Speaker
Geoffrey Fox
(Indiana University)
Description
MLCommons Research is described as a community to collaborate with and as a model for similar communities. Its working groups cover Algorithms, Datasets, Platforms, Storage and Science and Medical applications. MLCommons involves 62 companies, 6 DOE laboratories, 11 Universities with a flagship benchmark set MLPerf and the mission of “Accelerating machine learning innovation to benefit everyone.” MLCommons thrusts are Best Practice (software and ontologies), Datasets (Identify and package good ones; develop new ones) and Benchmarks. We describe 7 Science benchmarks with open data and models and a possible direction of Foundation AI models for Science. We note the importance of Clouds, Supercomputers and their integration HPC Clouds.
