Abstract
Wavelet neural networks employing wavelets as the activation functions recently have been researched as an alternative approach to the traditional neural networks with sigmoidal activation functions. In this paper, we proposed a new type of wavelet neural network by introducing local linear models, which are used in some neuro-fuzzy systems, as powerful weights instead of straightforward weights employed in the previous wavelet neural networks. The proposed network is called the local linear adaptive wavelet neural network. Its effectiveness is examined by the network performances on function approximation and chaotic time series prediction problems. In these experiments, the proposed local linear adaptive wavelet neural network performed well and compared favorably to the previous wavelet neural network.
