BriefGPT.xyz
Sep, 2023
高效在线主动蒸馏的查询策略
On the Query Strategies for Efficient Online Active Distillation
HTML
PDF
Michele Boldo, Enrico Martini, Mirco De Marchi, Stefano Aldegheri, Nicola Bombieri
TL;DR
通过评估一组查询策略，在人体姿势估计应用中，通过两种方法（离线方法和基于知识蒸馏的在线评估）评估训练过程中所选择的帧对训练结果的影响，证明了在实时场景中有效地将边缘轻量级模型适应于新环境的训练的可能性。
Abstract
deep learning
(DL) requires lots of time and data, resulting in high computational demands. Recently, researchers employ
active learning
(AL) and
→