Optimizing RBF Networks with Cooperative/Competitive Evolution of Units and Fuzzy Rules

Authors:
Antonio Jesús Rivera Rivas;Julio Ortega;Ignacio Rojas;Alberto Prieto
Affiliations:
-;-;-;-
Venue:
IWANN '01 Proceedings of the 6th International Work-Conference on Artificial and Natural Neural Networks: Connectionist Models of Neurons, Learning Processes and Artificial Intelligence-Part I
Year:
2001

Citing 8
Cited 1

The cascade-correlation learning architecture

Advances in neural information processing systems 2
A resource-allocating network for function interpolation

Neural Computation
Combining evolution with credit apportionment: a new learning algorithm for neural nets

Neural Networks
A sequential learning scheme for function approximation using minimal radial basis function neural networks

Neural Computation
Implicit niching in a learning classifier system: Nature's way

Evolutionary Computation
Fast learning in networks of locally-tuned processing units

Neural Computation
Fitness sharing and niching methods revisited

IEEE Transactions on Evolutionary Computation
Cooperative-competitive genetic evolution of radial basis function centers and widths for time series prediction

IEEE Transactions on Neural Networks

Technical data mining with evolutionary radial basis function classifiers

Applied Soft Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents a new evolutionary method to design optimal networks of Radial Basis Functions (RBFs). The main characteristics of this method lie in the estimation of the fitness of a neurone in the population and in the choice of the operator to apply; for this latter objective, a set of fuzzy rules is used. Thus, the estimation of the fitness considered here, is done by considering three main factors: the weight of the neuron in the RBF Network, the overlapping among neurons, and the distances from neurons to the points where the approximation is worst. These factors allow us to define a fitness function in which concepts such as cooperation, speciation, and niching are taken into account. These three factors are also used as linguistic variables in a fuzzy logic system to choose the operator to apply. The proposed method has been tested with the Mackey-Glass series.