Uncertainty Analysis in Earthquake Prediction using Deep Learning Methods for Improved Risk Management

Kalisz, Aleksandra (2024) Uncertainty Analysis in Earthquake Prediction using Deep Learning Methods for Improved Risk Management. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (3MB) | Preview

Abstract

Predicting earthquake magnitude remains a challenging aspect of natural disaster management. This study explores deep learning techniques to improve earthquake predictions, with a focus on uncertainty in seismic forecasting. The experiment includes Bayesian Long Short-Term Memory (LSTM), Bayesian Convolutional Neural Networks (CNN), Bayesian Temporal Convolutional Networks (TCN), and a Hybrid Bayesian CNN/LSTM model integrated with the Monte Carlo Dropout Method to enhance the reliability of the predictions by effectively quantifying uncertainty. All models underwent a training process of 1000 epochs. The Adam optimizer and Stochastic Variational Inference (SVI) were used to adjust parameters and control uncertainty, improving the learning process. The models were evaluated using various metrics such as Standard Deviation, Uncertainty Estimate, MAE, RMSE and R2 to assess their accuracy and uncertainty. The results indicated that the Bayesian LSTM model was the most effective, delivering the precise forecast while maintaining well-calculated uncertainty, with the lowest Mean Absolute Error (MAE) of 0.0322 and Root Mean Squared Error (RMSE) of 0.0453 and an R-squared score of 0.9903 indicating that it accounted for nearly all the variability in data. Also, the hybrid Bayesian CNN/LSTM model performed well, showing a good balance between accuracy and uncertainty. This research highlights the importance of deep learning methods and their potential in helping to manage the risk of natural disasters, saving lives, and reducing economic losses caused by earthquakes.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Mulwa, Catherine UNSPECIFIED
Subjects:	G Geography. Anthropology. Recreation > GE Environmental Sciences H Social Sciences > HD Industries. Land use. Labor > HD61 Risk Management Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	20 Aug 2025 09:13
Last Modified:	20 Aug 2025 09:13
URI:	https://norma.ncirl.ie/id/eprint/8577

Actions (login required)

View Item