Recurrent dendritic neuron model artificial neural network for time series forecasting

Egrioglu E., Baş E., Chen M.

Information Sciences, vol.607, pp.572-584, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 607
  • Publication Date: 2022
  • Doi Number: 10.1016/j.ins.2022.06.012
  • Journal Name: Information Sciences
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Applied Science & Technology Source, Business Source Elite, Business Source Premier, Communication Abstracts, Computer & Applied Sciences, INSPEC, Library, Information Science & Technology Abstracts (LISTA), Metadex, MLA - Modern Language Association Database, zbMATH, Civil Engineering Abstracts
  • Page Numbers: pp.572-584
  • Keywords: Forecasting, Recurrent Artificial Neural Networks, Dendritic Neuron Model, M3 and M4 competitions
  • Karadeniz Technical University Affiliated: No

Abstract

© 2022 Elsevier Inc.Various neuron models have been proposed in the literature. Their structures are the simplest imitation for the biological neuron models. The dendritic neuron model is closer to biological neuron than the others. The single dendritic neuron model artificial neural networks have produced good forecasting results in the literature. In this study, a new recurrent dendritic neuron model artificial neural network is proposed. Moreover, a training algorithm based on particle swarm optimization is proposed. The performance of the proposed method is examined on SP500 stock exchange data sets. The proposed method produces better forecasts than long short term memory and Pi-Sigma artificial neural network. Moreover, the forecasting performance of the proposed method is investigated by using M3 and M4 competitions yearly data, totally used 23645-time series. According to application results, the proposed method produces very competitive results and it is the second-best method for M3 competition yearly data and it is the fifth-best method for M4 competition yearly data. Moreover, it is the best machine learning method for both M3 and M4 performance.