Regularization and neural computation: application to aerial images analysis

摘要

The main feature of this paper is to show that the key point of two different problems tackled by neural approaches-pairing pattern and function approximationlies in the choice of the regulatization term in the function which is minmized by the neural approach. After the description of links between approximation of functions, combinatorial optimization and regularization theory, we describe a neural approach for solving the hard problem of change detection in aerial images.