Relative importance of input signals of neural network

by

Relative importance of input signals of neural network

Marcin Perzyk, Andrzej Kochański, Jacek Kozłowski

Instytut Technogii Materiałowych Politechniki Warszawskiej.

DOI:

https://doi.org/10.7494/cmms.2003.3.0045

Abstract:

Several types of relative importance factors of input signals of neural network are analyzed and assessed using simulated training sets. Behavior of the factors for characteristic cases of synergetic influences of input signals on network output are presented. Difficulties in utilization of the importance factors for detection of casting defects resulting from manufacturing process are discussed.

Cite as:

Perzyk, M., Kochański, A., Kozłowski, J. (2003). Relative importance of input signals of neural network. Computer Methods in Materials Science, 3(3-4), 172 – 179. https://doi.org/10.7494/cmms.2003.3.0045

Article (PDF):

Keywords:

,

References:

Garson, D. (1991). Implementing neural network connection weights. Al Expert, 6 (4), 45-51.

Glorfeld, L.W. (1996). A Methodology for Simplification and Interpretation of Backpropagation-Based Neural Network Models. Expert Systems With Applications, 10 (1), 37-4.

Kochański, A., Perzyk, M. (2002). Identyfikacja przyczyn wady porowatości w odlewach staliwnych z wykorzystaniem sztucznych sieci neuronowych. Archiwum Odlewnictwa, 2, (5), 87-92.

Masters, T. (1993). Practical Neural Network Recipes in C++. Academic Press, 1993.

Perzyk, M., Kochanski A. (2001). Prediction of ductile cast iron quality by artificial neural networks. J. Mat. Proc. Techn., 109 (3), 305-307.

Perzyk, M., Kochański, A. (2002a). Methodology of detection of casting defects causes by neural networks. Mat. Sci., 8, (4), 375-379.

Perzyk, M., Kochański. (2002b). System komputerowy wspomagający projektowanie procesów metalurgicznych wykorzystujący modelowanie sztucznymi sieciami neuronowymi. Rozdział w monografii Polska metalurgia w latach 1998 – 2002, red. Świątkowski, K., tom 2. Wyd. Komitet Metalurgii PAN, Krakow, 236-242.

Perzyk, M., Kochański, A. (2003). Detection of causes of casting defects assisted by artificial neural networks. J. Eng. Manufacture, Proceedings of the Institution of Mechanical Engineers, 217, Part B, 1279-284.