Abstract
Due to their simplicity and small computational complexity gradient-type algorithms play an important role in the field of adaptive filtering and system identification. In practical cases, however, these algorithms often fail to converge to the global minimum when the error surface is multimodal. There are algorithms - known and to be presented here-which try to eliminate this problem by modifying the simple gradient search method in such a way that they do not track the true output error gradients. (Equation Error Algorithm, Steiglitz McBride Method, Composite Algorithms). This paper investigates these algorithms from the viewpoint of the fictitious mean square error surface. This approach leads to a new Composite Gradient Algorithm with dynamic error surface. Illustrative simulation examples are presented showing good global convergence properties of the new algorithm in cases when other methods fail.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Adaptive Systems in Control and Signal Processing 1995
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
