Heterogeneous acceleration method, device and system for longitudinal federated logistic regression learning

Abstract

The invention relates to a heterogeneous acceleration method, device and system for longitudinal federated logistic regression learning. The method comprises the following steps: enabling a first participant and a second participant to perform plaintext operation and encryption operation according to respective feature data to respectively obtain an encrypted model parameter of the first participant and an encrypted model parameter of the second participant; performing secret state addition operation on the encrypted model parameter of the first participant and the encrypted model parameter of the second participant to obtain a forward gradient; performing secret multiplication on the forward gradient to obtain a secret multiplication result; and performing mask accumulation operation on the secret-state multiplication operation result to obtain a mask accumulation operation result, and updating a gradient calculation result. Through algorithm splitting and module allocation, the resource utilization efficiency is improved through heterogeneous computing architecture advantages.

Description

technical field [0001] The present application relates to the technical field of data security and privacy protection, and in particular to a heterogeneous acceleration method, device and system for longitudinal federated logistic regression learning. Background technique [0002] With the development of application fields such as artificial intelligence and big data mining analysis, the demand for data volume is increasing. For example, training artificial intelligence application models requires a large amount of training data with appropriate data labels or feature values. High-quality data often come from the application data generated and accumulated in business activities. However, application data is often scattered in the hands of different organizations and individuals. For example, transaction data is scattered in various financial institutions, and medical diagnosis data is scattered in various medical institutions. Cross-industry and cross-field application dat...

AI Technical Summary

Problems solved by technology

In addition, the encrypted data is generally data with a large integer bit width. For example, the original data is a floating-point number but after encryption, it is generally at least 1024 bits, and federated learning also involves a large number of encrypted operations, so it is very important for federated learning or similar Model training or logistic regression learning in application scenarios such as private computing poses huge challenges from storage resources to computing performance

Images