Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
TOPOLOGY RECOGNITION MODEL COMPRESSION METHOD AND APPARATUS, ELECTRONIC DEVICE AND MEDIUM
Document Type and Number:
WIPO Patent Application WO/2024/037393
Kind Code:
A1
Abstract:
The present disclosure relates to the technical field of computers, in particular to a topology recognition model compression method and apparatus, an electronic device and a medium. The topology recognition model compression method comprises: performing pruning on a model to be compressed, so as to obtain a pruned model, and training the pruned model to obtain a trained pruned model; quantizing the number of bits of weighting parameters of each neural network layer in the trained pruned model from a first number of bits to a second number of bits, so as to obtain a highly-quantized pruned model, and training the highly-quantized pruned model to obtain a trained highly-quantized pruned model; and taking the trained highly-quantized pruned model as the model to be compressed, continuing to perform pruning and quantization on the model to be compressed until a compressed topology recognition model is obtained, and deploying the compressed topology recognition model to an electric power Internet of Things terminal. Therefore, the present disclosure can reduce the size of the topology recognition model while ensuring that the precision of the topology recognition model meets requirements.

Inventors:
YUAN JIANAN (CN)
HUO CHAO (CN)
BAI HUIFENG (CN)
ZHANG GANGHONG (CN)
GAO JIAN (CN)
ZHENG LIBIN (CN)
YU HUADONG (CN)
YIN ZHIBIN (CN)
LUO ANQIN (CN)
XIE FAN (CN)
SHEN YIFAN (CN)
YANG SHUANGSHUANG (CN)
DING XIAO (CN)
Application Number:
PCT/CN2023/111880
Publication Date:
February 22, 2024
Filing Date:
August 09, 2023
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
BEIJING SMARTCHIP MICROELECTRONICS TECH CO LTD (CN)
International Classes:
G06N3/08
Domestic Patent References:
WO2021189906A12021-09-30
WO2022160856A12022-08-04
Foreign References:
CN115049058A2022-09-13
CN108615049A2018-10-02
CN112698123A2021-04-23
CN113033897A2021-06-25
Attorney, Agent or Firm:
SCIHEAD IP LAW FIRM (CN)
Download PDF: