Secure Computing Method Using Shamir Secret Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
When machine learning computation is outsourced to cloud service providers, there is a risk of learning data leakage, and existing secure computing technologies like homomorphic encryption have high computational complexity, making secret sharing a preferable method, but it can result in low accuracy of decoding inference result data if not properly managed.
Innovation Solution
A secure computing method using Shamir's secret sharing, where learning and inference data are split into shares and processed across multiple computing devices, with coefficients set to ensure each share is less than the maximum value of the field, allowing accurate decoding of inference results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If secret sharing is used to perform machine learning with distributed learning data, then data secrecy is improved, but decoding accuracy of inference result data deteriorates
Solution Approach 1:
The patent changes the parameter of the secret sharing scheme by setting the maximum value of the field Q to be greater than the maximum absolute value of the inference result data. This parameter adjustment ensures that the inference result data can be accurately decoded while maintaining data secrecy through distributed secret sharing.
Solution Approach 2:
The patent segments the inference result data into multiple shares distributed across different computing devices. By properly managing these segments with appropriate field size selection, the system enables accurate reconstruction of the original data while maintaining security through the secret sharing mechanism.
2Reliability
If homomorphic encryption is used for secure computing, then data secrecy is improved, but computational complexity increases
Solution Approach 1:
The patent extracts the secrecy protection mechanism from homomorphic encryption and implements it through secret sharing instead. This extraction replaces the computationally complex homomorphic encryption operations with simpler secret sharing operations, reducing computational complexity while maintaining data secrecy.
Solution Approach 2:
The patent substitutes the cryptographic mechanism of homomorphic encryption with the algebraic mechanism of secret sharing over a finite field. This substitution replaces complex cryptographic operations with simpler field arithmetic operations, significantly reducing computational complexity while achieving the same security goal.
Data Source
AI summary
According to one embodiment, a secure computing method includes setting a coefficient selected from a ring of integers Q based on first data X, generating n pieces of first fragment data from the first data X based on the coefficient, causing a learning model held in the computing device to learn the first fragment data, generating n pieces of second fragment data from second data Z based on the coefficient, performing, by each of the n computing devices, inference based on the second fragment data using the learning model, and obtaining decoded data dec by decoding k pieces of inference result data. The coefficient is set to make each of the n pieces of first fragment data less than a maximum value of the ring of integers Q.