Pseudo neural networks for Iris data classification

Abstract

This research deals with a novel approach to classification. Iris data was used for the experiments. Classical artificial neural networks, where a relation between inputs and outputs is based on the mathematical transfer functions and optimized numerical weights, was an inspiration for this work. Artificial neural networks need to optimize weights, but the structure and transfer functions are usually set up before the training. The proposed method utilizes the symbolic regression for synthesis of a whole structure, i.e. the relation between inputs and output(s). This paper differs from the previous approach where only one output pseudo node was used even for more classes. In this case, there were synthesized more node output equations as in classical artificial neural networks. The benchmark was iris data as in previous research. For experimentation, Differential Evolution (DE) for the main procedure and also for meta-evolution version of analytic programming (AP) was used. Proceedings 28th European Conference on Modelling and Simulation © ECMS Flaminio Squazzoni, Fabio Baronio, Claudia Archetti, Marco Castellani (Editors).