Title :
An adaptive solution to the mixture problem with drift spectra
Author :
Perez, Rosa M. ; Martinez, P. ; Silva, Alonso
Author_Institution :
Avenida de la Univ., Caceres
Abstract :
In this work we show the development of a robust method for determining and quantifying the components in a composite spectrum obtained from a given mixture of elements. It is assumed that the patterns of the individual spectra belonging to the mixture are known in advance and there are miscalibration problems in the composite spectrum measure. The proposed method is implemented by a linear recurrent neural network based on the Hopfield model (HRNN). The neural model guarantees the convergence of this problem using the gradient method for minimizing errors
Keywords :
Hopfield neural nets; calibration; spectral analysis; Hopfield model; adaptive solution; composite spectrum; convergence; drift spectra; error minimization; gradient method; linear recurrent neural network; miscalibration problems; mixture problem; neural model; robust method; Convergence; Gradient methods; Neural networks; Parallel processing; Recurrent neural networks; Remote sensing; Robustness; Signal processing; Signal processing algorithms; Vectors;
Conference_Titel :
Signal Processing, 1996., 3rd International Conference on
Conference_Location :
Beijing
Print_ISBN :
0-7803-2912-0
DOI :
10.1109/ICSIGP.1996.567103