An adaptive solution to the mixture problem with drift spectra

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.