Artificial Speciation of Neural Network Ensembles
Vineet R. Khare · 2005
Modular approach of solving a complex problem can reduce the total complexity of the system while solving a difficult problem satisfactorily. To implement this idea, an EANN system is developed here for classifying data. The system evolved is speciated in such a manner that members of a particular species solve certain parts of the problem and complement each other in solving one big problem. Fitness sharing is used in evolving the group of ANNs to achieve the required speciation. Sharing was performed at phenotypic level using modified Kullback-Leibler entropy as the distance measure. Since the group as a unit solves the classification problem, outputs of all the ANNs are used in finding the final output. For the combination of ANN outputs 3 different methods – Voting, averaging and recursive least square are used. The evolved system is tested on two data classification problems (Heart Disease Dataset and Breast Cancer Dataset) taken from UCI machine learning benchmark repository.