ABSTRACT
In this contribution a new supervised classification model is proposed, namely the Fuzzy Evolutionary Probabilistic Neural Network (FEPNN). The proposed model incorporates a fuzzy class membership function into the recently proposed Evolutionary Probabilistic Neural Network (EPNN). EPNN employs an evolutionary algorithm, namely the Particle Swarm Optimization (PSO), for the selection of the spread parameters and prior probabilities of Probabilistic Neural Networks. FEPNN combines efficient and effective evolutionary algorithms as well as techniques from fuzzy set theory. This combination provides an adequate model that achieves similar and superior performance than the well known and widely used Feed Forward Neural Networks (FNNs). FEPNN is applied to a credit card approval task with promising results.
