Title:
INCREMENTAL LEARNING IMPLEMENTATION METHOD AND APPARATUS FOR CLASSIFICATION MODEL, AND ELECTRONIC DEVICE AND MEDIUM
Document Type and Number:
WIPO Patent Application WO/2022/174436
Kind Code:
A1
Abstract:
An incremental learning implementation method and apparatus for a classification model, and an electronic device and a medium. The method comprises: acquiring at least one unlabeled incremental sample (S110); inputting the unlabeled incremental sample into an established classification model one by one for category prediction; wherein the classification model is obtained, by using an extreme learning machine, by modeling at least some initial samples with incomplete labels (S120); and performing incremental learning on the established classification model according to a category prediction result for the unlabeled incremental sample and the corresponding unlabeled incremental sample, so as to realize the training and updating of the classification model (S130).
Inventors:
HE YULIN (CN)
HUANG QIHANG (CN)
HUANG QIHANG (CN)
Application Number:
PCT/CN2021/077147
Publication Date:
August 25, 2022
Filing Date:
February 22, 2021
Export Citation:
Assignee:
UNIV SHENZHEN (CN)
International Classes:
G06N3/04; G06K9/62
Foreign References:
| CN112132179A | 2020-12-25 | |||
| CN110244689A | 2019-09-17 | |||
| CN108920446A | 2018-11-30 | |||
| CN104820687A | 2015-08-05 | |||
| CN109034190A | 2018-12-18 | |||
| CN108764281A | 2018-11-06 | |||
| US20180322416A1 | 2018-11-08 |
Attorney, Agent or Firm:
BEYOND ATTORNEYS AT LAW (CN)
Download PDF:
Previous Patent: DOUBLE SETTLERS – BASED CATALYTIC CRACKING DEVICE AND METHOD
Next Patent: HYDROPHILIC HOT MELT ADHESIVE AND USES THEREOF
Next Patent: HYDROPHILIC HOT MELT ADHESIVE AND USES THEREOF