Patent · US Active

Learning convolution neural networks on heterogeneous CPU-GPU platform

US10002402B2 · kind B2 · utility

6Cited by

1References

14Claims

0Family size

Assignee

SONY GROUP CORPORATION · JP

Inventors

Ming-Chang Liu · San Jose, US
Xun Xu · Shanghai, CN
Da Li · San Francisco, US

Key dates

Filing date	Jul 22, 2016
Grant date	Jun 19, 2018
Priority date	—
Expiry date	Oct 11, 2036

Classification

Technology area (CPC G)Physics
CPC primaryG06N3/09
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

Convolution neural networks are able to be trained using a GPU and a CPU. To efficiently utilize a device's resources, the HetNet and HybNet approaches have been developed. The HetNet approach separates batches into partitions such that the GPU and CPU process separate batches. The HybNet approach separates the layers of a convolution neural network for the GPU and CPU.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.