An adaptive feedforward artificial neural network with the applications to multiple criteria decision making

An adaptive feedforward artificial neural network (FANN) is presented and used to and solve multiple criteria decision making (MCDM) problems. Given a FANN topology, an adaptive learning algorithm to adjust its parameters is presented. The algorithm consists of two steps: generating a direction, and performing a one-dimensional search along that direction. A methodology to obtain the most appropriate topology and assess its parameters, using the adaptive learning algorithm as a subroutine, is developed. When the set of training patterns changes, the FANN model can adapt itself to the changes by retraining or expanding the existing model. The theoretical basis for the adaptive FANN and some examples are included. How to formulate MCDM problems and use the adaptive FANN to select the best alternative when each alternative is associated with a set of conflicting and noncommensurate criteria is shown. It is demonstrated how to incorporate basic MCDM definitions and properties into FANN