Roulette: A Pruning Framework to Train a Sparse Neural Network From Scratch

doi:10.1109/ACCESS.2021.3065406

	Roulette: A Pruning Framework to Train a Sparse Neural Network From Scratch
	Zhong, Qiaoling 1; Zhang, Zhibin ; Qiu, Qiang ; Cheng, Xueqi
	2021
发表期刊	IEEE ACCESS
ISSN	2169-3536
卷号	9 页码:51134-51145
摘要	Due to space and inference time restrictions, finding an efficient and sparse sub-network from a dense and over-parameterized network is critical for deploying neural networks on edge devices. Recent efforts explore obtaining a sparse sub-network by performing network pruning during training procedures to reduce training costs, such as memory and fioating-point operations (FLOPs). However, these works take more than 1.4 x the total number of iterations and try all possible pruning parameters manually to obtain sparse sub-networks. In this paper, we present a pruning framework Roulette to train a sparse network from scratch. First, we propose a novel method to train a sparse network by Pruning through the lens of Loss Landscape iteratively and automatically (PLL). We do a theoretical analysis that the curvature of the loss function is higher in the initial phase and can conduct us to start network pruning. According to our results on CIFAR-10/100 and ImageNet dataset, PLL saves up to 4x training FLOPs than prior works while maintaining comparable or even better accuracy. Then we design push and pull operations to synchronize the pruned weights on different GPUs during training, scaling PLL to multiple GPUs linearly. To our knowledge, Roulette is the first network pruning framework supporting multiple GPUs linearly.
关键词	Network pruning inference acceleration model compression multiple GPUs
DOI	10.1109/ACCESS.2021.3065406
收录类别	SCI
语种	英语
资助项目	Strategic Priority Research Program of Chinese Academy of Sciences (CAS)[XDA19020400]
WOS研究方向	Computer Science ; Engineering ; Telecommunications
WOS类目	Computer Science, Information Systems ; Engineering, Electrical & Electronic ; Telecommunications
WOS记录号	WOS:000638386800001
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计	被引频次：1[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/16768
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Zhong, Qiaoling
作者单位	1.Chinese Acad Sci, Inst Comp Technol, CAS Key Lab Network Data Sci & Technol, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Sch Comp & Control Engn, Beijing 100049, Peoples R China
推荐引用方式 GB/T 7714	Zhong, Qiaoling,Zhang, Zhibin,Qiu, Qiang,et al. Roulette: A Pruning Framework to Train a Sparse Neural Network From Scratch[J]. IEEE ACCESS,2021,9:51134-51145.
APA	Zhong, Qiaoling,Zhang, Zhibin,Qiu, Qiang,&Cheng, Xueqi.(2021).Roulette: A Pruning Framework to Train a Sparse Neural Network From Scratch.IEEE ACCESS,9,51134-51145.
MLA	Zhong, Qiaoling,et al."Roulette: A Pruning Framework to Train a Sparse Neural Network From Scratch".IEEE ACCESS 9(2021):51134-51145.