期刊文献+

一种神经网络的可配置激活函数模块设计 被引量:2

Implementation of Configurable Activation Function Module Based on Neural Network
下载PDF
导出
摘要 神经网络是机器学习领域的一个重要分支,因其优秀的性能在很多基于嵌入式设备的应用中大放异彩。尽管已经有很多研究已经成功地将神经网络算法移植到小型化平台中,但是大多数的方法依然是基于某个固定模型的设计。本文提出了一种可配置的激活函数模块,比较了不同拟合方法的误差和耗费的资源,选择使用分段四阶多项式拟合激活函数的方法,拟合了6种常用激活函数,可以满足绝大多数神经网络的激活部分。这样在设计可配置神经网络的芯片时,通过配置该模块即可满足不同神经网络的需求。 Neural network is an important branch of machine learning,it is used in many embedded applications.Although many studies have successfully transplanted neural network algorithms into miniaturized platforms,most of them are still based on a fixed model.In the paper a configurable activation function module is proposed,compares the error and resource cost of different fitting methods,chooses the method of piecewise fourth-order polynomial fitting activation function,and fitting six commonly used activation functions.It can satisfy the activation part of most neural networks.In this way,when designing chips that can be configured with neural network,the modules can meet the requirements of different neural networks.
作者 苏潮阳 应三丛 Su Chaoyang;Ying Sancong(National Key Laboratory of Fundamental Science on Synethetic Vision,College of Computer,Sichuan University,Chengdu 610065,China)
出处 《单片机与嵌入式系统应用》 2020年第4期6-9,共4页 Microcontrollers & Embedded Systems
关键词 ZC706 FPGA 激活函数 神经网络 多项式逼近 ZC706 FPGA activation function neural networks polynomial approximation
  • 相关文献

参考文献5

二级参考文献91

  • 1张炜,杨虎,张尔扬.一种改进的混合预测冗余CORDIC算法[J].系统工程与电子技术,2006,28(5):779-782. 被引量:2
  • 2VOLDER J E.The CORDIC trigonometric computing technique[J].IRETrans.Electron.Comput.,1959,8(3):330-334.
  • 3JUANG T B.Low latency angle recoding methods for the higher bit-width parallel CORDIC rotator implementations[J].IEEE Transactions on Circuits and Systems-II:Express Briefs,2008,55(11):1139-1143.
  • 4WANG S Y,VINCENZO P,SWARZLANDER E E.Hybrid CORDIC algorithms[J].IEEE Trans.on Computers,1997,46(11):1202-1207.
  • 5FU D W,WILLSON A N.A two-stage angle-rotation architecture and its error analysis for efficient digital mixer implementation[J].IEEE Trans.on Circuits and Systems,2006,53(3):604-614.
  • 6ANTELO E,LANG T,BRUGUERA J D.Very-high radix circular CORDIC:vectoring and unified rotation/vectoring[J].IEEE Trans.on Computers,2000,49(7):727-739.
  • 7TIMMERMAN D,HAHN H,HOSTICHKA B J.Low latency time CORDIC algorithms[J].IEEE Trans.on Computers,1992,41(8):1010-1014.
  • 8TAKAGI N,ASADA T,YAJIMA S.Redundant CORDIC methods with a constant scale factor for sine and cosine computation[J].IEEE Trans.on Computers,1991,40(9):989-995.
  • 9DAWID H,MEYR H.The differential CORDIC algorithm:Constant scale factor redundant implementation without correcting iterations[J].IEEE Trans.Computers,1996,45(3):307-318.
  • 10CARO D D,PATA N,GIUSEPPE M S A.A 380 MHz direct digital synthesiizer/mixer with hybrid CORDIC architecture in 0.25μm CMOS[J].IEEE J.Solid-State Circuits,2007,42(1):151-160.

共引文献609

同被引文献11

引证文献2

二级引证文献3

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部