Loss Function Value Determination Using Secret Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In cooperative security modeling scenarios, calculating the loss function value for a data processing model is challenging due to data privacy concerns, as aggregating data among parties involved can lead to leakage of enterprise data.
Innovation Solution
A method and device that use secret sharing algorithms to enable data parties to cooperatively calculate the loss function value without leaking their own data, by secretly sharing data based on parameters and random numbers, allowing each party to determine the value of the loss function while maintaining privacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data of parties involved in cooperative modeling is aggregated at an independent third party to calculate the loss function value, then the calculation can be completed, but data privacy leakage occurs
Solution Approach 1:
The patent segments the data and computation process by dividing the loss function calculation into multiple parties, each holding secret shares of the data. No single party possesses the complete data, thus preventing privacy leakage while enabling collaborative computation of the loss function value through distributed secret sharing mechanisms.
Solution Approach 2:
The patent introduces secret sharing as an intermediary mechanism that allows parties to compute the loss function without directly sharing or exposing their underlying data. The secret shares act as intermediaries that enable calculation while maintaining data confidentiality, resolving the contradiction between calculation accuracy and privacy protection.
2Object-affected harmful factors
If data is not aggregated to calculate the loss function value, then data privacy is protected, but the loss function value cannot be determined
Solution Approach 1:
The patent merges the computational capabilities of multiple parties while keeping their data separate. Each party contributes secret shares and performs local computations, and the results are combined to determine the loss function value without merging the underlying data, thus achieving both privacy protection and calculation capability.
Solution Approach 2:
Secret sharing serves as an intermediary that enables loss function determination without direct data aggregation. The mathematical properties of secret sharing allow the system to compute the loss function value from distributed shares while preventing any single party from accessing complete sensitive data.
3Object-affected harmful factors
If secret sharing is used to protect data privacy during loss function calculation, then data security is improved, but the computational complexity increases
Solution Approach 1:
The patent segments the computational workload across multiple parties, where each party performs simpler local computations on their secret shares rather than one party performing complex aggregation. This distribution of computational tasks enhances security while managing complexity through parallelization.
Solution Approach 2:
The patent transforms the computational problem by changing parameters from raw data values to secret share representations. This parameter transformation enables secure computation through algebraic operations on shares, which can be more efficient than secure multi-party computation protocols while maintaining data security.
Data Source
AI summary
Implementations of the present specification provide a value determination for a loss function. In an implementation, a first share of data that was generated based on a first parameter is received from a partner. A first term is determined based on the first share of the data. A second share of the data is generated based on a second parameter and a random number and transmitted to the partner. A coefficient of a second term is determined based on the random number. A second term is determined based on the coefficient of the second term and the second share of the data. The value of the loss function is determined based on the first term and the second term.