Regularization paths of L₁-penalized ROC Curve-Optimizing Support Vector Machines

The receiver operator characteristic (ROC) curve is one of the most popular tools to evaluate the performance of binary classifiers in a variety of applications. Rakotomamonjy (2004) proposed the ROC-SVM that directly optimizes the area under the ROC curve instead of the prediction accuracy. In this article, we study the L₁-penalized ROC-SVM that directly optimizes the ROC curve. We first show that the L₁-penalized ROC-SVM has piecewise linear regularization paths and then develop an efficient algorithm to compute the entire paths, which greatly facilitates its tuning procedure.