US Patent Issued to NVIDIA on Sept. 16 for "Bayesian optimization of sparsity ratios in model compression" (California, Minnesota, Utah Inventors)

ALEXANDRIA, Va., Sept. 17 -- United States Patent no. 12,417,403, issued on Sept. 16, was assigned to NVIDIA Corp. (Santa Clara, Calif.).

"Bayesian optimization of sparsity ratios in model compression" was invented by Saurav Muralidharan (San Jose, Calif.), Vinu Joseph (Salt Lake City), Animesh Garg (Berkeley, Calif.) and Michael Garland (Lake Elmo, Minn.).

According to the abstract* released by the U.S. Patent & Trademark Office: "One embodiment of a method includes determining, by a Bayesian optimizer, a first sparsity ratio associated with a limit on an accuracy loss caused by compressing the machine learning model. The method further includes selecting, by the Bayesian optimizer, a second sparsity ratio that optimizes a predefined obj...