Feature Selection

Tony Bellotti, Ilia Nouretdinov, Meng Yang, Alexander Gammerman

Research output: Chapter in Book/Conference proceedingBook Chapterpeer-review

5 Citations (Scopus)

Abstract

In this chapter we consider the implementation of feature selection approaches within the Conformal Predictor framework. We begin with a review of feature selection, then consider several approaches to implementation. First, we use existing feature selection methods within conformal predictors, which raises some computational issues. Second, we use techniques specifically designed for conformal predictors: (1) the strangeness minimization feature selection (SMFS) method and (2) the average confidence maximization (ACM) method. SMFS minimizes the overall nonconformity values of a sequence of examples, whereas ACM maximizes the average confidence output by the conformal predictor using a subset of features. We also give some results and illustrations based on a medical dataset for abdominal pain diagnosis collected in a Scottish hospital.

Original languageEnglish
Title of host publicationConformal Prediction for Reliable Machine Learning
Subtitle of host publicationTheory, Adaptations and Applications
PublisherElsevier Inc.
Pages115-130
Number of pages16
ISBN (Print)9780123985378
DOIs
Publication statusPublished - Apr 2014
Externally publishedYes

Keywords

  • Confidence Maximization
  • Feature Selection
  • Microarray Classification
  • Nearest Centroid Classifier
  • Strangeness Minimization
  • Support Vector Machine

ASJC Scopus subject areas

  • General Computer Science

Fingerprint

Dive into the research topics of 'Feature Selection'. Together they form a unique fingerprint.

Cite this