Optimizing Area Under Roc Curve with SVMs
Alain Rakotomamonjy · 2004
For many years now, there is a growing interest around ROC curve for characterizing machine learning performances. This is particularly due to the fact that in real-world problems misclassifi- cation costs are not known and thus, ROC curve and related metrics such as the Area Under ROC curve (AUC) can be a more mean- ingful performance measures. In this paper, we propose a quadratic programming based algorithm for AUC maximization and show that under certain conditions 2-norm soft margin Support Vector Ma- chines can also maximize AUC. We present experiments that com- pare SVMs performances to those of other AUC maximization based algorithms and provide empirical analysis of SVMs behavior with regards to ROC- based metrics. Our main conclusion is that SVMs can maximize both AUC and accuracy compared to other algorithms like RankBoost that optimize only AUC 2 .