Refinement of Embedding Parameters of Nonlinear Time Series by Neural Network with Fuzzy-Controlled Regularizer

Abstract

This work proposes a refinement scheme of estimation of optimal embedding parameters of a nonlinear time series by a feed-forward neural network with Fuzzy-Controlled Regularizer(FCR). Although the estimation of embedding parameters using neural network has been proposed by authors earlier, learning procedure was heavy and time consuming. Here we propose an adaptive regularization technique by fuzzy reasoning and added it to the learning procedure for refinement task. From the simulation results, it has been found that the proposed scheme can refine the estimation of embedding parameters successfully and takes lesser time in learning of the neural network than our previously proposed method.