Distilling Knowledge by Mimicking Features
Knowledge distillation (KD) is a popular method to train efficient networks ("student") with the help of high-capacity networks ("teacher"). Traditional methods use the teacher's soft logits as extra supervision to train the student network. In this paper, we argue that it i...
Veröffentlicht in: | IEEE transactions on pattern analysis and machine intelligence. - 1979. - 44(2022), 11 vom: 31. Nov., Seite 8183-8195 |
---|---|
1. Verfasser: | |
Weitere Verfasser: | , |
Format: | Online-Aufsatz |
Sprache: | English |
Veröffentlicht: |
2022
|
Zugriff auf das übergeordnete Werk: | IEEE transactions on pattern analysis and machine intelligence |
Schlagworte: | Journal Article Research Support, Non-U.S. Gov't |
Online verfügbar |
Volltext |