trends.barnas.net

method · first seen 28 Oct 2026

soft cross-entropy loss

Also: SCE

Classification loss that trains against soft probability targets instead of one-hot labels.

Mentions per week

0119 May11 Aug10 NovPapersCodeDiscussion

PapersCodeDiscussion

Earliest sources

Latest sources

Mentioned together with

Plackett-Luce modelMixUp