| บทคัดย่อ(English) |
Radial Basis Function Neural Network (RBF NN) is one of the mostconsidered neural networks with various data classification and functionapproximation applications. Here, we concern only the data classificationapplications. The classification correctness of an RBF network is definedby the data and the radius of the radial basis function. To cover a classof data, the data must be covered by the same neurons as much as possiblewithout any data from the other classes. However, the correctclassification of the training and testing data may not mean that thenetwork can achieve its generalization. To overcome this, some statisticalmethod for estimating the density distribution of the data must be appliedto correctly adjust the center and radius of the RBF neuron. Bootstraptechnique is considered and applied to estimate the center and size of eachRBF neuron in this thesis. The experimental results show that thistechnique significantly increase the generalization. |
