the effectiveness of Bayesian feature selection",
abstract = "A practical Bayesian approach for inference in neural network models has been available for ten years, and yet it is not used frequently in medical applications. In this chapter we show how both regularisation and feature selection can bring significant benefits in diagnostic tasks through two case studies: heart arrhythmia classification based on ECG data and the prognosis of lupus. In the first of these, the number of variables was reduced by two thirds without significantly affecting performance, while in the second, only the Bayesian models had an acceptable accuracy. In both tasks, neural networks outperformed other pattern recognition approaches.",
keywords = "Bayesian approach, inference, neural network models, diagnostic tasks, heart arrhythmia, prognosis of lumps",
author = "Nabney, {Ian T.} and Evans, {David J.} and Yann Brule and Caroline Gordon",
year = "2005",
doi = "10.1007/1-84628-119-9_12",
language = "English",
isbn = "9781846281198",
series = "Advanced information and knowledge processing",
publisher = "Springer",
pages = "371--389",
editor = "R. Dybowski and D. Husmeier and Roberts, {S. J.}",
booktitle = "Applications of probabilistic modelling in medical informatics and bioinformatics",
address = "Germany",

}

Nabney, IT, Evans, DJ, Brule, Y & Gordon, C 2005, Evaluating the effectiveness of Bayesian feature selection. in R Dybowski, D Husmeier & SJ Roberts (eds), Applications of probabilistic modelling in medical informatics and bioinformatics. Advanced information and knowledge processing, Springer, pp. 371-389. https://doi.org/10.1007/1-84628-119-9_12

Evaluating the effectiveness of Bayesian feature selection. / Nabney, Ian T.; Evans, David J.; Brule, Yann; Gordon, Caroline.

Applications of probabilistic modelling in medical informatics and bioinformatics. ed. / R. Dybowski; D. Husmeier; S. J. Roberts. Springer, 2005. p. 371-389 (Advanced information and knowledge processing).

Research output: Chapter in Book/Report/Conference proceedingChapter

TY - CHAP

T1 - Evaluating the effectiveness of Bayesian feature selection

AU - Nabney, Ian T.

AU - Evans, David J.

AU - Brule, Yann

AU - Gordon, Caroline

PY - 2005

Y1 - 2005

N2 - A practical Bayesian approach for inference in neural network models has been available for ten years, and yet it is not used frequently in medical applications. In this chapter we show how both regularisation and feature selection can bring significant benefits in diagnostic tasks through two case studies: heart arrhythmia classification based on ECG data and the prognosis of lupus. In the first of these, the number of variables was reduced by two thirds without significantly affecting performance, while in the second, only the Bayesian models had an acceptable accuracy. In both tasks, neural networks outperformed other pattern recognition approaches.

AB - A practical Bayesian approach for inference in neural network models has been available for ten years, and yet it is not used frequently in medical applications. In this chapter we show how both regularisation and feature selection can bring significant benefits in diagnostic tasks through two case studies: heart arrhythmia classification based on ECG data and the prognosis of lupus. In the first of these, the number of variables was reduced by two thirds without significantly affecting performance, while in the second, only the Bayesian models had an acceptable accuracy. In both tasks, neural networks outperformed other pattern recognition approaches.

KW - Bayesian approach

KW - inference

KW - neural network models

KW - diagnostic tasks

KW - heart arrhythmia

KW - prognosis of lumps

UR - http://www.springerlink.com/content/l37672/?p=768d9aa0a4ba4f54874bf9aa43c36988&pi=0

U2 - 10.1007/1-84628-119-9_12

DO - 10.1007/1-84628-119-9_12

M3 - Chapter

SN - 9781846281198

T3 - Advanced information and knowledge processing

SP - 371

EP - 389

BT - Applications of probabilistic modelling in medical informatics and bioinformatics

A2 - Dybowski, R.

A2 - Husmeier, D.

A2 - Roberts, S. J.

PB - Springer

ER -

Nabney IT, Evans DJ, Brule Y, Gordon C. Evaluating the effectiveness of Bayesian feature selection. In Dybowski R, Husmeier D, Roberts SJ, editors, Applications of probabilistic modelling in medical informatics and bioinformatics. Springer. 2005. p. 371-389. (Advanced information and knowledge processing). https://doi.org/10.1007/1-84628-119-9_12