Timezone: »

 
Oral
On Symmetric Losses for Learning from Corrupted Labels
Nontawat Charoenphakdee · Jongyeong Lee · Masashi Sugiyama

Tue Jun 11 02:40 PM -- 03:00 PM (PDT) @ Seaside Ballroom
in Supervised Learning »

This paper aims to provide a better understanding of a symmetric loss. First, we show that using a symmetric loss is advantageous in the balanced error rate (BER) minimization and area under the receiver operating characteristic curve (AUC) maximization from corrupted labels. Second, we prove general theoretical properties of symmetric losses, including a classification-calibration condition, excess risk bound, conditional risk minimizer, and AUC-consistency condition. Third, since all nonnegative symmetric losses are non-convex, we propose a convex barrier hinge loss that benefits significantly from the symmetric condition, although it is not symmetric everywhere. Finally, we conduct experiments on BER and AUC optimization from corrupted labels to validate the relevance of the symmetric condition.

[ Slides]  [ Video] 

Author Information

Nontawat Charoenphakdee (The University of Tokyo / RIKEN)
Jongyeong Lee (The University of Tokyo/RIKEN)
Masashi Sugiyama (RIKEN / The University of Tokyo)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors