TY - GEN
T1 - Efficient AUC optimization for classification
AU - Calders, T.
AU - Jaroszewicz, S.
PY - 2007
Y1 - 2007
N2 - In this paper we show an efficient method for inducing classifiers that directly optimize the area under the ROC curve. Recently, AUC gained importance in the classification community as a mean to compare the performance of classifiers. Because most classification methods do not optimize this measure directly, several classification learning methods are emerging that directly optimize the AUC. These methods, however, require many costly computations of the AUC, and hence, do not scale well to large datasets. In this paper, we develop a method to increase the efficiency of computing AUC based on a polynomial approximation of the AUC. As a proof of concept, the approximation is plugged into the construction of a scalable linear classifier that directly optimizes AUC using a gradient descent method. Experiments on real-life datasets show a high accuracy and efficiency of the polynomial approximation.
AB - In this paper we show an efficient method for inducing classifiers that directly optimize the area under the ROC curve. Recently, AUC gained importance in the classification community as a mean to compare the performance of classifiers. Because most classification methods do not optimize this measure directly, several classification learning methods are emerging that directly optimize the AUC. These methods, however, require many costly computations of the AUC, and hence, do not scale well to large datasets. In this paper, we develop a method to increase the efficiency of computing AUC based on a polynomial approximation of the AUC. As a proof of concept, the approximation is plugged into the construction of a scalable linear classifier that directly optimizes AUC using a gradient descent method. Experiments on real-life datasets show a high accuracy and efficiency of the polynomial approximation.
U2 - 10.1007/978-3-540-74976-9_8
DO - 10.1007/978-3-540-74976-9_8
M3 - Conference contribution
SN - 978-3-540-74975-2
T3 - Lecture Notes in Computer Science
SP - 42
EP - 53
BT - Proceedings of the 11th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD 2007) 17-21 September 2007, Warsaw, Poland
A2 - Kok, J.N.
A2 - Koronacki, J.
A2 - Lopez de Mantaras, R.
A2 - Matwin, S.
A2 - Mladenic, D.
A2 - Skowron, A.
PB - Springer
CY - Berlin, Germany
T2 - conference; PKDD 2007, Warsaw, Poland; 2007-09-17; 2007-09-21
Y2 - 17 September 2007 through 21 September 2007
ER -