The learning dynamics of a universal approximator


The learning properties of a universal approximator, a normalized committee machine with adjustable biases, are studied for on-line back-propagation learning. Within a statistical mechanics framework, numerical studies show that this model has features which do not exist in previously studied two-layer network models without adjustable biases, e.g., attractive suboptimal symmetric phases even for realizable cases and noiseless data.

Divisions: Engineering & Applied Sciences > Mathematics
Engineering & Applied Sciences > Systems analytics research institute (SARI)
Engineering & Applied Sciences > Computer Science
Additional Information: Copyright of the Massachusetts Institute of Technology Press (MIT Press)
Uncontrolled Keywords: approximator,back-propagation,symmetric phases,realizable cases,noiseless data
Full Text Link:
Related URLs: http://mitpress ... type=2&tid=3990 (Publisher URL)
http://www.scop ... tnerID=8YFLogxK (Scopus URL)
PURE Output Type: Article
Published Date: 1997-05
Authors: West, Ansgar H. L.
Saad, David ( 0000-0001-9821-2623)
Nabney, Ian T. ( 0000-0003-1513-993X)



Version: Published Version

| Preview

Export / Share Citation


Additional statistics for this record