Gene Selection in Cancer Classification Using Sparse Logistic Regression with L1/2 Regularization

Abstract

In recent years, gene selection for cancer classification based on the expression of a small number of gene biomarkers has been the subject of much research in genetics and molecular biology. The successful identification of gene biomarkers will help in the classification of different types of cancer and improve the prediction accuracy. Recently, regularized logistic regression using the L 1 regularization has been successfully applied in high-dimensional cancer classification to tackle both the estimation of gene coefficients and the simultaneous performance of gene selection. However, the L 1 has a biased gene selection and dose not have the oracle property. To address these problems, we investigate L 1 / 2 regularized logistic regression for gene selection in cancer classification. Experimental results on three DNA microarray datasets demonstrate that our proposed method outperforms other commonly used sparse methods ( L 1 and L E N ) in terms of classification performance.

Optimal Classification of Lung Cancer Related Genes using Enhanced reliefF Algorithm and Multiclass Support Vector Machine

“…In this research work, MATLAB (version 2018a) was used for experimental simulation with i5 processor and 3.2 GHz. In order to estimate the effectiveness of proposed system, the performance of proposed system was compared with a few existing systems such as, sparse logistic regression with L1/2 regularization [12], GEP multi classification using decomposition schemes [14], and GEP models [16] on GEO dataset. The proposed system performance was evaluated by means of TPR, accuracy, FPR and error rate.…”

“…In this segment, some major contributions to the earlier research papers are presented. S. Wu, H. Jiang, H. Shen, and Z. Yang, [12] developed a new methodology for gene selection named as L1/2 logistic regression model. In this research work, the developed method uses new univariate half thresholding for updating the estimated coefficients.…”

“…Table 5 analysis the performance of proposed and existing systems. S. Wu, H. Jiang, H. Shen, and Z. Yang, [12] developed a new methodology (L1/2 regularized logistic regression) to select the optimal genes for classification. In this research, GEO dataset (GEO IDs: GSE10072 and GSE19804) is utilized to evaluate the performance of the developed system.…”

Biomarker discovery for predicting spontaneous preterm birth from gene expression data by regularized logistic regression

“…Obviously, when , the penalty refers to ridge penalty, and when , the penalty refers to lasso penalty, respectively. can be set literally to be , the penalty is an elastic net penalty [22] .…”

“…Effective gene selection methods are desirable to classify different phenotypic states of SPTB. The classification accuracy is our objective function of optimization in biomarker discovery [21] , [22] . From a statistical perspective, too many variables may lead to multicollinearity [23] .…”