University of Texas at Austin

Researchers show that for certain kinds of games, an overlooked class of algorithms performs much better than expected.

New LiGO technique accelerates training of large machine-learning models, reducing the monetary and environmental cost of developing AI applications.