US Patent Issued to NVIDIA on Dec. 30 for "Distributed weight update for backpropagation of a neural network" (California, Connecticut, Texas Inventors)

ALEXANDRIA, Va., Dec. 31 -- United States Patent no. 12,511,543, issued on Dec. 30, was assigned to NVIDIA Corp. (Santa Clara, Calif.).

"Distributed weight update for backpropagation of a neural network" was invented by Sharan Chetlur (San Jose, Calif.), Natalia Gimelshein (San Jose, Calif.), Thor Mikal Johnsen (Houston) and Simon Layton (West Hartford, Conn.).

According to the abstract* released by the U.S. Patent & Trademark Office: "Speed of training a neural network is improved by updating the weights of the neural network in parallel. In at least one embodiment, after back propagation, gradients are distributed to a plurality of processors, each of which calculate a portion of the updated weights of the neural network."

The patent w...