1 paper
Yubin Ge, Site Li, Xuyang Li +4
The widely-used cross-entropy (CE) loss-based deep networks achieved significant progress w.r.t. the classification accuracy. However, the CE loss can essentially ignore the risk o…