This paper introduces a hybrid neural structure using radial-basis function (RBF) and multilayer perceptron (MLP) networks. The hybrid network is composed of one RBF network and a number of MLPs, and is trained using a combined genetic/ unsupervised/ supervised learning algorithm. Genetic and unsupervised learning algorithms are used to locate centres of the RBF part in the hybrid network. In addition, supervised learning algorithm, based on backpropagation algorithm, is used to train connection weights of the MLP part in the hybrid network. Performance of the hybrid network is initially tested using the two-spiral benchmark problem. Several simulation results are reported for applying the algorithm in the classification of Electromyogram signals (EMG) where the genetic-based network proved most efficient.
展开▼
