Homomorphic Approximation Functions With Error Assignment

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing homomorphic encryption methods face inefficiencies and accuracy issues when performing operations on approximate polynomials, leading to decreased efficiency and increased error in calculation results.

Innovation Solution

Assigning an error value to encrypted data and performing a homomorphic operation using an approximation function, such as a polynomial, to improve accuracy and efficiency by approximating a target function.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a higher-order approximate polynomial is used to improve accuracy, then the accuracy of the calculation result improves, but the computational efficiency decreases

Engineering Contradiction:
Improveaccuracy of calculation resultVSAvoidcomputational efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent changes the parameter of the approximation function from high-order to low-order, and compensates for the accuracy loss by introducing error value assignment and homomorphic operations. This transforms the problem from using high-order polynomials (which are computationally expensive) to using low-order polynomials with error compensation (which are efficient but require additional error handling steps).

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent introduces an intermediary error value that mediates between the low-order approximation and the target function. By assigning error values to encrypted data and performing homomorphic operations, the system bridges the gap between the simplified low-order polynomial and the actual target function, maintaining accuracy without requiring high-order polynomials.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If a low-order approximate polynomial is used to improve computational efficiency, then the computational efficiency increases, but the accuracy of the calculation result decreases

Engineering Contradiction:
Improvecomputational efficiencyVSAvoidaccuracy of calculation result
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies beforehand cushioning by pre-assigning error values to encrypted data before performing homomorphic operations. This error assignment acts as a cushion that compensates for the inherent accuracy loss from using low-order approximations, ensuring that the final result maintains sufficient accuracy despite the simplified polynomial order.

Inventive Principle:
Principle #11Beforehand cushioning (Prior cushioning)

3Productivity

If approximate polynomials are used to perform non-polynomial operations, then the computational efficiency improves, but the accuracy decreases due to intrinsic limitations in specific ranges

Engineering Contradiction:
Improvecomputational efficiencyVSAvoidaccuracy in specific range
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies dynamics by making the approximation function adaptive through error value assignment. Instead of using a fixed low-order polynomial that performs poorly in specific ranges, the system dynamically adjusts by assigning error values that compensate for range-specific inaccuracies, allowing the low-order approximation to maintain accuracy across different input ranges while preserving computational efficiency.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3843318B1Apparatus and method for performing homomorphic operation using approximation function
Publication Date: 2025.12.17 SAMSUNG SDS CO LTD
  • EP3843318B1 patent drawingFigure 1
  • EP3843318B1 patent drawingFigure 2
  • EP3843318B1 patent drawingFigure 3A~3B

AI summary

An apparatus and method for performing an operation using an approximation function are provided. The method for performing an operation according to one embodiment includes assigning an error value to encrypted data; and performing a homomorphic operation for an approximation function, which is obtained by approximating a target function, by using, as an input value, the encrypted data to which the error value is assigned.