A framework for executing multiple computational intelligent programs

Abstract

Computational intelligent programs are capable of discovering interesting relationships contained in "raw" data. These programs, including artificial neural networks, set covering algorithms and decision trees, have been successfully used to address a number of real-world problems in, among others, the retail, medical, financial and educational fields. A computationally intelligent program can be very effective and useful, given that the learning problems are sufficiently narrowly defined and the data set contains a distribution of attributes favoured by the program. Many complex real-world problems, however, pose learning problems which cannot effectively be solved by a single program. These problems may be successfully addressed by using a combination of computational intelligent programs. A framework, which combines computational intelligent programs into a computational network, is presented. Employing more than one program potentially leads to more powerful and versatile results.