The following paper derives a general gradient process for extremization ion by imposing an " exponential decay" condition on the controlled variable. The general process is applied to two adaptive linear filter problems; a sonar application by Widrow [3] and a radar problem of Brennan & Reed [4]. The results are discussed in sections 11 and Ill, respectively. A central feature of the Fletcher-Powell process [1] is a matrix modification algorithm due to Davidon [2]. It is shown in Appendix A that this algorithm may be derived using the same exponential condition imposed on the proper error term.
A General Steepest Descent Algorithm
IEEE Transactions on Aerospace and Electronic Systems ; AES-12 , 1 ; 12-22
1976-01-01
1736131 byte
Article (Journal)
Electronic Resource
English
Steepest Descent Day-to-Day Dynamic Toll
Online Contents | 2007
|Steepest Descent Day-to-Day Dynamic Toll
Transportation Research Record | 2007
|Global Least Squares Matching Using a Steepest Descent Method
British Library Conference Proceedings | 1997
|