Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
DEEP LEARNING MODEL ACQUISITION METHOD, SYSTEM AND APPARATUS, AND STORAGE MEDIUM
Document Type and Number:
WIPO Patent Application WO/2022/198606
Kind Code:
A1
Abstract:
Provided in the present application is a deep learning model acquisition method, the method comprising: acquiring a first deep learning model and an expected parameter that represents the performance of the deep learning model, wherein the performance of the deep learning model at least comprises one of the following: the size, operation speed and operation precision of the deep learning model; pruning the first deep learning model according to the expected parameter, so as to obtain a second deep learning model; and fixing a parameter of the first deep learning model, and performing joint training on the first deep learning model and the second deep learning model, so as to obtain a first target deep learning model satisfying the expected parameter. By means of applying the method provided in the embodiments of the present application, a deep learning model meeting the requirements of different platforms or different application scenarios can be obtained. In addition, the waste of computing resources in the relevant art can be eliminated, and the defect of failing to acquire a suitable deep learning model when there is a computing resource shortage can be overcome.

Inventors:
ZHANG XUE (CN)
Application Number:
PCT/CN2021/083129
Publication Date:
September 29, 2022
Filing Date:
March 26, 2021
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
SZ DJI TECHNOLOGY CO LTD (CN)
International Classes:
G06N3/04
Foreign References:
CN110633747A2019-12-31
CN111091177A2020-05-01
CN109711528A2019-05-03
CN110555417A2019-12-10
CN109657780A2019-04-19
CN111695375A2020-09-22
CN108334934A2018-07-27
CN112016674A2020-12-01
US20200104716A12020-04-02
Attorney, Agent or Firm:
BEIJING BESTIPR INTELLECTUAL PROPERTY LAW CORPORATION (CN)
Download PDF: