Weighted Partly Adaptive Elastic Net And Its Application In Cancer Diagnosis

The exploited of statistical machine learning methods in microarray data for gene expression at the level of tumor diagnosis helps cancer early detection and accurate judg-ment, which has been attracted extensive attention in the life sciences, medicine and information science. Adaptive and resilient network expansion model can adaptively se-lect groups of genes, and thus be successfully applied to cancer diagnosis and screening of key genes. However, when there is noise in the sample, the classification accuracy will be seriously affected. In order to solve this problem, this paper presents a new statistical machine learning model.Specifically, the article has the following contributions:1?Based on the information of the class and the class, a double weighted mechanism is proposed, and a weighted partial adaptive elastic network model is constructed. By using the glmnet toolkit of R language, a completely regularized solution algorithm is developed, and suppress the impact of noise on the classification accuracy successfully;2?The adaptive group gene selection effect of machine learning model under the sample-weighted is described by using mathematical language, and the method of deriva-tion is used to prove that the proposed model can encourage the effect.3?This paper apply the model and algorithm to the diagnosis of three types of cancer,and screen cancer-related genes successfully.