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 in a book

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
Place of PublicationGermany
PublisherSpringer
Pages371-389
Number of pages19
ISBN (Print)9781846281198
DOIs
Publication statusPublished - 2005

Publication series

NameAdvanced information and knowledge processing
PublisherSpringer

Keywords

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

Fingerprint Dive into the research topics of 'Evaluating the effectiveness of Bayesian feature selection'. Together they form a unique fingerprint.

  • 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