Title :
Neural least-squares design of higher order digital differentiators
Author_Institution :
Dept. of Comput. & Inf. Sci., Mil. Acad., Kaohsiung, Taiwan
Abstract :
The least-squares (LS) design of higher order digital differentiators can be formulated as to solve a system of linear equations. This is completed by solving a matrix inversion or by using some matrix properties to simplify the optimization. Here, a Hopfield neural network (NN) with closed-form evaluation for the related parameters is utilized to solve the mentioned set of linear equations in real-time. Simulation results are given to illustrate the excellent performance of the proposed neural-based method.
Keywords :
FIR filters; Hopfield neural nets; least squares approximations; matrix inversion; optimisation; Hop-field neural network; filter optimization; higher order digital differentiators; linear equations; matrix inversion; neural least-squares design; Biological system modeling; Closed-form solution; Differential equations; Digital signal processing; Finite impulse response filter; Frequency response; Hopfield neural networks; Matrix decomposition; Military computing; Neural networks;
Journal_Title :
Signal Processing Letters, IEEE
DOI :
10.1109/LSP.2004.839693
