Radial basis function neural network model for discontinuous signals
Various Radial Basis Functions (RBFs) are adopted to approximate the discontinuous input data points. Empirical relations among the RBF parameters, input data sets and convergence have been derived and used to develop an adaptive RBF method. These adaptive methods enhance convergence near the discontinuity. The first derivative technique is successfully used to detect discontinuities and it has been used to update the adaptivity criteria. Additionally, this project has so far led to the development of a 2-dimensional RBF image reconstruction toolbox for sharp edged images.