Abslracl-In this paper, a new architecture of divide-andconquer based radial basis function network (DCRBF) and its learning algorithm are presented. The DCRBF network is a :hybrid system consisting of several sub-RBF networks, each of which individually takes a sub-input space as its input. The output of this new. architecture is a linear combination of the sub-networks' outputs with the coefficients tuned together with each snb-netwnrk system parameters. Since this system divides a high-dimensional modelling problem into several low-dimensional ones, it can considerably reduce the structural complexity of a RBF network, whereby the net's learning speed as a whole is significantly improved with the comparable generalization ability. We apply DCRBF to model a recurrent version of RBF networks. The experimental results have shown its outstanding performance.
Abstract:This paper experimentally investigates Independent Component Analysis (ICA) and Principle Component Analysis (PCA) on reducing the input dimension of a Radial Basis Function (RBF) network such that the net's complexity is reduced. The results have shown that a RBF network with ICA as an input pre-process has the similar generalization ability to the one without pre-processing, but the former's performance converges much faster. In contrast, a PCA based RBF however leads to a deteriorated result in both of convergent speed and the generalization ability.
From the dual structural radial basis function network (DSRBF) (Cheung and Xu 2001), this paper presents a new divide-and-conquer learning approach to radial basis function networks (DCRBF). The DCRBF network is a hybrid system consisting of several sub-RBF networks, each of which takes a sub-input space as its input. Since this system divides a high-dimensional modeling problem into serveral low-dimensional ones, it can considerably reduce the structural complexity of a RBF network, whereby the net's learning is much faster. We have experimentally shown its outstanding learning performance on forecasting two real time series as well as synthetic data in comparison with a conventional RBF one.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.