ROC curves. Representative ROC curves for three kernel-based SVM classifiers generated using the threefold cross-validation with the mRNA expression microarray dataset for training only. The ROC curves were generated by varying the decision threshold of each SVM classifier. The average AUC for the linear kernel, polynomial kernel and RBF kernel are 0.89, 0.85, and 0.95, respectively. ROC: receiver operating characteristic; TPR: true positive rate; FPR: false positive rate; AUC: area under the curve.