Teacher Assistant-Based Knowledge Distillation Extracting Multi-Level Features on Single Channel Sleep EEG

Venue: IJCAI

Authors: Heng Liang, Yucheng Liu, Haichao Wang (equal contribution), Ziyu Jia

We propose SleepKD, a novel general knowledge distillation framework for sleep stage classification. The multi-level module transfers multi-level knowledge from the teacher model to the student model. The teacher assistant module bridges the large gap between teacher and student networks, further improving the distillation.

Tags: Knowledge Distillation, Sleep EEG, Multi-Level Features