A comprehensive survey on pi-sigma neural network for time series prediction

• Main Authors: Akram, Urooj, Ghazali, Rozaida, Mushtaq, Muhammad Faheem
• Format: Article
• Language: English
• Published: Universiti Teknikal Malaysia Melaka 2017
• Subjects: QA273-280 Probabilities. Mathematical statistics
• Online Access: http://eprints.uthm.edu.my/4414/1/AJ%202018%20%28763%29%20A%20comprehensive%20survey%20on%20pi-sigma%20neural%20network%20for%20time%20series%20prediction.pdf; http://eprints.uthm.edu.my/4414/

Prediction of time series grabs received much attention because of its effect on the vast range of real life applications. This paper presents a survey of time series applications using Higher Order Neural Network (HONN) model. The basic motivation behind using HONN is the ability to expand the input space, to solve complex problems it becomes more efficient and perform high learning abilities of the time series forecasting. Pi-Sigma Neural Network (PSNN) includes indirectly the capabilities of higher order networks using product cells as the output units and less number of weights. The goal of this research is to present the reader awareness about PSNN for time series prediction, to highlight some benefits and challenges using PSNN. Possible fields of PSNN applications in comparison with existing methods are presented and future directions are also explored in advantage with the properties of error feedback and recurrent networks.