In a Perspective, researchers report that a machine-learning competition to predict the timing of laboratory-generated earthquake simulations attracted more than 4,500 teams, and that the winning teams employed unexpected computational strategies that yielded insights into fault processes; the results suggest the value of engaging the machine-learning community through competitions in other scientific problems of significance.
Article #20-11362: "Laboratory earthquake forecasting: A machine learning competition," by Paul A. Johnson et al.
MEDIA CONTACT: Charles Poling, Los Alamos National Laboratory, NM; tel: 505-257-8007; email: cpoling@lanl.gov
###
Proceedings of the National Academy of Sciences