Patent · US Active

Method for feature data recalibration and apparatus thereof

US10824944B2 · kind B2 · utility

1Cited by
0References
12Claims
0Family size

Assignee

Inventor

Key dates

Filing dateNov 7, 2019
Grant dateNov 3, 2020
Priority date
Expiry dateNov 7, 2039

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG06T3/4046
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

A method of recalibrating a feature data of each channel generated by a convolution layer of a convolution neural network is provided. According to some embodiments, since an affine transformation is applied to the feature data of each channel independently of the feature data of the other channel, the overall number of parameters defining the affine transformation is minimized. As a result, the amount of computations required in performing the feature data recalibration can be reduced.

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