公开(公告)号：US20240378507A1

公开(公告)日：2024-11-14

申请号：US18690017

申请日：2022-08-15

Applicant: CHINA UNIONPAY CO., LTD.

Inventor： Dong CHENG ,  Xin CHENG ,  Yongkai ZHOU ,  Pengfei GAO ,  Tiecheng JIANG

IPC: G06N20/00

Abstract: The present application provides a model gradient update method and device, for use in improving the accuracy of model training. A central server repeatedly executes a gradient update process until a stop condition is satisfied. One gradient update process comprises: receiving first gradients respectively sent by multiple nodes, the first gradients being obtained by each node using sample data to train a model to be trained of the node one or more times; obtaining a second gradient one the basis of the multiple first gradients and the probability of each node in the present gradient update process, the probability of each node in the present gradient update process being determined by an Actor-Critic network one the basis of the probability of each node in the last gradient update process; and sending the second gradient to the multiple nodes, respectively.

公开(公告)号：US20240323023A1

公开(公告)日：2024-09-26

申请号：US18579216

申请日：2022-07-20

Applicant: CHINA UNIONPAY CO., LTD.

Inventor： Yongkai ZHOU ,  Yanming YANG ,  Hongbao LIU ,  Jianbin ZHENG ,  Yang YANG ,  Zhenyao QIU ,  Chen OUYANG ,  Ling JIN ,  Pengfei GAO ,  Dong CHENG ,  Fen ZOU ,  Hang YUAN ,  Qi WANG

IPC: H04L9/32 ,  G06F21/60

CPC classification number: H04L9/3234 ,  G06F21/602

Abstract: A method for sample alignment is applied to a first participant system, where a first trusted execution environment is deployed at the first participant system. The method includes, in the first trusted execution environment, obtaining at least one first sample identifier of the first participant system; through the first trusted execution environment, obtaining at least one second sample identifier of the second participant system from the second trusted execution environment, where the second trusted execution environment is deployed at the second participant system; in the first trusted execution environment, determining the first initial intersection of the at least one first sample identifier and the at least one second sample identifier and performing the shuffle processing on all first target sample identifiers in the first initial intersection to obtain the first target intersection; and based on the first target intersection, determining the first sample alignment result.