Model Selection of RBF Networks Via Genetic Algorithms

AUTOR(ES)

Estefane George Macedo de Lacerda

DATA DE PUBLICAÇÃO

2003

RESUMO

One of the main obstacles to the widespread use of artificial neural networks is the difficulty of adequately defining values for their free parameters. This work discusses how Radial Basis Function (RBF) neural networks can have their free parameters defined by Genetic Algorithms (GAs). For such, it firstly presents an overall view of the problems involved and the different approaches used to genetically optimize RBF networks. It also proposes a genetic algorithm for RBF networks with a nonredundant genetic encoding based on clustering methods. Secondly, this work addresses the problem of finding the adjustable parameters of a learning algorithm via GAs. This problem is also known as the model selection problem. Some model selection techniques (e.g., crossvalidation and bootstrap) are used as objective functions of the GA. The GA is modified in order to adapt to that problem by means of occamâs razor, growing, and other heuristics. Some modifications explore features of the GA, such as the ability for solving multiobjective optimization problems and handling objective functions with noise. Experiments using a benchmark problem are performed and the results achieved, using the proposed GA, are compared to those achieved by other approaches. The proposed techniques are quite general and may also be applied to a large range of learning algorithms

ASSUNTO(S)

rbf networks via genetic algorithms - model selection ciencia da computacao

Documentos Relacionados

• Projeto Ótimo de redes de distribuição de água via algoritmos genéticos multiobjetivos
• Optimization of laminated composite plates and shells using genetic algorithms, neural networks and finite elements
• Concurrent blind channel equalization with phase transmittance rbf neural networks
• Rehabilitation analysis of the water distribution networks by multiobjective genetic algorithms
• A Frequency-Dependent Natural Selection Model for the Evolution of Social Cooperation Networks