Title:
MACHINE LEARNING DEVICE, MACHINE LEARNING METHOD, AND MACHINE LEARNING PROGRAM
Document Type and Number:
WIPO Patent Application WO/2022/172569
Kind Code:
A1
Abstract:
For a first model of a neural network trained using training data of a first domain, a domain adaptation data adequacy determination unit (52) performs transfer learning with training data in a second domain and determines the domain adaptation data adequacy on the basis of the amount of the training data in the second domain. On the basis of the domain adaptation data adequacy, a learning layer determination unit (54) determines a layer of a second model, obtained by duplicating the first model, that is to be trained. A transfer learning execution unit (56) uses training data in the second domain to perform transfer learning of the layer of the second model that is to be trained.
More Like This:
Inventors:
TAKEHARA HIDEKI (JP)
KIDA SHINGO (JP)
YANG YINCHENG (JP)
KIDA SHINGO (JP)
YANG YINCHENG (JP)
Application Number:
PCT/JP2021/045349
Publication Date:
August 18, 2022
Filing Date:
December 09, 2021
Export Citation:
Assignee:
JVCKENWOOD CORP (JP)
International Classes:
G06N3/08; G06N20/00
Domestic Patent References:
| WO2020188971A1 | 2020-09-24 | |||
| WO2020054058A1 | 2020-03-19 |
Foreign References:
| JP2020502665A | 2020-01-23 |
Other References:
KIM SEUNGHYEON; KIM WOOYOUNG; NOH YUNG-KYUN; PARK FRANK C.: "Transfer learning for automated optical inspection", 2017 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), IEEE, 14 May 2017 (2017-05-14), pages 2517 - 2524, XP033112309, DOI: 10.1109/IJCNN.2017.7966162
Attorney, Agent or Firm:
MORISHITA Sakaki (JP)
Download PDF: