Method and hardware circuit for performing average pooling in hardware

CN114239797BActive Publication Date: 2025-02-18GOOGLE LLC
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202111383490.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2017-03-23
Filing Date
2017-09-29
Publication Date
2025-02-18
Estimated Expiration
2037-09-29

Smart Images

  • Figure CN114239797B_ABST
    Figure CN114239797B_ABST
Patent Text Reader

Abstract

A method and hardware circuit for performing average pooling in hardware, for processing an average pooling neural network layer of a neural network. The hardware circuit is configured to process the average pooling neural network layer by performing operations including: sending an input tensor to the average pooling neural network layer from a first portion of a memory of the hardware circuit to a matrix multiplication circuit of the hardware circuit, wherein the matrix multiplication circuit is configured to generate a tensor corresponding to an output of the average pooling neural network layer; sending a kernel from a second portion of the memory of the hardware circuit to the matrix multiplication circuit of the hardware circuit; and generating, by the matrix multiplication circuit of the hardware circuit, the tensor corresponding to the output of the average pooling neural network layer from the input tensor to the average pooling neural network layer.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Computing convolutions using a neural network processor

    WO2016186811A1

  • Power weight quantification-based nerve network forward operation hardware structure

    CN106066783A

  • Reconfigurable processing unit

    US20150170021A1