Secure Two-Party Data Utility Evaluation via Homomorphic Encryption
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the context of machine learning and data sharing, existing technologies face challenges in securely evaluating the utility of feature data without revealing sensitive information, particularly in scenarios where data owners are hesitant to share data until its value is determined, and there is a need for a method to assess the data's potential to improve a classifier without compromising privacy.
Innovation Solution
A system and method for securely estimating the utility of combining data from multiple parties by exchanging encrypted data and statistics, using Paillier homomorphic encryption to compute an encrypted utility statistic without sharing private decryption keys, allowing parties to determine if the data is useful for improving a classifier without revealing the feature vector or data content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data is shared to improve classifier accuracy, then classification performance improves, but data privacy and security are compromised
Solution Approach 1:
The patent introduces encrypted statistical measures (chi-square statistic, mutual information) as intermediaries that enable classifier improvement without direct data sharing. These statistical intermediaries capture the essential relationship between features and classes while preserving data privacy through encryption, allowing parties to assess data utility without exposing sensitive information.
Solution Approach 2:
The patent replaces the mechanical process of direct data sharing and manual utility assessment with cryptographic operations and automated statistical computations. Homomorphic encryption and encrypted statistical calculations substitute for traditional data exchange mechanisms, enabling automatic utility evaluation while maintaining security.
2Productivity
If data utility is evaluated before sharing, then data trading efficiency improves, but computational complexity increases
Solution Approach 1:
The patent performs preliminary encrypted statistical computations (chi-square statistic, mutual information) before data sharing decisions are made. By calculating these utility measures in advance using encryption, the system enables efficient data trading decisions without requiring complex real-time analysis, thus improving productivity while managing computational complexity through pre-computation.
3Reliability
If encrypted computations are performed to preserve privacy, then data security improves, but measurement precision of utility assessment deteriorates
Solution Approach 1:
The patent creates encrypted copies of statistical measures (chi-square statistic, mutual information) that preserve the essential utility information while maintaining security. These encrypted statistical copies enable accurate utility assessment without exposing the actual data, allowing precise measurement of data value while preserving privacy through cryptographic representation.
Data Source
AI summary
Embodiments described herein provide a system for improving a classifier by computing a statistic for the utility of sharing data with a second party. The system may encrypt a set of class labels based on a public key/private key pair to obtain a set of encrypted class labels. The system may send a public key and the set of encrypted class labels to a second computing device. The system may receive an encrypted value computed by the second computing device based on the public key. The system may decrypt the encrypted value based on a private key to obtain a decrypted value. The system may then send a pair of encrypted values computed based on the decrypted value to the second computing device. The system may subsequently receive an encrypted utility statistic from the second computing device, and decrypt the encrypted utility statistic to obtain a decrypted utility statistic.