This article presents a new learning algorithm, CO-RBFNN, for complex classifications, which attempts to construct the radial basis function neural network (RBFNN) models by using a cooperative coevolutionary algorithm (Co-CEA). The Co-CEA utilizes a divide-and-cooperative mechanism by which subpopulations are coevolved in separate populations of evolutionary algorithms executing in parallel. A modified K-means method is employed to divide the initial hidden nodes into modules that are represented as subpopulation of the Co-CEA. Collaborations among the modules are formed to obtain complete solutions. The algorithm adopts a matrix-form mixed encoding to represent the RBFNN hidden layer structure, the optimum of which is achieved by coevolving all parameters. Experimental results on eight UCI datasets illustrate that CO-RBFNN is able to produce a higher accuracy of classification with a much simpler network structure in fewer evolutionary trials when compared with other alternative standard algorithms. © Springer-Verlag Berlin Heidelberg 2007.
CITATION STYLE
Tian, J., Li, M., & Chen, F. (2007). A cooperative coevolution algorithm of RBFNN for classification. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 4426 LNAI, pp. 809–816). Springer Verlag. https://doi.org/10.1007/978-3-540-71701-0_89
Mendeley helps you to discover research relevant for your work.