Hierarchical Radial Basis Function Neural Networks for Classification Problems

Abstract

Abstract. Hierarchical neural networks consist of multiple neural networks assembled in the form of an acyclic graph. The purpose of this study is to identify the hierarchical radial basis function neural networks and select important input features for each sub-RBF neural network automatically. Based on the pre-defined instruction/operator sets, a hierarchical RBF neural network can be created and evolved by using treestructure based evolutionary algorithm. This framework allows input variables selection, over-layer connections for the various nodes involved. The HRBF structure is developed using an evolutionary algorithm and the parameters are optimized by particle swarm optimization algorithm. Empirical results on benchmark classification problems indicate that the proposed method is efficient.

Topology and simulations of a Hierarchical Markovian Radial Basis Function Neural Network classifier

“…The flexible hierarchical RBF Neural Network model was proposed by Chen and co-workers [8,9], also called flexible neural tree [8,10] from which it was originated. This model is consisted of multiple neural networks assembled in the form of an acyclic graph.…”

“…Popular algorithms [9] select the RBF centers randomly, or perform supervised center selection or employ unsupervised clustering for center selection, which is the most common strategy. Clustering algorithms such as k-means, fuzzy c-means or Subtractive Clustering are all well-documented for this RBF training stage.…”

ICA Based on KPCA and Hierarchical RBF Network for Face Recognition

“…Because the pattern is complex and nonlinear, hierarchical radial basis function (HRBF) classification model are proposed for face recognition [12,13].…”

“…Recently, RBF neural network has been used in many engineering and scientific applications including face recognition [7]. Hierarchical RBF network consist of multiple RBF networks assembled in different level or cascade architecture, and has been proved effective than single RBF neural network [12,13]. In this paper, an automatic method for constructing HRBF network is proposed to identify the faces.…”