The Intelligence Advanced Research Projects Activity has launched a new competition that seeks to develop systems designed to predict events in North Africa and the Middle East by processing data from publicly available sources.
IARPA said the Mercury Challenge aims to encourage machine learning engineers, data scientists and technologists from industry and academia to develop new automated platforms that will forecast military activity, infectious diseases and non-violent civil unrest in the two regions.
IARPA will evaluate participants through three scoring periods based on event-type matching; data accuracy; location accuracy; and forecast lead time and will award cash prizes of up to $100,000 to developers of winning forecasting systems.
The agency will officially launch the contest on Aug. 1 and close the first scoring period on Oct. 31.
The second and third scoring terms will respectively end on Jan. 31, 2019 and April 30, 2019.
IARPA will announce the winners by spring 2019.