论文部分内容阅读
RBF径向基神经网络在工程中,尤其是各种智能控制中的应用十分广泛。其隐含层的非线性激活函数经常采用高斯径向基函数,这一函数为一指数函数。指数函数用硬件实现起来比较困难。本文提供了一种应用CORDIC算法在FPGA上实现ex的方法。这种方法具有硬件实现简单、误差小和便于VLSI等优点。
Radial RBF neural network in engineering, especially in a variety of intelligent control is widely used. The nonlinear activation function of its hidden layer often uses Gaussian radial basis function, which is an exponential function. Exponential functions are hard to implement in hardware. This article provides a method of using CORDIC algorithm to implement ex on FPGA. This method has the advantages of simple hardware implementation, small error and easy VLSI.