| Autor: |
Beig, R., Ehlers, J., Frisch, U., Hepp, K., Jaffe, R. L., Kippenhahn, R., Ojima, I., Weidenmüller, H. A., Wess, J., Zittartz, J., Beiglböck, W., Eisenächer, Monika, Clark, John W., Lindenau, Thomas, Ristig, Manfred L., Gernoth, Klaus A. |
| Zdroj: |
Scientific Applications of Neural Nets; 1999, p139-169, 31p |
| Abstrakt: |
This article surveys modeling and prediction of various nuclidic properties with feedforward artificial neural networks. Special emphasis is placed on neural network modeling of nuclear ground state masses, the cleanprop algorithm for training neural networks on data with error bars, global neural network models of nuclear stability and branching ratios of decay of unstable nuclides, and higher-order probabilistic perceptrons for classifying nuclides as stable or unstable. The various network architectures and training algorithms devised for and successfully tested in these applications are discussed in detail and the best of numerical neural network results are presented. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Supplemental Index |
| Externí odkaz: |
|
