ALEXANDRIA, Va., July 14 -- United States Patent no. 12,682,230, issued on July 14, was assigned to QUALCOMM Inc. (San Diego).
"Convolution with kernel expansion and tensor accumulation" was invented by Jamie Menjay Lin (San Diego), Yash Sanjay Bhalgat (San Diego) and Edwin Chongwoo Park (San Diego).
According to the abstract* released by the U.S. Patent & Trademark Office: "Certain aspects of the present disclosure provide techniques for kernel expansion. An input data tensor is received at a first layer in a neural network, and a first convolution is performed for a first kernel, where the first kernel has a size greater than a preferred size. Performing the first convolution comprises generating a plurality of intermediate tensors by p...