Title:
METHOD AND DEVICE FOR GENERATING PREDICTION MODEL OF ANALYSIS OBJECT, AND STORAGE MEDIUM
Document Type and Number:
WIPO Patent Application WO/2022/110213
Kind Code:
A1
Abstract:
A method and device for generating a prediction model of an analysis object, and a storage medium. The method (100) comprises: obtaining (101) first data (201) of an analysis object; obtaining (102) second data (204) of the analysis object, wherein the second data (204) and the first data (201) represent the same physical quantity of the analysis object, and a data type of the second data (204) is different from that of the first data (201); using the first data (201) to train (103) a neural network model (202) containing N hidden layers into a prediction model (203) of the analysis object, wherein N is a positive integer at least equal to 2; and using the second data (204) to update (104) at least one of the N hidden layers contained in the prediction model (203). In the method, multiple types of data training models are used, such that the dependency of the models on single data can be reduced, the model performance is improved, migration learning and acceleration of a model training process are implemented.
Inventors:
QU YING (CN)
BAI XIN (CN)
SCHNEEGASS DANIEL (CN)
GUAN JINYAN (CN)
WANG JIAOJIAN (CN)
LIU XIAONAN (CN)
BAI XIN (CN)
SCHNEEGASS DANIEL (CN)
GUAN JINYAN (CN)
WANG JIAOJIAN (CN)
LIU XIAONAN (CN)
Application Number:
PCT/CN2020/132933
Publication Date:
June 02, 2022
Filing Date:
November 30, 2020
Export Citation:
Assignee:
SIEMENS LTD CHINA (CN)
International Classes:
G06N3/08
Foreign References:
| CN110995475A | 2020-04-10 | |||
| CN111664823A | 2020-09-15 | |||
| CN109685277A | 2019-04-26 | |||
| CN110703899A | 2020-01-17 | |||
| US20190197550A1 | 2019-06-27 |
Attorney, Agent or Firm:
KANGXIN PARTNERS, P.C. (CN)
Download PDF: