神经网络传递函数sigmoid函数及其导数的fpga实现我要分享

FPGA implementation of neural network transfer function sigmoid function and its derivatives

关注次数: 419

下载次数: 0

文件大小: 2.03 kB

代码分类: 其他

开发平台: matlab

下载需要积分: 2积分

版权声明:如果侵犯了您的权益请与我们联系,我们将在24小时内删除。

代码描述

中文说明:用查找表的方式实现了sigmoid函数的verilog实现,sigmoid函数的特点是关于(0,0.5)奇对称,只用matlab求出自变量大于0的函数值,小于0的函数值可以由1-f(x)得到。导函数关于y轴偶对称,由f(x)乘以f(-x)即可得到。使用matlab生成mif文件,再调用quartus2中的rom核,把mif文件添加进去,仿真即可.


English Description:

The sigmoid function is characterized by (0,0.5) odd symmetry. We only use matlab to calculate the function value of the independent variable greater than 0, and the function value less than 0 can be obtained by 1-f (x). The derivative function is even symmetric about the Y axis, which can be obtained by multiplying f (x) by F (- x). Use matlab to generate MIF file, and then call the ROM core in quartus2 to add MIF file to the simulation


代码预览

相关推荐