Home // BIOTECHNO 2016, The Eighth International Conference on Bioinformatics, Biocomputational Systems and Biotechnologies // View article

Matroska Feature Selection Method for Microarray Data

Authors:
Shuichi Shinmura

Keywords: Minimum Number of Misclassifications (MNM); Revised IP-OLDF; SVM; Fisher’s LDF; Basic Gene Subspaces (BGS); Lasso.

Abstract:
In this paper, we propose a Matroska feature selection method (the method) for microarray data (the data). We had already established a new theory of the discriminant analysis (the theory) and developed a linear discriminant function (LDF) named Revised IP-OLDF. This LDF can make feature selection for the data naturally. We confirmed this feature selection of Revised IP-OLDF by Swiss banknote data and Japanese automobile data. It finds the data consists of several small genes subspaces that are linearly separable. Therefore, we need not struggle with high-dimension genes space. In this paper, we develop a LINGO program to find all small genes subspaces called the small Matroska (SM). Because it is very easy for us to analyze these SMs, we may be able to find new facts of gene analysis.

Pages: 1 to 8

Copyright: Copyright (c) IARIA, 2016

Publication date: June 26, 2016

Published in: conference

ISSN: 2308-4383

ISBN: 978-1-61208-488-6

Location: Lisbon, Portugal

Dates: from June 26, 2016 to June 30, 2016