The past few years have seen an increase in the application of machine learning methods for seismic data analysis. This is due to the increased adoption and visibility of freely available and easy-to-use machine learning toolkits, faster computation, reduced cost of data storage, and the very large sets of continuous geophysical and laboratory experimental data. The combination of these factors means that now is the time to consider machine learning as one of the key strategies modeling tools in both improving routine data processing and for better understanding the underlying geophysical processes.

Already, significant progress has been made in seismic waveform detection and classification of seismic waves for automatic onset picking. Such advances are allowing us to vastly speed up and improve the accuracy of previously laborious processing flows. In other notable recent applications, waveforms and ground motions, from both laboratory and natural datasets, are being used to understand the precursory physics of sudden- and slow-slip and to predict aftershock locations within supervised learning frameworks.

In this session, we will see machine learning focussed presentations covering topics such as seismic waveform processing, earthquake cataloging, earthquake classification, and earthquake cycle behavior from numerical and laboratory experiments.

In particular we would like to highlight invited talks from
Beroza et al.: Earthquake Monitoring with Deep Learning
Hulbert et al.: Probing Fault Physics Applying Machine Learning
De Hoop et al.: Unsupervised learning for identification of seismic signals
and
Kriegerowski et al.: Deep learning for localizing and detecting earthquake swarm activity based on full waveforms: Chances, challenges and questions