Title:
SYSTEMS AND METHODS FOR ARTIFICIAL-INTELLIGENCE MODEL TRAINING USING UNSUPERVISED DOMAIN ADAPTATION WITH MULTI-SOURCE META-DISTILLATION
Document Type and Number:
WIPO Patent Application WO/2024/032386
Kind Code:
A1
Abstract:
A method has the steps of obtaining a set of training samples from one or more domains, using the set of training samples to query a plurality of artificial-intelligence (AI) models, combining the outputs of the queried AI models, and adapting a target AI model via knowledge distillation using the combined outputs.
Inventors:
CHI ZHIXIANG (CA)
GU LI (CA)
ZHONG TAO (CA)
YU YUANHAO (CA)
WANG YANG (CA)
TANG JIN (CA)
GU LI (CA)
ZHONG TAO (CA)
YU YUANHAO (CA)
WANG YANG (CA)
TANG JIN (CA)
Application Number:
PCT/CN2023/109728
Publication Date:
February 15, 2024
Filing Date:
July 28, 2023
Export Citation:
Assignee:
HUAWEI TECH CO LTD (CN)
International Classes:
G06N3/08; G06N20/00
Foreign References:
| US20200272940A1 | 2020-08-27 | |||
| US20190287515A1 | 2019-09-19 | |||
| US20220012637A1 | 2022-01-13 | |||
| CN113610173A | 2021-11-05 | |||
| CN111160409A | 2020-05-15 |
Download PDF: