Earthquake prediction test shows potential of artificial intelligence

A new attempt to predict earthquakes with the help of artificial intelligence has raised hopes that the technology could one day be used to limit their devastating impact on lives and the economy. The artificial intelligence algorithm, developed by researchers at the University of Texas at Austin (UT), correctly predicted 70% of earthquakes in the week before they occurred during a seven-month trial in China.

Researchers are working to use artificial intelligence to predict earthquakes. By training the artificial intelligence, it is possible to quickly detect fluctuation statistics in real-time seismic data that match historical earthquakes. The results showed that artificial intelligence successfully predicted 14 earthquakes, about 320 kilometers away from the earthquake site, and their intensity was almost completely consistent with the calculated intensity. However, it also missed one earthquake and sent out eight false alarms. It's not yet known whether the same approach will work elsewhere, but the work is part of a growing body of research into artificial intelligence-driven earthquake forecasting. milestone.

“Earthquake prediction is considered the Holy Grail,” said research team member Sergey Fomel, a professor at the Utah Bureau of Economic Geology, a research unit of the Jackson School of Geosciences. making predictions, but our achievement shows that problems that were considered unsolvable in the past are in principle solvable.” Ranked first among other designs. The test results have been published in the Bulletin of the Seismological Society of America. "You can't see the earthquake coming," said Alexandros Savvaidis, a senior research scientist who leads the service's Texas Earthquake Network project (TexNet). "It's just a matter of milliseconds, and the only thing you can control is your level of preparedness. Even if it could reach 70%, that would be a huge result that could help minimize economic and human losses and potentially significantly improve global safety." Earthquake preparedness levels."

The researchers say their approach succeeded by following a relatively simple machine learning approach. The AI ​​was given a set of statistical features based on the team's knowledge of earthquake physics and then told to train on a five-year database of earthquake records. After training, the AI ​​predicts by listening for signs of coming earthquakes in the "background rumble" of the Earth.

"We are so proud of this team and their first-place finish in this prestigious competition," said Scott Tinker, director of the Utah Bureau of Economic Geology. "Of course, it's not just (the earthquake) that's important. ) location and magnitude, as well as time. Earthquake prediction is a thorny problem."

Researchers believe that in places with strong earthquake tracking networks, such as California and Texas in the United States, Italy, Japan, and Greece and Turkey, artificial intelligence can improve the prediction success rate and narrow the prediction range to within 100 kilometers. The next step is to test the AI ​​in Texas, a state that experiences frequent mild and moderate earthquakes. TexNet in Texas has 300 seismic stations and more than six years of continuous recording, making it an ideal location to validate the method. Ultimately, the researchers hope to combine the system with physics-based models, This may be more important where the data is poor. This research was supported by TexNet, the Texas Computational Seismology Consortium, and Zhejiang University

The above is the detailed content of Earthquake prediction test shows potential of artificial intelligence. For more information, please follow other related articles on the PHP Chinese website!