PERFORMANCE EVALUATION OF A SEQUENTIAL MINIMAL RADIAL BASIS FUNCTION (RBF) NEURAL-NETWORK LEARNING ALGORITHM

This paper presents a detailed performance analysis of the recently de veloped minimal resource allocation network (M-RAN) learning algorithm . M-RAN is a sequential learning radial basis function neural network which combines the growth criterion of the resource allocating network (RAN) of Platt with a pruning strategy based on the relative contribu tion of each hidden unit to the overall network output. The resulting network leads toward a minimal topology for the RAN. The performance o f this algorithm is compared with the multilayer feedforward networks (MFN's) trained with 1) a variant of the standard back-propagation alg orithm, known as RPROP and 2) the dependence identification (DI) algor ithm of Moody and Antsaklis on several benchmark problems in the funct ion approximation and pattern classification areas. For all these prob lems, the M-RAN algorithm is shown to realize networks with far fewer hidden neurons with better or same approximation/classification accura cy. Further, the time taken for learning (training) is also considerab ly shorter as M-RAN does not require repeated presentation of the trai ning data.