Federated Learning Model Validation and Encryption
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing federated learning systems lack the ability to effectively explain the validity of output results, making it difficult to determine the applicability and accuracy of the models generated, especially in cross-sectional data analysis scenarios like illegal money transfer detection.
Innovation Solution
A federated learning system that enables cooperative learning between local servers and a central server through a network, utilizing encryption and decryption of models, mean gradient calculation, model updating, and validation error calculation to select and update global models based on local data, ensuring confidentiality and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If deep learning is used to improve learning efficiency, then learning efficiency is improved, but the ability to explain the validity of output results deteriorates
Solution Approach 1:
The patent segments the learning process into multiple rounds where each local server performs learning independently and submits results to the central server. This segmentation allows for both efficient parallel processing (improving productivity) and maintains traceability of each server's contribution (preserving explainability through validation errors and gradients).
Solution Approach 2:
The patent implements feedback mechanisms by calculating and transmitting validation errors and gradients from local servers to the central server. This feedback loop enables the system to track the validity and reliability of each local model's output, thereby maintaining explainability while achieving efficient distributed learning.
2Reliability
If federated learning with multiple local servers is implemented, then data privacy is protected, but communication overhead increases
Solution Approach 1:
The patent extracts only the necessary information (encrypted models, validation errors, and gradients) from each local server and transmits them to the central server, rather than transmitting all raw data. This extraction approach maintains data privacy while significantly reducing communication overhead.
Solution Approach 2:
The patent transmits partial information (validation errors and gradients instead of complete datasets) to achieve the learning objective. This partial action approach maintains privacy protection while minimizing the communication burden on the federated learning system.
3Measurement precision
If continuous learning with past data is performed, then model accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent performs preliminary learning actions at each local server using their own past data before submitting to the central server. This preliminary action allows continuous learning and accuracy improvement to occur in parallel at distributed locations, reducing the computational burden on any single server while maintaining high model accuracy.
Data Source
AI summary
A federated learning system in which a plurality of local servers repeatedly learn cooperatively through communications between the plurality of local servers and a central server via a network. The local server includes a decryption unit, a mean gradient calculation unit, a model updating unit, a validation error calculation unit, an encryption unit, and a local transmission unit that transmits at least one of a current local mean gradient and a current local validation error. The central server includes a central reception unit, a model selection unit, a weight determination unit, and a central transmission unit. The central reception unit receives encrypted current local models and at least one of current local training data counts, the current local mean gradients, and the current local validation errors from the plurality of respective local servers.