Comparison of neural models, off-line and on-line learning algorithms for a benchmark problem
This papcr comparcs thc application of diffcrcnt ncural modcls-multilaycr pcrccptrons, radial basis functions and B-splincs - for a bcnchmark problem, and illustrates the applicability of a common learning algorithm for all models considered. The learning algorithm is employed both for off-line trai...
| Autor principal: | |
|---|---|
| Formato: | bookPart |
| Idioma: | eng |
| Publicado em: |
2013
|
| Texto completo: | http://hdl.handle.net/10400.1/2290 |
| País: | Portugal |
| Oai: | oai:sapientia.ualg.pt:10400.1/2290 |
| Resumo: | This papcr comparcs thc application of diffcrcnt ncural modcls-multilaycr pcrccptrons, radial basis functions and B-splincs - for a bcnchmark problem, and illustrates the applicability of a common learning algorithm for all models considered. The learning algorithm is employed both for off-line training and for on-line model adaptation. In the latter case. a sliding window of past learning data is employed. |
|---|