Secure Two-Party Matrix Hybrid Multiplication Protocol
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solutions for secure two-party matrix hybrid multiplication face issues such as high communication overheads, inadequate numerical precision, and security risks due to reliance on homomorphic encryption and oblivious transfer technologies, as well as vulnerabilities in data leakage and third-party cloud service trustworthiness.
Innovation Solution
The implementation of a privacy-preserving computation method and system using a secure two-party matrix multiplication protocol (2PMP) with parallel hybrid multiplication, supporting precision up to 64-bit floating numbers, and a rank-preserving secure 2-party matrix addition decomposition technique to ensure stable computation and security without relying on third-party cloud services.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If homomorphic encryption technology is used for secure two-party matrix hybrid multiplication, then input data privacy and computational verifiability are ensured, but computation efficiency is greatly reduced
Solution Approach 1:
The patent segments the matrix hybrid multiplication into four separate matrix multiplication operations (A1×B2, A2×B1, A1×B1, A2×B2) that can be executed in parallel. Each multiplication result is then split into two parts using obfuscation technique, allowing independent computation without homomorphic encryption overhead while maintaining security through distributed computation.
Solution Approach 2:
The patent introduces an obfuscation technique as an intermediary mechanism that splits matrix multiplication results into two parts (Va and Vb) without requiring homomorphic encryption. This intermediary approach enables secure computation by ensuring that no single party possesses the complete result, thereby maintaining privacy while improving computational efficiency.
2Reliability
If secret sharing technology with public-key cryptography tools is used, then computational security is improved, but computation overheads are substantially increased
Solution Approach 1:
The patent extracts the essential security function from complex public-key cryptography tools and implements a simplified secret sharing mechanism using only obfuscation technique. By taking out the unnecessary cryptographic overhead while retaining the core security benefit of distributed secret sharing, the system achieves computational security with reduced computation overhead.
3Reliability
If conventional secure two-party matrix multiplication protocol is used, then data privacy is maintained, but numerical precision is inadequate
Solution Approach 1:
The patent changes the precision parameter of the matrix multiplication protocol from conventional low-precision fixed-point arithmetic to high-precision 64-bit floating-point arithmetic. This parameter change enables the system to maintain data privacy through the secure protocol while achieving adequate numerical precision for practical applications such as machine learning model training.
Data Source
AI summary
Disclosed are a privacy-preserving computation method and system for secure two-party matrix hybrid multiplication, and a medium. The method includes: A first computation participant and a second computation participant perform secure two-party matrix multiplication based on a private matrix A1 and a private matrix B2 and based on a private matrix A2 and a private matrix B1 by using a secure two-party matrix multiplication protocol, randomly split each matrix multiplication result into two random data items. At the same time, the first computation participant performs private matrix multiplication based on the private matrix A1 and the private matrix A2, and the second computation participant performs private matrix multiplication based on the private matrix B1 and the private matrix B2. The first computation participant and the second computation participant each obtain a final outcome matrix, and send the final outcome matrix to a computation requester to compute a privacy-preserving computation result.