Encrypted ID Union Matching for Secure Federated Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In federated learning, vertical federated learning faces challenges with data security due to the need to align user intersections, which can lead to information leakage.
Innovation Solution
A method for data protection involving the acquisition of a target identification information union set, encrypted using secret keys from both parties, to determine a target sample data set for joint training without revealing original identification information, thereby preventing information leakage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If user intersection alignment is performed in vertical federated learning, then common training can be completed, but information leakage occurs
Solution Approach 1:
The patent introduces encrypted identification information as an intermediary element. Instead of directly aligning user intersections which causes information leakage, the system uses encrypted identification information that can be aligned without revealing the actual user identities. The encryption acts as a mediator that enables alignment functionality while preventing the harmful information leakage effect.
Solution Approach 2:
The patent creates a copy of the identification information through encryption. The encrypted identification information serves as a placeholder or representation that maintains the structural needed for alignment operations while being semantically different from the original identification information, thus enabling common training without exposing sensitive data.
2Measurement precision
If original identification information is revealed for alignment, then accurate matching is achieved, but data security is compromised
Solution Approach 1:
The encrypted identification information serves as an intermediary that enables accurate alignment matching without requiring the revelation of original identification information. The encryption layer maintains the structural integrity needed for precise matching while protecting the underlying data, thus achieving both alignment accuracy and data security simultaneously.
3Loss of information
If encryption is applied to identification information, then information leakage is prevented, but processing complexity increases
Solution Approach 1:
The patent applies encryption to identification information in advance, before the alignment process begins. By pre-encrypting the identification information, the system establishes security measures beforehand, which prevents information leakage during subsequent processing operations. This preliminary action allows the system to maintain security without adding complexity to the core alignment functionality.
Data Source
AI summary
The present disclosure relates to a method and a device for data protection, a readable medium and an electronic apparatus, and the method comprises: acquiring a target identification information union set, wherein the target identification information union set comprises target encryption identification information of a first party of a joint training model and target encryption identification information of a second party of the joint training model, the target encryption identification information in the target identification information union set being obtained by encrypting according to a secret key of the first party and a secret key of the second party; and determining, according to the target identification information union set, a target sample data set for training the joint training model. Therefore, an identification information intersection of the first party and the second party does not need to be determined in advance as in the related technology.