Smoothness Bias in Relevance Estimators for Feature Selection in Regression - Archive ouverte HAL Access content directly
Conference Papers Year : 2018

Smoothness Bias in Relevance Estimators for Feature Selection in Regression

(1, 2, 3) , (1) , (4)
1
2
3
4

Abstract

Selecting features from high-dimensional datasets is an important problem in machine learning. This paper shows that in the context of filter methods for feature selection, the estimator of the criterion used to select features plays an important role; in particular the estimators may suffer from a bias when comparing smooth and non-smooth features. This paper analyses the origin of such bias and investigates whether this bias influences the results of the feature selection process. Results show that non-smooth features tend to be penalised especially in small datasets.
Fichier principal
Vignette du fichier
467708_1_En_25_Chapter.pdf (1.3 Mo) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01821055 , version 1 (22-06-2018)

Licence

Attribution - CC BY 4.0

Identifiers

Cite

Alexandra Degeest, Michel Verleysen, Benoît Frenay. Smoothness Bias in Relevance Estimators for Feature Selection in Regression. 14th IFIP International Conference on Artificial Intelligence Applications and Innovations (AIAI), May 2018, Rhodes, Greece. pp.285-294, ⟨10.1007/978-3-319-92007-8_25⟩. ⟨hal-01821055⟩
27 View
72 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More