ABSTRACT
In order to effectively explore useful information from a huge amount of high-dimensional small sample data, feature selection has become critical for the analysis and processing of information security high-dimensional data. Experimental data adopt three high-dimensional small sample data sets, Colon, Leukemia, and Lung for algorithm performance verification. According to the feature selection problem of the high-dimensional small sample safety data, this chapter aims to combine the Memetic Algorithm (MA) and Least-Squares Support Vector Machines (LSSVM) to design a type of wrapper feature selection method (MA-LSSVM). The experimental results demonstrate that MALSSVM can be more efficiently and stably used to obtain features that largely contribute to the classification precision and then reduce the data dimension and improve the classification efficiency. The MA-LSSVM depends on certain or a variety of learning algorithms at feature selection process. The LS-SVM algorithm accelerates the iterative evolution process of the population on the aspect of classification effects and speed for nonlinear high-dimensional data.
