Title :
Learning the architecture and parameters of RBF network based on hybrid IPL algorithm
Author :
Zhu, Gen-Biao ; Zhang, Feng-Ming ; Wang, Jin-Gan ; Shi, Jun-Yong
Author_Institution :
Coll. of Eng., Air Force Eng. Univ., Xi´´an, China
Abstract :
A new method of constructing RBF network based on hybrid incremental projection-learning algorithm (HIPLA) is presented. The method substitutes aspiration criteria for original error to simplify network architecture and improve its approximation. It only needs a small number of sampling data, and its training speed is higher than the traditional one. The computer simulation results show that the output of the system is accurate.
Keywords :
learning (artificial intelligence); radial basis function networks; RBF network architecture optimization; hybrid incremental projection-learning algorithm; radial basis function network; Approximation algorithms; Computer architecture; Computer errors; Computer simulation; Equations; Function approximation; Kernel; Radial basis function networks; Sampling methods; Vectors; Aspiration criteria; Background theory; Hybrid IPL algorithm (HIPLA); RBF Network Architecture optimization;
Conference_Titel :
Machine Learning and Cybernetics, 2005. Proceedings of 2005 International Conference on
Conference_Location :
Guangzhou, China
Print_ISBN :
0-7803-9091-1
DOI :
10.1109/ICMLC.2005.1527441
