Efficient anonymous trace query method under semi-honesty model in federated learning

• Main Authors: PENG YANXIONG, TAN YANG
• Format: Patent
• Language: Chinese; English
• Published: 15.09.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention relates to an efficient anonymous trace query method under a semi-honesty model in federated learning, which comprises the following steps of: carrying out privacy intersection between participants of federated learning, randomly screening a to-be-queried database containing a privacy intersection by a query party, and sending a corresponding privacy intersection intermediate result as an index {index} to a private database holder; screening a database to be queried by a private database holder; the private database holder performs addition secret sharing on the to-be-queried database to obtain a share and a share2, and sends the share and the share2 to the query party and a semi-honest third party respectively; screening the obtained database secret sharing fragment share1 by the query party to obtain share11; a semi-honesty third party screens the fragment share2 after database secret sharing through a query index {index} to obtain a new data secret sharing fragment share22, and finally a quer