中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 计算技术研究所
  3. 中国科学院计算技术研究所
  4. 中国科学院计算技术研究所期刊论文
  5. 英文
LINAC: A Spatially Linear Accelerator for Convolutional Neural Networks

文献类型：期刊论文

作者Xiao, Hang1,2; Xu, Haobo1,2; Wang, Ying1,2,3; Wang, Yujie1,2; Han, Yinhe1,2,3
刊名IEEE COMPUTER ARCHITECTURE LETTERS
出版日期2022
卷号21期号:1页码:29-32
关键词Linear particle accelerator Correlation Kernel Convolution Linear regression System-on-chip Quantization (signal) Neural network acceleration convolution linear regression bit-sparsity
ISSN号1556-6056
DOI10.1109/LCA.2022.3167465
英文摘要We first propose a method to utilize the spatially linear correlation between activations and weights of CNN models and accelerate the inference process via linear regression. With the proposed linear regression-based acceleration method, stronger bit-sparsity is excavated to reduce ineffectual computations during the inference in CNN acceleration systems without introducing any errors to the algorithm. We also propose the corresponding hardware accelerator based on the linear convolution mechanism: LINAC, which implements the linear convolution that exploits bit-sparsity to boost performance and energy efficiency significantly. In our experiments, LINAC boosts the inference performance by 11x on average over the present value-agnostic accelerator and outperforms other state-of-the-art bit-sparse accelerators.
资助项目National Natural Science Foundation of China (NSFC)[61834006] ; National Natural Science Foundation of China (NSFC)[62025404] ; National Natural Science Foundation of China (NSFC)[61876173] ; National Natural Science Foundation of China (NSFC)[62104229] ; Strategic Priority Research Program of Chinese Academy of Sciences[XDC05030100] ; Zhejiang Lab[2021PC0AC01]
WOS研究方向Computer Science
语种英语
WOS记录号WOS:000790807700003
出版者IEEE COMPUTER SOC
源URL[http://119.78.100.204/handle/2XEOYT63/19632]  
专题中国科学院计算技术研究所期刊论文_英文
通讯作者Wang, Ying; Han, Yinhe
作者单位1.Univ Chinese Acad Sci, Beijing 100049, Peoples R China
2.Chinese Acad Sci, Inst Comp Technol, Beijing 100190, Peoples R China
3.Zhejiang Lab, Hangzhou 311100, Zhejiang, Peoples R China
推荐引用方式
GB/T 7714		 Xiao, Hang,Xu, Haobo,Wang, Ying,et al. LINAC: A Spatially Linear Accelerator for Convolutional Neural Networks[J]. IEEE COMPUTER ARCHITECTURE LETTERS,2022,21(1):29-32.
APA Xiao, Hang,Xu, Haobo,Wang, Ying,Wang, Yujie,&Han, Yinhe.(2022).LINAC: A Spatially Linear Accelerator for Convolutional Neural Networks.IEEE COMPUTER ARCHITECTURE LETTERS,21(1),29-32.
MLA Xiao, Hang,et al."LINAC: A Spatially Linear Accelerator for Convolutional Neural Networks".IEEE COMPUTER ARCHITECTURE LETTERS 21.1(2022):29-32.

入库方式： OAI收割

来源：计算技术研究所

浏览0
下载0
收藏0
其他版本

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。