Extracting Refined Rules from Knowledge-Based Neural Networks
Machine Learning
Hi-index | 0.00 |
In the recent years intelligent systems proved to be a useful and successful tool which has often been used for decision support, data mining and knowledge discovery in medicine. The ability of decision support systems enabling early and accurate diagnosing of various diseases is of vital importance in all fields of medicine. Data mining and knowledge discovery enable to prove existing hypothesis or to generate new ones - thus they in longer term support the advancement of medicine as a scientific discipline and as a consequence developing new treatment, prediction and diagnosing methods directly helping clinicians. However, in many cases the proposed treatment, prediction or diagnose can differ from one intelligent system to another as in real world where different medical specialists can have different opinions. Even more often, specialists' opinions complement one another and joined together form better solution. This paper presents a novel idea for combining different specialists' opinions in a form of intelligent systems with a purpose to improve the accuracy of final decision. We empirically show that self-organization ability and voting of classification cellular automata together with many various methods for classifier construction used to cover as many different aspect of a problem as possible, can be used to improve a search for an optimal solution.