Communications of the ACM
Google and Harvard Made AI to Predict Earthquake Aftershocks
Google and Harvard University researchers have trained a neural network to accurately predict the locations of future aftershocks.
Cfredit: Reuters/Carlos Jasso/Reuters
Google and Harvard University researchers trained a neural network on a database of more than 131,000 earthquakes and the sites of their aftershocks to accurately predict the locations of future aftershocks.
The network is the same type of artificial intelligence responsible for powering Facebook photo tagging and Alexa's voice transcription.
The researchers say the algorithm's accuracy stems from two complex metrics that were previously thought not to correlate with aftershocks: maximum shear stress change, and the von-Mises yield criterion. These metrics are typically applied in the fields of bendable materials such as copper or aluminum.
From Quartz
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found