Balanced Quantization: An Effective and Efficient Approach to Quantized Neural Networks

doi:10.1007/s11390-017-1750-y

	Balanced Quantization: An Effective and Efficient Approach to Quantized Neural Networks
	Zhou, Shu-Chang 1,2,3; Wang, Yu-Zhi 3,4; Wen, He 3,5; He, Qin-Yao 3,5; Zou, Yu-Heng 5,6
	2017-07-01
发表期刊	JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY
ISSN	1000-9000
卷号	32 期号:4 页码:667-682
摘要	Quantized neural networks (QNNs), which use low bitwidth numbers for representing parameters and performing computations, have been proposed to reduce the computation complexity, storage size and memory usage. In QNNs, parameters and activations are uniformly quantized, such that the multiplications and additions can be accelerated by bitwise operations. However, distributions of parameters in neural networks are often imbalanced, such that the uniform quantization determined from extremal values may underutilize available bitwidth. In this paper, we propose a novel quantization method that can ensure the balance of distributions of quantized values. Our method first recursively partitions the parameters by percentiles into balanced bins, and then applies uniform quantization. We also introduce computationally cheaper approximations of percentiles to reduce the computation overhead introduced. Overall, our method improves the prediction accuracies of QNNs without introducing extra computation during inference, has negligible impact on training speed, and is applicable to both convolutional neural networks and recurrent neural networks. Experiments on standard datasets including ImageNet and Penn Treebank confirm the effectiveness of our method. On ImageNet, the top-5 error rate of our 4-bit quantized GoogLeNet model is 12.7%, which is superior to the state-of-the-arts of QNNs.
关键词	quantized neural network percentile histogram equalization uniform quantization
DOI	10.1007/s11390-017-1750-y
收录类别	SCI
语种	英语
WOS研究方向	Computer Science
WOS类目	Computer Science, Hardware & Architecture ; Computer Science, Software Engineering
WOS记录号	WOS:000405580700002
出版者	SCIENCE PRESS
引用统计	被引频次：60[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/2XEOYT63/7021
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Zhou, Shu-Chang
作者单位	1.Univ Chinese Acad Sci, Beijing 100049, Peoples R China 2.Chinese Acad Sci, Inst Comp Technol, State Key Lab Comp Architecture, Beijing 100190, Peoples R China 3.Megvii Inc, Beijing 100190, Peoples R China 4.Tsinghua Univ, Dept Elect Engn, Beijing 100084, Peoples R China 5.Tsinghua Univ, Dept Comp Sci & Technol, Beijing 100084, Peoples R China 6.Peking Univ, Sch Elect Engn & Comp Sci, Beijing 100871, Peoples R China
推荐引用方式 GB/T 7714	Zhou, Shu-Chang,Wang, Yu-Zhi,Wen, He,et al. Balanced Quantization: An Effective and Efficient Approach to Quantized Neural Networks[J]. JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY,2017,32(4):667-682.
APA	Zhou, Shu-Chang,Wang, Yu-Zhi,Wen, He,He, Qin-Yao,&Zou, Yu-Heng.(2017).Balanced Quantization: An Effective and Efficient Approach to Quantized Neural Networks.JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY,32(4),667-682.
MLA	Zhou, Shu-Chang,et al."Balanced Quantization: An Effective and Efficient Approach to Quantized Neural Networks".JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY 32.4(2017):667-682.