Secure Computation Data Tuples Without Boolean-to-Arithmetic Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing secure computation protocols face significant communication and computation overheads due to the conversion of Boolean shares to arithmetic shares, which is necessary for performing operations in multi-party secure computation, particularly in machine learning applications.
Innovation Solution
A method and apparatus for generating a data tuple through joint execution of the random oblivious transfer (R-OT) protocol twice by two participants, allowing direct multiplication of Boolean and arithmetic shares without the need for conversion, thereby reducing communication and computation overheads.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If Boolean shares are converted to arithmetic shares for secure computation operations, then computation capability is improved, but communication and computation overheads increase significantly
Solution Approach 1:
The patent segments the share representation by maintaining Boolean shares and arithmetic shares as separate entities, each suited for specific operations. Boolean shares are used for bitwise operations while arithmetic shares handle arithmetic operations, eliminating the need for complete conversion and reducing overheads.
Solution Approach 2:
The system dynamically switches between Boolean and arithmetic share representations based on the operation type. When arithmetic operations are needed, only the necessary arithmetic shares are generated and used, rather than converting all shares, enabling flexible adaptation to different computational needs.
2Adaptability or versatility
If Boolean shares are converted to arithmetic shares to perform arithmetic operations, then arithmetic computation is enabled, but communication volume increases
Solution Approach 1:
Instead of converting all Boolean shares to arithmetic shares (excessive action), the system performs partial conversion only when arithmetic operations are required. This selective approach enables arithmetic computation capability while minimizing unnecessary communication volume.
Solution Approach 2:
The patent applies different representations (Boolean or arithmetic) to different parts of the computation based on local requirements. Boolean shares are used locally for bitwise operations, while arithmetic shares are used locally for arithmetic operations, reducing overall communication volume.
3Adaptability or versatility
If conversion protocols are implemented for Boolean-to-arithmetic share transformation, then operation flexibility is improved, but protocol complexity increases
Solution Approach 1:
The patent extracts the conversion operation from a mandatory step, making it optional and conditional. The conversion from Boolean to arithmetic shares is performed only when needed for arithmetic operations, rather than being a required part of every operation, thereby reducing protocol complexity.
Solution Approach 2:
Instead of requiring Boolean shares to be converted to arithmetic shares for all operations, the system inverts the approach by allowing both representations to coexist and be used directly for their respective operations, eliminating the need for conversion in many cases.
Data Source
Figure 1
Figure 2~3
AI summary
The present disclosure relates to a method and apparatus for generating a data tuple for secure computation, a medium, and a device. The method is applied to a first participant, and the method includes: separately obtaining a first pseudo-random number pair and a second pseudo-random number pair by jointly executing the R-OT protocol with a second participant twice; and generating a first element slice of a data tuple for secure computation based on the two pseudo-random number pairs, where the data tuple includes the first element slice and a second element slice, and the second element slice is generated by the second participant.