Federated Data Binning Using Secure Label Distribution Exchange
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge of implementing secure and efficient joint binning between data islands formed by independent storage and maintenance of data between different companies and departments in longitudinal federated learning, particularly due to the need for label information transmission that risks information leakage and credential stuffing.
Innovation Solution
A federated-learning-based data processing method where a feature-party device performs initial binning and transmits target identifier sets to a label-party device, receiving label distribution information to determine bin label distribution, and merges bins based on a binning policy, avoiding direct transmission of encrypted label information to maintain data security.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If label information is transmitted directly between data islands for joint binning, then binning accuracy is improved, but data security deteriorates due to information leakage and credential stuffing risks
Solution Approach 1:
The patent introduces an intermediary mechanism where instead of directly transmitting sensitive label information between data islands, the system uses encrypted identifiers and aggregated statistical information as intermediaries. The label party sends encrypted object identifiers to the feature party, and receives aggregated label distribution statistics back, preventing direct exposure of raw label data while still enabling accurate binning.
Solution Approach 2:
The patent creates copies of data in different forms - encrypted copies of object identifiers are transmitted instead of original identifiers, and aggregated statistical copies of label distributions are exchanged instead of raw label information. This copying approach maintains the utility needed for binning while protecting the security and privacy of original data.
2Reliability
If data is stored independently in different companies and departments, then data security and privacy are improved, but joint binning efficiency deteriorates due to data island formation
Solution Approach 1:
The patent segments the joint binning process into distinct independent steps that can be executed across data islands: (1) feature party performs initial binning locally, (2) encrypted identifier exchange occurs, (3) label party computes label distributions locally, (4) aggregated statistics are shared back. This segmentation allows each party to work with their own data securely while still achieving coordinated binning results.
Solution Approach 2:
The patent transitions the problem from a spatial dimension (physical data location) to a statistical dimension (aggregated information exchange). Instead of moving data across data islands, the system exchanges aggregated statistical information and encrypted identifiers, changing the dimension of interaction from data-level to statistic-level communication.
Data Source
AI summary
This application provides a federated-learning-based data processing method, apparatus, and device, and a computer-readable storage medium. The method includes obtaining data to be processed, the data to be processed comprising multiple object identifiers and a feature value corresponding to each object identifier; binning the data to be processed based on the feature value corresponding to each object identifier to obtain a number of bins; determining multiple target identifier sets from each bin, and transmitting each target identifier set to a label-party device; receiving each piece of set label distribution information corresponding to each target identifier set from the label-party device, and determining bin label distribution information corresponding to each bin based on each piece of set label distribution information; and merging bins based on a binning policy and each piece of bin label distribution information to obtain a final binning result.