TY - GEN
T1 - AUC Maximization in Bayesian hierarchical models
AU - Gonen, Mehmet
PY - 2016
Y1 - 2016
N2 - The area under the curve (AUC) measures such as the area under the receiver operating characteristics curve (AUROC) and the area under the precision-recall curve (AUPR) are known to be more appropriate than the error rate, especially, for imbalanced data sets. There are several algorithms to optimize AUC measures instead of minimizing the error rate. However, this idea has not been fully exploited in Bayesian hierarchical models owing to the difficulties in inference. Here, we formulate a general Bayesian inference framework, called Bayesian AUC Maximization (BAM), to integrate AUC maximization into Bayesian hierarchical models by borrowing the pairwise and listwise ranking ideas from the information retrieval literature. To showcase our BAM framework, we develop two Bayesian linear classifier variants for two ranking approaches and derive their variational inference procedures. We perform validation experiments on four biomedical data sets to demonstrate the better predictive performance of our framework over its error-minimizing counterpart in terms of average AUROC and AUPR values.
AB - The area under the curve (AUC) measures such as the area under the receiver operating characteristics curve (AUROC) and the area under the precision-recall curve (AUPR) are known to be more appropriate than the error rate, especially, for imbalanced data sets. There are several algorithms to optimize AUC measures instead of minimizing the error rate. However, this idea has not been fully exploited in Bayesian hierarchical models owing to the difficulties in inference. Here, we formulate a general Bayesian inference framework, called Bayesian AUC Maximization (BAM), to integrate AUC maximization into Bayesian hierarchical models by borrowing the pairwise and listwise ranking ideas from the information retrieval literature. To showcase our BAM framework, we develop two Bayesian linear classifier variants for two ranking approaches and derive their variational inference procedures. We perform validation experiments on four biomedical data sets to demonstrate the better predictive performance of our framework over its error-minimizing counterpart in terms of average AUROC and AUPR values.
UR - http://www.scopus.com/inward/record.url?scp=85013042405&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85013042405&partnerID=8YFLogxK
U2 - 10.3233/978-1-61499-672-9-21
DO - 10.3233/978-1-61499-672-9-21
M3 - Conference contribution
AN - SCOPUS:85013042405
VL - 285
T3 - Frontiers in Artificial Intelligence and Applications
SP - 21
EP - 27
BT - Frontiers in Artificial Intelligence and Applications
PB - IOS Press
T2 - 22nd European Conference on Artificial Intelligence, ECAI 2016
Y2 - 29 August 2016 through 2 September 2016
ER -