REGRESSION-TEST MODEL FOR HIGH DIMENSIONAL FEATURE SELECTION

摘要

Feature selection is a classical problem in pattern recognition. Feature selection when the number of features is much higher than the number of samples is a new issue that we seldom face before. But now more and more such problems emerge out. Gene selection, an important issue in medicine and biology, is such a problem. Many of traditional feature selection methods cannot perform well in that case. This paper proposes a Regression-Test model for high dimension feature selection. This method has a good performance when the dimension of feature space is very high. Experiment result in public data has demonstrated the validity of it.