RBF Neural Network Learning Algorithm Based on Error Projection and Local Projection

宋绍云,张玉忠,沐士光
DOI: https://doi.org/10.3969/j.issn.1009-9506.2009.04.010
2009-01-01
Abstract:A new RBF network learning algorithm: error projection and local projection algorithm is proposed.The new algorithm can not only reduce the number of adjusting parameters significantly,but also predict the accuracy of network structure and the change in the overall performance of the algorithm through the adjustment of parameters.The result of MATLA simulation shows that the new algorithm is superior to M-RAN algorithm,especially in the setting of LPN critical threshold value.Its intelligent self-regulating ability makes the setting of initial parameters simpler and more transparent.
What problem does this paper attempt to address?