LDP Protocol Accuracy Evaluation Using Hash-Domain Variance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing systems face challenges in efficiently evaluating the accuracy of local differential privacy (LDP) protocols, requiring extensive computational resources and time to determine the quantity of times source data was the cause of a message, and lack an unbiased estimator for different LDP protocols.

Innovation Solution

A method is introduced to evaluate the accuracy of LDP protocols by predicting the quantity of times source data was the cause of a message using a hash function, adding noise, and updating a matrix with decrypted vectors, allowing for a computationally efficient and unbiased estimation of LDP protocol performance without performing the operations of each protocol.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional systems perform operations of different LDP protocols to evaluate accuracy, then measurement precision is improved, but productivity deteriorates due to extensive computational resources and time required

Engineering Contradiction:
Improveaccuracy evaluationVSAvoidcomputation efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent creates a theoretical model that copies the essential mathematical relationships of LDP protocols without executing actual protocol operations. By using variance formulas and probability distributions to represent protocol behavior, the system evaluates accuracy through computation of expected values and variances rather than running multiple protocol instances, dramatically improving computational efficiency while maintaining evaluation precision

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent performs preliminary analysis by deriving closed-form variance expressions for different LDP protocols before actual evaluation is needed. By pre-computing the mathematical relationships between protocol parameters and accuracy metrics, the system avoids performing repeated protocol operations during evaluation, resolving the contradiction between thorough accuracy assessment and computational efficiency

Inventive Principle:
Principle #10Preliminary action

2Reliability

If LDP protocols introduce noise into data, then data security is improved, but measurement precision deteriorates due to reduced data accuracy

Engineering Contradiction:
Improvedata securityVSAvoiddata accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent systematically varies noise parameters (such as noise scale and distribution) in the theoretical model to evaluate how different levels of noise introduction affect accuracy. By changing these parameters in the mathematical model rather than in actual data processing, the system can optimize the balance between security (noise level) and accuracy (measurement precision) without compromising real data

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent uses variance as a feedback metric to evaluate the impact of noise introduction on accuracy. By computing the variance of estimated frequencies under different noise conditions, the system provides quantitative feedback on the security-accuracy tradeoff, enabling informed selection of noise parameters that achieve desired security levels while minimizing accuracy degradation

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4672665A1Method for evaluating the accuracy of LDP protocols
Publication Date: 2025.12.31 LEMON INC(GB)
  • EP4672665A1 patent drawingFigure 1
  • EP4672665A1 patent drawingFigure 2
  • EP4672665A1 patent drawingFigure 3

AI summary

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for evaluating a protocol for hashing data on client devices. One of the methods includes obtaining (202) a size of a hashing domain for one or more hash functions, each configured to process source data to generate an output value; obtaining (204) a probability of including each output value in a corresponding message; obtaining (206) a differential privacy parameter; and generating (208) a variance of the quantity of times the source data was a cause of a message over the one or more hash functions using the size of the hashing domain, the probability, and a value based on the differential privacy parameter.