Effects of data complexity on the intelligent diagnostic reasoning

Abstract

The objective was to train several Artificial Neural Networks (ANNs) with different training functions in order to gain an understanding of the effect of dataset complexity on performance. The utilization of varying training functions permitted ANN diversity; and allowing for enhanced diagnostic reasoning in classification. This improvement is achieved by expediting training stage, calibrating classification. The proposed technique is applied to a number of dataset to verify performance improvements. Particular application of the proposed technique is demonstrated by applying methodology for medical diagnostics.