Neural networks for long-term earthquake prediction using modified meta-learning

Article type: Research Article

Authors: Maya, Mario^a | Yu, Wen^{a; *} | Telesca, Luciano^b

Affiliations: [a] Departamento de Control Automatico CINVESTAV-IPN (National Polytechnic Institute) Mexico City, Mexico | [b] Institute of Methodologies for Environmental Analysis, National Research Council, Tito (PZ), Italy

Correspondence: [*] Corresponding author. Wen Yu, Departamento de Control Automatico CINVESTAV-IPN (National Polytechnic Institute) Mexico City, Mexico. E-mail: yuw@ctrl.cinvestav.mx.

Abstract: Neural networks have been successfully applied for modeling time series. However, the results of long-term prediction are not satisfied. In this paper, the modified Meta-Learning is applied to the neural model. The normal Meta-Learning is modified by time-varying learning rates and adding a momentum term to improve convergence speed and robustness property. The stability of the learning process is proven. Finally, two experiments are presented to evaluate the proposed method. The first one shows an improvement in earthquakes prediction in the long-term, and the second one is a classical Benchmark problem. In both experiments, the modified Meta-Learning technique minimizes remarkably the mean square error index.

Keywords: Meta-learning, neural networks, long-term earthquake prediction

DOI: 10.3233/JIFS-210173

Journal: Journal of Intelligent & Fuzzy Systems, vol. 41, no. 6, pp. 6375-6388, 2021