Data auditing method and device in federated learning, medium and program product

Abstract

The invention discloses a data auditing method and device in federated learning, a medium and a program product, and the data auditing method in federated learning comprises the steps: obtaining communication data flow information of a federated learning device, the communication data flow information at least comprising one of plaintext data and encrypted data; if the communication data flow information comprises plaintext data, performing basic feature auditing on the plaintext data, and performing content feature auditing on the plaintext data when the basic feature auditing of the plaintext data is passed; if the communication data flow information comprises the secret state data, performing basic feature auditing on the secret state data, and selectively performing content feature auditing on the secret state data according to whether a decryption permission for the secret state data exists or not when the basic feature auditing of the secret state data is passed; and generating an auditing result when the auditing is finished or an abnormal state is found in the auditing process. The technical problem of low federal learning efficiency caused by malicious data in federal equipment communication is solved.

Description

technical field [0001] This application relates to the field of big data technology, and in particular to a data audit method, device, medium and program product in federated learning. Background technique [0002] With the continuous development of artificial intelligence technology, researchers have proposed a machine learning solution called federated learning. Federated learning is a machine learning technology that trains models on multiple decentralized edge devices or servers, allowing multiple On the premise of not sharing the original data, two data parties jointly establish a shared and powerful machine learning model to solve key issues such as data privacy, data security, data access rights, and heterogeneous data access. Currently, the federal When learning, each federated learning device usually needs to transmit confidential data, such as encrypted gradients, encrypted losses, or encrypted model network parameters themselves. Once a malicious device sends mali...

AI Technical Summary

Problems solved by technology

[0003] The main purpose of this application is to provide a data audit method, device, medium, and program product in federated learning, aiming to solve the technical problem of low federated learning efficiency in the prior art due to malicious data in the federated learning communication process

Images