公开(公告)号：US11409991B2

公开(公告)日：2022-08-09

申请号：US16422797

申请日：2019-05-24

Applicant: Google LLC

Inventor： Vineet Gupta ,  Philip M. Long ,  Hanie Sedghi

IPC: G06K9/62 ,  G06N3/08 ,  G06F17/16 ,  G06F17/14

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a convolutional neural network using a regularization scheme. One of the methods includes repeatedly performing the following operations: obtaining a kernel of a particular convolutional layer; applying a Fourier transform to the kernel; generating a decomposition using singular-value decomposition (SVD); generating a regularized diagonal matrix; generating a recomposition; applying an inverse Fourier transform to the recomposition; and training the convolutional neural network on training inputs.

公开(公告)号：US20200372300A1

公开(公告)日：2020-11-26

申请号：US16422797

申请日：2019-05-24

Applicant: Google LLC

Inventor： Vineet Gupta ,  Philip M. Long ,  Hanie Sedghi

IPC: G06K9/62 ,  G06N3/08 ,  G06F17/14 ,  G06F17/16

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a convolutional neural network using a regularization scheme. One of the methods includes repeatedly performing the following operations: obtaining a kernel of a particular convolutional layer; applying a Fourier transform to the kernel; generating a decomposition using singular-value decomposition (SVD); generating a regularized diagonal matrix; generating a recomposition; applying an inverse Fourier transform to the recomposition; and training the convolutional neural network on training inputs.