「Deep Learning」综述:模型压缩与加速

Sina Weibo:小锋子Shawn
Tencent E-mail:[email protected]
http://blog.csdn.net/dgyuanshaofeng/article/details/83389069

1 方法分类:

1、参数修剪和共享(parameter pruning and sharing)
2、低秩分解(low-rank factorization)
3、转移/紧凑卷积核(transfered/compact convolutional filters)
4、知识萃取(knowledge distillation)

[1] A Survey of Model Compression and Acceleration for DNNs 2017 [paper]
[2] A Survey on Acceleration of DCNNs 2018 [paper]

你可能感兴趣的:(「Deep Learning」综述:模型压缩与加速)