One-Pass AUC Optimization

Wei Gao, Rong Jin, Shenghuo Zhu, Zhi-Hua Zhou
Proceedings of the 30th International Conference on Machine Learning, PMLR 28(3):906-914, 2013.

Abstract

AUC is an important performance measure and many algorithms have been devoted to AUC optimization, mostly by minimizing a surrogate convex loss on a training data set. In this work, we focus on one-pass AUC optimization that requires only going through the training data once without storing the entire training dataset, where conventional online learning algorithms cannot be applied directly because AUC is measured by a sum of losses defined over pairs of instances from different classes. We develop a regression-based algorithm which only needs to maintain the first and second order statistics of training data in memory, resulting a storage requirement independent from the size of training data. To efficiently handle high dimensional data, we develop a randomized algorithm that approximates the covariance matrices by low rank matrices. We verify, both theoretically and empirically, the effectiveness of the proposed algorithm.

Cite this Paper


BibTeX
@InProceedings{pmlr-v28-gao13, title = {One-Pass AUC Optimization}, author = {Gao, Wei and Jin, Rong and Zhu, Shenghuo and Zhou, Zhi-Hua}, booktitle = {Proceedings of the 30th International Conference on Machine Learning}, pages = {906--914}, year = {2013}, editor = {Dasgupta, Sanjoy and McAllester, David}, volume = {28}, number = {3}, series = {Proceedings of Machine Learning Research}, address = {Atlanta, Georgia, USA}, month = {17--19 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v28/gao13.pdf}, url = {https://proceedings.mlr.press/v28/gao13.html}, abstract = {AUC is an important performance measure and many algorithms have been devoted to AUC optimization, mostly by minimizing a surrogate convex loss on a training data set. In this work, we focus on one-pass AUC optimization that requires only going through the training data once without storing the entire training dataset, where conventional online learning algorithms cannot be applied directly because AUC is measured by a sum of losses defined over pairs of instances from different classes. We develop a regression-based algorithm which only needs to maintain the first and second order statistics of training data in memory, resulting a storage requirement independent from the size of training data. To efficiently handle high dimensional data, we develop a randomized algorithm that approximates the covariance matrices by low rank matrices. We verify, both theoretically and empirically, the effectiveness of the proposed algorithm.} }
Endnote
%0 Conference Paper %T One-Pass AUC Optimization %A Wei Gao %A Rong Jin %A Shenghuo Zhu %A Zhi-Hua Zhou %B Proceedings of the 30th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2013 %E Sanjoy Dasgupta %E David McAllester %F pmlr-v28-gao13 %I PMLR %P 906--914 %U https://proceedings.mlr.press/v28/gao13.html %V 28 %N 3 %X AUC is an important performance measure and many algorithms have been devoted to AUC optimization, mostly by minimizing a surrogate convex loss on a training data set. In this work, we focus on one-pass AUC optimization that requires only going through the training data once without storing the entire training dataset, where conventional online learning algorithms cannot be applied directly because AUC is measured by a sum of losses defined over pairs of instances from different classes. We develop a regression-based algorithm which only needs to maintain the first and second order statistics of training data in memory, resulting a storage requirement independent from the size of training data. To efficiently handle high dimensional data, we develop a randomized algorithm that approximates the covariance matrices by low rank matrices. We verify, both theoretically and empirically, the effectiveness of the proposed algorithm.
RIS
TY - CPAPER TI - One-Pass AUC Optimization AU - Wei Gao AU - Rong Jin AU - Shenghuo Zhu AU - Zhi-Hua Zhou BT - Proceedings of the 30th International Conference on Machine Learning DA - 2013/05/26 ED - Sanjoy Dasgupta ED - David McAllester ID - pmlr-v28-gao13 PB - PMLR DP - Proceedings of Machine Learning Research VL - 28 IS - 3 SP - 906 EP - 914 L1 - http://proceedings.mlr.press/v28/gao13.pdf UR - https://proceedings.mlr.press/v28/gao13.html AB - AUC is an important performance measure and many algorithms have been devoted to AUC optimization, mostly by minimizing a surrogate convex loss on a training data set. In this work, we focus on one-pass AUC optimization that requires only going through the training data once without storing the entire training dataset, where conventional online learning algorithms cannot be applied directly because AUC is measured by a sum of losses defined over pairs of instances from different classes. We develop a regression-based algorithm which only needs to maintain the first and second order statistics of training data in memory, resulting a storage requirement independent from the size of training data. To efficiently handle high dimensional data, we develop a randomized algorithm that approximates the covariance matrices by low rank matrices. We verify, both theoretically and empirically, the effectiveness of the proposed algorithm. ER -
APA
Gao, W., Jin, R., Zhu, S. & Zhou, Z.. (2013). One-Pass AUC Optimization. Proceedings of the 30th International Conference on Machine Learning, in Proceedings of Machine Learning Research 28(3):906-914 Available from https://proceedings.mlr.press/v28/gao13.html.

Related Material