Multiclass Cancer Classification by Using Fuzzy Support Vector Machine and Binary Decision Tree With Gene Selection

摘要

We investigate the problems of multiclass cancer classificationwith gene selection from gene expression data. Two differentconstructed multiclass classifiers with gene selection areproposed, which are fuzzy support vector machine (FSVM) with geneselection and binary classification tree based on SVM with geneselection. Using F test and recursive feature elimination based onSVM as gene selection methods, binary classification tree based onSVM with F test, binary classification tree based on SVM withrecursive feature elimination based on SVM, and FSVM withrecursive feature elimination based on SVM are tested in ourexperiments. To accelerate computation, preselecting the strongestgenes is also used. The proposed techniques are applied to analyzebreast cancer data, small round blue-cell tumors, and acuteleukemia data. Compared to existing multiclass cancer classifiersand binary classification tree based on SVM with F test or binaryclassification tree based on SVM with recursive featureelimination based on SVM mentioned in this paper, FSVM based onrecursive feature elimination based on SVM can find most importantgenes that affect certain types of cancer with high recognition accuracy.