US Patent Issued to NVIDIA on April 21 for "Pruning and accelerating neural networks with hierarchical fine-grained structured sparsity" (Massachusetts, California Inventors)

ALEXANDRIA, Va., April 21 -- United States Patent no. 12,608,612, issued on April 21, was assigned to NVIDIA Corp. (Santa Clara, Calif.).

"Pruning and accelerating neural networks with hierarchical fine-grained structured sparsity" was invented by Yannan Wu (Cambridge, Mass.), Po-An Tsai (Somerville, Mass.), Saurav Muralidharan (San Jose, Calif.) and Joel Springer Emer (Acton, Mass.).

According to the abstract* released by the U.S. Patent & Trademark Office: "Hierarchical structured sparse parameter pruning and processing improves runtime performance and energy efficiency of neural networks. In contrast with conventional (non-structured) pruning which allows for any distribution of the non-zero values within a matrix that achieves the des...