A transfer-learning approach shows promise for predicting fault-slip behavior and possibly earthquakes from limited field observations (IMAGE)

DOE/Los Alamos National Laboratory

A transfer-learning approach shows promise for predicting fault-slip behavior and possibly earthquakes from limited field observations

Caption

Stick-slip events in the earth cause damage like this, but limited data from these relatively rare earthquakes makes them difficult to model with machine learning. Transfer learning may provide a path to understanding when such deep faults slip.

Credit

Dreamstime

Usage Restrictions

none

License

Licensed content

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.