A Filter Approach to Feature Selection Based on Mutual Information

Author

Huang, Jinjie ; Cai, Yunze ; Xu, Xiaoming

Author_Institution

Dept. of Autom., Shanghai Jiao Tong Univ.

Volume

1

fYear

2006

fDate

17-19 July 2006

Firstpage

84

Lastpage

89

Abstract

In pattern recognition, feature selection aims to choose the smallest subset of features that is necessary and sufficient to describe the target concept. In this paper, a mutual information-based constructive criterion under arbitrary information distributions of input features is presented for feature selection. This criterion can capture both the relevance to the output classes and the redundancy with respect to the already-selected features without any parameters like beta in MIFS or MIFS-U methods to be preset. Furthermore, a modified greedy feature selection algorithm called MICC is proposed, and experimental results demonstrate the good performance of MICC on both synthetic and benchmark data sets

Keywords

feature extraction; filtering theory; greedy algorithms; learning (artificial intelligence); pattern classification; MICC; filter approach; greedy feature selection algorithm; machine learning; mutual information-based constructive criterion; pattern classification; pattern recognition; Computational complexity; Concrete; Information filtering; Information filters; Machine learning; Machine learning algorithms; Mutual information; Noise measurement; Pattern classification; Pattern recognition; Pattern classification; feature selection; filter approach; machine learning; mutual information;

fLanguage

English

Publisher

ieee

Conference_Titel

Cognitive Informatics, 2006. ICCI 2006. 5th IEEE International Conference on

Conference_Location

Beijing

Print_ISBN

1-4244-0475-4

Type

conf

DOI

10.1109/COGINF.2006.365681

Filename

4216396