Dimensionality Reduction Methods For Gene Expression Data Base On SVM

Bioinformatics has got great progress in recent years, with the rapid development of Gene Chip, we can get reliable gene expression data more efficiently, and these data provide a foundation of uncovering the secret of life. But the data with high dimensionality, few samples and nonlinear characteristics, leading to"dimensionality tragedy", which is a new challenge for some traditional machine learning methods.Support Vector Machine (SVM) is a type of new machine learning method based on statistical learning theory. It uses the principle of minimizing structural risk, which overcomes the small sample learning problem. Moreover, it employs kernel function, changing nonlinear problems into linear ones by using mapping the dimension original space to high dimension feature space, which makes the algorithm be realized easily. Because of such advantage, SVMs become a hot spot of machine learning theory.This paper applies dimensionality reduction methods to the gene expression data, aimed at the characteristic of high dimensionality and small sample, uses linear and nonlinear methods to reduce the dimensionality of the original data, then uses the SVM to classify the reduced data, achieving better results. Lastly, the papers compares two data sets usually used, the results show that the method we put forward performs better than traditional classify methods.