Title: Artificial chemical reaction optimisation of recurrent functional link neural networks for efficient modelling and forecasting of financial time series

Authors: Sarat Chandra Nayak

Addresses: Department of Computer Science and Engineering, CMR College of Engineering & Technology, Hyderabad, 501401, India

Abstract: In contrast to multilayer neural networks, a functional link artificial neural network uses functional expansion units for transferring lower input space to higher dimensions. It achieves enhanced discrimination capability through generating hyperplanes in the input space. The feedback properties of recurrent networks made them more proficient and dynamic to model nonlinear systems accurately. This paper develops a recurrent functional link artificial neural network (RFLN)-based forecasting model where the optimal model parameters are efficiently searched with artificial chemical reaction optimisation (ACRO). The reason behind using ACRO is its faster convergence toward optimal solution with fewer tuning parameters. The optimal model is achieved through the process of artificial chemical reaction of potential RFLN structures, therefore termed as ACRRFLN. Also, three other optimisation techniques, i.e., particle swarm optimisation (PSO), teaching learning-based optimisation (TLBO), and genetic algorithm (GA) are employed to train RFLN separately. All the models are experimented and validated on forecasting closing indices of six stock markets. Results from extensive simulations clearly reveal the outperformance of ACRRFLN over other models similarly trained. Further, results from Deibold-Mariano test supported the statistical significance of the proposed model.

Keywords: recurrent neural network; artificial chemical reaction optimisation; ACRO; stock market prediction; recurrent FLANN; particle swarm optimisation; PSO; genetic algorithm; financial time series forecasting.

DOI: 10.1504/IJAAC.2021.118524

International Journal of Automation and Control, 2021 Vol.15 No.6, pp.669 - 691

Received: 30 Jan 2019
Accepted: 17 Feb 2020
Published online: 28 Oct 2021 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article