In this paper, we propose a fuzzy neural network based on immune feedback learning (FNNBIFL) for the availability classifier of satellite images, which accelerates the learning speed, solves the problem of being trapped in the local minimum and improves the learning performance of fuzzy neural network. Using 122 satellite images, we compare the recognition results of the availability classifier trained by FNNBIFL or the traditional BP algorithm. Those results show that the recognition errors of FNNBIFL are reduced by 5.1%, and its learning speed is improved by 55%.
Research on availability of satellite images based on immune feedback learning fuzzy neural network
2009
4 Seiten, 6 Quellen
Aufsatz (Konferenz)
Englisch
Training a neural network to detect jet contrails in satellite images
British Library Conference Proceedings | 1994
|Training a neural network to detect jet contrails in satellite images
British Library Conference Proceedings | 1994
|