Deep Neural Network with Bayesian Optimization for Earthquake and Tsunami Forecasting

Abstract

Earthquake is one of the most natural disasters that can kill thousands of people and causes millions of dollars losses every year. The early prediction of earthquakes has given more attention in the last few years and many researchers focus not only on the prediction of the earthquake but also estimation of its magnitude. This paper proposes a deep learning framework with Bayesian Optimization called DLBO to predict early the occurrence of an earthquake and tsunami. The proposed DLBO consists of Ensemble Neural Network model and Long Short-Term Memory (LSTM). Bayesian Optimization is utilized to tune the hyperparameters of the DLBO. The results reveal that the proposed model can predict earthquakes with accuracy86% and predict the occurrence of tsunami with 94% compared to other methods from the literature.