This paper deals with the possibility of using a feedforward neural network to test the discrepancies between a real astronomical image and a predefined template. This task can be accomplished thanks to the capability of neural networks to solve a nonlinear approximation problem, i.e. to construct an hypersurface that approximates a given set of scattered data couples. Images are encoded associating each of them with some conveniently chosen statistical moments, evaluated along the axes; in this way a parsimonious method is obtained that allows a really effective approach to Seidel aberration diagnostics.
In questo articolo viene studiata la possibilità di usare una rete neurale feedforward per identificare eventuali discrepanze tra un'immagine astronomica reale ed un suo modello predefinito. Questo compito viene affrontato grazie alla capacità delle reti neurali di risolvere un problema di approssimazione non lineare di funzioni attraverso la costruzione di un'ipersuperficie approssimante un insieme dato di punti sparsi. La codifica delle immagini viene effettuata associando ciascuna di esse ad alcuni momenti statistici opportunamente scelti, calcolati relativamente agli assi , ottenendo in tal modo un metodo computazionalmente economico che permette un approccio realmente efficace alla diagnostica delle aberrazioni di Seidel.
@article{BUMI_2004_8_7B_3_687_0,
author = {Rossella Cancelliere and Mario Gai},
title = {Function approximation of Seidel aberrations by a neural network},
journal = {Bollettino dell'Unione Matematica Italiana},
volume = {7-A},
year = {2004},
pages = {687-696},
zbl = {1182.41017},
mrnumber = {2101659},
language = {en},
url = {http://dml.mathdoc.fr/item/BUMI_2004_8_7B_3_687_0}
}
Cancelliere, Rossella; Gai, Mario. Function approximation of Seidel aberrations by a neural network. Bollettino dell'Unione Matematica Italiana, Tome 7-A (2004) pp. 687-696. http://gdmltest.u-ga.fr/item/BUMI_2004_8_7B_3_687_0/
[1] -, Principles of optics, Pergamon, New York, 1985.
[2] -, A Comparative Analysis of Neural Network Performances in Astronomy Imaging, Applied Numerical Mathematics, 45, n. 1 (2003), 87-98. | Zbl 1019.85001
[3] , approximation by superpositions of a sigmoidal function, Math. Control-Signals Systems, 2 (1989), 303-314. | MR 1015670 | Zbl 0679.94019
[4] , Aspects of the numerical analysis of neural networks, Acta Numerica, 1994, 145-202. | MR 1288097 | Zbl 0807.65007
[5] , On the approximate realization of continuous mappings by neural networks, Neural Networks, 2 (1989), 183-192.
[6] ---, Location estimators for interferometric fringe, Publ. Astron. Soc. Pac., 110 (1998), 848-862.
[7] -------, Location accuracy limitations for CCD cameras, A&A, 367 (2001), 362-370.
[8] , Neural Networks, a Comprehensive FoundationIEEE Computer Society Press, 1994. | Zbl 0828.68103
[9] --, Multilayer feedforward networks are universal approximators, Neural Networks, 2 (1989), 359-366.
[10] --------, First Results of an On-line Adaptive Optics System with Atmospheric Wavefront Sensing by an Artificial Neural Network, ApJ, 390 (1992), L41-44.
[11] -, Perceptrons, Cambridge, MA:MIT Press, 1969. | Zbl 0197.43702
[12] --, Learning internal representation by error propagation. Parallel Distribuited Processing (PDP): Exploration in the Microstructure of Cognition, MIT Press, Cambridge, Massachussetts, 1 (1986), 318-362.
[13] --, Adaptive Optics for Array Telescopes Using Neural Networks Techniques on Transputers, Transputing '91, IOS Press, Washington D.C., 1 (1991), 170-183.