Synthetic Data Generation via Cumulative Distribution Function Modeling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The challenge is to generate synthetic datasets that maintain the statistical properties of actual data while ensuring privacy, especially in scenarios where sharing or using actual data is undesirable due to privacy concerns or lack of sufficient data, and existing methods like generative adversarial networks (GANs) are inefficient or less effective.

Innovation Solution

A method and system for generating synthetic datasets by creating a cumulative distribution function (CDF) model based on a source dataset, using random numbers to generate discretized synthetic data, and selecting values from the source dataset based on the CDF bins, allowing for the creation of synthetic data that mimics the statistical properties of the actual data without revealing sensitive information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If generative adversarial networks (GANs) are used to generate synthetic data, then privacy protection is improved, but computational efficiency and scalability deteriorate

Engineering Contradiction:
Improveprivacy protectionVSAvoidcomputational efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent creates synthetic data by copying the statistical properties and distribution patterns from real data through CDF modeling, rather than using complex GAN training. The method generates synthetic samples that replicate the essential characteristics of real data while being computationally efficient and scalable.

Inventive Principle:
Principle #26Copying

2Measurement precision

If actual data is used for data science and machine learning applications, then data quality and accuracy are improved, but privacy concerns and data security deteriorate

Engineering Contradiction:
Improvedata qualityVSAvoidprivacy concerns
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent introduces synthetic data as an intermediary between real data and machine learning applications. This intermediary preserves the statistical properties needed for accurate modeling while eliminating privacy risks, allowing applications to operate on data that mimics real data without containing sensitive information.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If synthetic data is generated using complex models, then statistical properties are preserved, but device complexity and computational resources increase

Engineering Contradiction:
Improvestatistical properties preservationVSAvoidmodel complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent transforms the complex problem of generating statistically accurate synthetic data into a parameter-based solution using CDF modeling. By changing the approach from complex generative models to parameterized distribution functions, the method preserves statistical properties while significantly reducing computational complexity and resource requirements.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4066138B1Method and system for generating synthethic data using a regression model while preserving statistical properties of underlying data
Publication Date: 2024.01.03 INTUIT INC
  • EP4066138B1 patent drawingFigure 1
  • EP4066138B1 patent drawingFigure 2
  • EP4066138B1 patent drawingFigure 3

AI summary

A method for generating a synthetic dataset involves generating discretized synthetic data based on driving a model of a cumulative distribution function (CDF) with random numbers. The CDF is based on a source dataset. The method further includes generating the synthetic dataset from the discretized synthetic data by selecting, for inclusion into the synthetic dataset, values from a multitude of entries of the source dataset, based on the discretized synthetic data, and providing the synthetic dataset to a downstream application that is configured to operate on the source dataset.