Interpolation and Approximation of Models
Bruno Pelletier · 2003
A global approach may not be well suited for the approximation of complex and high-dimensional input-output mappings, inadequacy being defined in terms of complexity and/or accuracy. In this paper, a global modelling method from a set of locally optimal simpler models is presented. Theoretical aspects of interpolation and approximation of models are addressed. The method has been applied to the remote sensing of ocean color from space where it has shown good results in comparison with global approaches based on multilayer perceptrons and radial basis functions networks.