Online sliding-window based for training MLP networks using advanced conjugate gradient

Abstract

This paper investigates the performance of conjugate gradient algorithms with sliding-window approach for training multilayer perceptron (MLP). Online learning is implemented when the system under investigation is time varying or when it is not convenient to obtain a full history of offline data about the system variables. Sliding window framework is proposed to combine the robustness of offline learning with the ability of online learning to track time varying elements of the process under investigation. A sliding window based second order conjugate gradient algorithms SWCG is presented. The performance of SWCG is compared with a sliding window based first order back propagation SWBP.