Advances in Earthquake Prevention and Reduction Based on Machine Learning: A Scoping Review

Earthquakes can cause disastrous casualties and infinite economic loss worldwide. People have tried various means to ease earthquake disasters. Machine learning (ML) is one efficient instrument for the above exploration, with its excellent implicit relationship extraction and complex task processing capabilities. In the past few decades, many effective machine learning methods have been applied to reduce the impact of earthquake disasters, and have achieved breakthrough achievements. However, as the number of accomplishments continues to grow, more overall and newer surveys and discussions about earthquake prevention and reduction based on ML should be presented. Based on this, we aim to provide a more fine-grained and up-to-date review of the advances in earthquake prevention and reduction based on ML from the perspectives of the effectiveness of earthquake forecasting and prediction, the development of anti-seismic structure, the ability, and quality of the post-earthquake rescue and emergency, the employment and combination of multi-type earthquake data, the improvement of the data quality and availability, etc. By the goal, we explored all existing literature on the theme of earthquake prevention and reduction based on ML, where screening for article titles, abstracts, keywords, and primary coverage from six academic databases. At last, 2,271 relevant articles were collected. We analyzed the screened articles (a number of 98) and categorized them into eight themes. In each theme, a detailed summary and discussion were presented. Based on the analysis, several discussions and suggestions were provided at the end.