Feature selection based on distance correlation: a filter algorithm
Open Access
- 7 September 2020
- journal article
- research article
- Published by Taylor & Francis in Journal of Applied Statistics
- Vol. 49 (2) , 411-426
Abstract
Feature selection (FS) is one of the most powerful techniques to cope with the curse of dimensionality. In the study, a new filter approach to feature selection based on distance correlation is presented (DCFS, for short), which keeps the model-free advantage without any pre-specified parameters. Our method consists of two steps: hard step (forward selection) and soft step (backward selection). In the hard step, two types of associations, between univariate feature and the classes and between group feature and the classes, are involved to pick out the most relevant features with respect to the target classes. Due to the strict screening condition in the first step, some of the useful features are likely removed. Therefore, in the soft step, a feature-relationship gain (like feature score) based on the distance correlation is introduced, which is concerned with five kinds of associations. We sort the feature gain values and implement the backward selection procedure until the errors stop declining. The simulation results show that our method becomes more competitive on several datasets compared with some of the representative feature selection methods based on several classification models.Keywords
Funding Information
- National Natural Science Foundation of China (61732019)
This publication has 20 references indexed in Scilit:
- Feature selection in machine learning: A new perspectiveNeurocomputing, 2018
- Class-specific mutual information variation for feature selectionPattern Recognition, 2018
- An effective feature selection method based on pair-wise feature proximity for high dimensional low sample size data2017
- Recent advances in feature selection and its applicationsKnowledge and Information Systems, 2017
- Feature selection via binary simultaneous perturbation stochastic approximationPattern Recognition Letters, 2016
- Model-Free Feature Screening for Ultrahigh Dimensional Discriminant AnalysisJournal of the American Statistical Association, 2015
- Data Preprocessing in Data Mining2015
- Correction: Discussion of Brownian distance covarianceThe Annals of Applied Statistics, 2013
- A new wrapper feature selection approach using neural networkNeurocomputing, 2010
- Normalized Mutual Information Feature SelectionIEEE Transactions on Neural Networks, 2009