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.
出处
《江南大学学报(自然科学版)》
CAS
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)
关键词
激活函数
随机算法
状态机
硬件描述语言
activation function, stochastic arithmetic, state machines, VHDL