A Multivariate Algorithm for Gene Selection Based on the Nearest Neighbor Probability

Authors:
Enrico Ferrari;Marco Muselli
Affiliations:
Institute of Electronics, Computer and Telecommunication Engineering, Italian National Research Council, Genoa, Italy 16149;Institute of Electronics, Computer and Telecommunication Engineering, Italian National Research Council, Genoa, Italy 16149
Venue:
Computational Intelligence Methods for Bioinformatics and Biostatistics
Year:
2009

Citing 4
Cited 0

Gene Selection for Cancer Classification using Support Vector Machines

Machine Learning
An introduction to variable and feature selection

The Journal of Machine Learning Research
Pattern Classification (2nd Edition)

Pattern Classification (2nd Edition)
Gene expression modeling through positive boolean functions

International Journal of Approximate Reasoning

Quantified Score

Hi-index	0.00

Visualization

Abstract

Experiments performed with DNA microarrays have very often the aim of retrieving a subset of genes involved in the discrimination between two physiological or pathological states (e.g. ill/healthy). Many methods have been proposed to solve this problem, among which the Signal to Noise ratio (S2N ) [5] and SVM-RFE [6]. Recently, the complementary approach to RFE, called Recursive Feature Addition (RFA ), has been successfully adopted. According to this approach, at each iteration the gene which maximizes a proper ranking function *** is selected, thus producing an ordering among the considered genes. In this paper an RFA method based on the nearest neighbor probability, named NN-RFA , is described and tested on some real world problems regarding the classification of human tissues. The results of such simulations show the ability of NN-RFA in retrieving a correct subset of genes for the problems at hand.