Application of machine learning, deep learning and optimization algorithms in geoengineering and geoscience: Comprehensive review and future challenge
Citation
Zhang, W., Gu, X., Tang, L., Yin, Y., Liu, D., Zhang, Y. (2022). Application of machine learning, deep learning and optimization algorithms in geoengineering and geoscience: Comprehensive review and future challenge. Gondwana Research, 109: 1-17. Link to paper
Abstract
The so-called Fourth Paradigm has witnessed a boom during the past two decades, with large volumes of observational data becoming available to scientists and engineers. Big data is characterized by the rule of the five Vs: Volume, Variety, Value, Velocity and Veracity. The concept of big data naturally matches well with the features of geoengineering and geoscience. Large-scale, comprehensive, multidirectional and multifield geotechnical data analysis is becoming a trend. On the other hand, Machine learning (ML), Deep Learning (DL) and Optimization Algorithm (OA) provide the ability to learn from data and deliver in-depth insight into geotechnical problems. Researchers use different ML, DL and OA models to solve various problems associated with geoengineering and geoscience. This work aims to make a comprehensive summary and provide fundamental guidelines for researchers and engineers in the discipline of geoengineering and geoscience or similar research areas on how to integrate and apply ML, DL and OA methods. In this context, a systematic review on the state-of-the-art application of ML, DL and OA algorithms in geoengineering and geoscience is presented, where various ML, DL, and OA approaches are firstly concisely introduced, concerning mainly the supervised learning, unsupervised learning, deep learning and optimization algorithms, then their representative applications in the geoengineering and geoscience are summarized via VOSviewer demonstration. Based on the comprehensive review, we discussed what have been learned from the application of ML, DL and OA in geoengineering and geoscience and identified the existing problems in the current work and future study recommendations.