基于VHDL语言的神经网络激活函数随机运算的实现  

Implementation of Activated Function of Neural Networks with Stochastic Arithmetic Theory Based on VHDL

作　　者：初丽娟[1] 沈艳霞[1] 

机构地区：[1]江南大学电气与自动化研究所,江苏无锡214122

出　　处：《江南大学学报（自然科学版）》2009年第2期154-158,共5页Joural of Jiangnan University (Natural Science Edition)　

基　　金：江苏省自然科学基金项目(BK2007540)

摘　　要：为节省硬件实现中的资源数量,基于状态机原理并采用随机运算实现神经网络非线性激活函数的方法,给出了S型激活函数的数字逻辑实现,用硬件描述语言(VHDL)对该算法进行了软件设计与实现,并在ModelsimSE6.2仿真平台上进行了仿真测试。该设计有利于程序的随时修改,可节省大量硬件乘法器,有效缩短设计周期,满足了神经网络超大规模集成电路的需要。The method based on stochastic arithmetic was presented to implement the nonlinear activation function of neural networks for saving hardware resoures. This method was based on the principle of state machines. In the paper,the digital logic implemention of Sigmoid function was given,and then the software was designed in hardware description language VHDL. The simulations based on the platform（Modelsim SE 6.2）were carried out. Using this method,the program can be modified anytime, the development, period can shorten greatly and the hardware resoures can be saved significantly. This meets the neural network VLSI requirements well.

关 键 词：激活函数 随机算法 状态机 硬件描述语言 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]