An improved boosting scheme based ensemble of Fuzzy Neural Networks for nonlinear time series prediction

This paper proposed a Modified AdaBoostRT (AdaBoost Regression and Threshold) algorithm based on Fuzzy Neural Networks (FNNs) and its application to the accurate prediction of complex nonlinear time-series. The algorithm is validated by using four typical time-series data, namely Lorenz, Mackey-Glass, Sunspot and Dow Jones Indices data. The performance comparison of the proposed method and several existing approaches is also performed to show its advantages for nonlinear time series prediction problems.