Academic literature on the topic 'Radial basis function networks (RBFNs)'

The application of statistical physics to the study of the learning curves of feedforward connectionist networks has to date been concerned mostly with perceptron-like networks. Recent work has extended the theory to networks such as committee machines and parity machines, and an important direction for current and future research is the extension of this body of theory to further connectionist networks. In this article, we use this formalism to investigate the learning curves of gaussian radial basis function networks (RBFNs) having fixed basis functions. (These networks have also been called generalized linear regression models.) We address the problem of learning linear and nonlinear, realizable and unrealizable, target rules from noise-free training examples using a stochastic training algorithm. Expressions for the generalization error, defined as the expected error for a network with a given set of parameters, are derived for general gaussian RBFNs, for which all parameters, including centers and spread parameters, are adaptable. Specializing to the case of RBFNs with fixed basis functions (basis functions having parameters chosen without reference to the training examples), we then study the learning curves for these networks in the limit of high temperature.

This paper investigates the capabilities of radial basis function networks (RBFN) and kernel neural networks (KNN), i.e. a specific probabilistic neural networks (PNN), and studies their similarities and differences. In order to avoid the huge amount of hidden units of the KNNs (or PNNs) and reduce the training time for the RBFNs, this paper proposes a new feedforward neural network model referred to as radial basis probabilistic neural network (RBPNN). This new network model inherits the merits of the two old odels to a great extent, and avoids their defects in some ways. Finally, we apply this new RBPNN to the recognition of one-dimensional cross-images of radar targets (five kinds of aircrafts), and the experimental results are given and discussed.

A radial basis function neural networks (RBFNs) mobile robot control system is automatically developed with the image processing and learned by the bacterial foraging particle swarm optimization (BFPSO) algorithm in this paper. The image-based architecture of robot model is self-generated to travel the routing path in the dynamical and complicated environments. The visible omni-directional image sensors capture the surrounding environment to represent the behavior model of the mobile robot system. Three parameterize RBFNs model with the centers and spreads of each radial basis function, and the connection weights to solve the mobile robot path traveling and routing problems. Several free parameters of radial basis functions can be automatically tuned by the direct of the specified fitness function. In additional, the proper number of radial basis functions of the constructed RBFNs can be chosen by the defined fitness function which takes this factor into account. The desired multiple objectives of the RBFNs control system are proposed to simultaneously approach the shorter path and avoid the unexpected obstacles. Evaluations of PSO and BFPSO show that the developed RBFNs robot systems skip the obstacles and efficiently achieve the desired targets as soon as possible.