Patent · US Active

Memory-adaptive processing method for convolutional neural network and system thereof

US10963390B2 · kind B2 · utility

0Cited by

1References

9Claims

0Family size

Assignee

NEUCHIPS CORPORATION · TW

Inventors

Ping Chao · Los Altos Hills, US
Chao-Yang Kao · Hsinchu, TW
Youn-Long Lin · Hsinchu, TW

Key dates

Filing date	Aug 7, 2019
Grant date	Mar 30, 2021
Priority date	—
Expiry date	Aug 7, 2039

Classification

Technology area (CPC Y)Emerging Cross-Sectional Technologies
CPC primaryY02D10/00
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

A memory-adaptive processing method for a convolutional neural network includes a feature map counting step, a size relation counting step and a convolution calculating step. The feature map counting step is for counting a plurality of input channels of an input feature map tile and a plurality of output channels of an output feature map tile for a convolutional layer operation of the convolutional neural network. The size relation counting step is for obtaining a cache free space size in a feature map cache and counting a size relation among a total input size, a total output size and the cache free space size of the feature map cache. The convolution calculating step is for performing the convolutional layer operation according to a memory-adaptive processing technique.

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