Patent · US Active

Methods and systems for training convolutional neural network using built-in attention

US11403486B2 · kind B2 · utility

1Cited by
0References
20Claims
0Family size

Assignee

Inventors

Key dates

Filing dateNov 11, 2020
Grant dateAug 2, 2022
Priority date
Expiry dateDec 19, 2040

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG06V30/19173
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

Methods and systems for updating the weights of a set of convolution kernels of a convolutional layer of a neural network are described. A set of convolution kernels having attention-infused weights is generated by using an attention mechanism based on characteristics of the weights. For example, a set of location-based attention multipliers is applied to weights in the set of convolution kernels, a magnitude-based attention function is applied to the weights in the set of convolution kernels, or both. An output activation map is generated using the set of convolution kernels with attention-infused weights. A loss for the neural network is computed, and the gradient is back propagated to update the attention-infused weights of the convolution kernels.

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