Synthetic Data Generation for Privacy-Preserving ML Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Edge computing devices are resource-constrained, particularly in terms of power and memory, and face challenges in securely managing and utilizing pooled resources across multi-tenant environments, which limits their ability to handle sensitive data for machine learning model training without compromising intellectual property or privacy.
Innovation Solution
A system that generates synthetic data from low-dimensional statistical features of customer data, allowing machine learning models to be trained without sharing actual proprietary data, using a feature analysis tool to create data feature descriptions that are used to generate synthetic datasets for training AI models, ensuring data privacy and protecting intellectual property.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If actual customer data is shared for machine learning model training, then model accuracy is improved, but data privacy and intellectual property are compromised
Solution Approach 1:
The patent creates synthetic copies of customer data that replicate statistical patterns and features without containing actual sensitive information. The synthetic data generation system produces artificial datasets that preserve the structural characteristics and statistical properties of real data while being completely fabricated, allowing model training without exposing actual customer data
Solution Approach 2:
The system extracts only the essential statistical features and patterns from actual customer data, separating these abstract characteristics from the sensitive underlying information. By taking out only the necessary statistical properties (such as distribution patterns, correlations, and feature relationships) and using them to generate synthetic data, the patent enables model training while leaving the actual sensitive data behind
2Adaptability or versatility
If customer data is shared with third-party providers, then machine learning model training is enabled, but intellectual property rights are at risk
Solution Approach 1:
The synthetic data generation system acts as an intermediary between the customer's actual data and the third-party model training provider. Instead of directly sharing sensitive customer data with external providers, the system creates and shares synthetic data that serves as a safe mediator, preserving the customer's intellectual property while still enabling the provider to train effective models
3Object-affected harmful factors
If data is anonymized to protect privacy, then data privacy is improved, but data utility for machine learning training deteriorates
Solution Approach 1:
The patent transforms data by changing its fundamental parameters - creating entirely new synthetic data points with different values and representations while preserving the statistical distribution and relational patterns. This parameter transformation approach maintains privacy (since no actual data values are used) while preserving utility (since the statistical structure remains intact)
Data Source
AI summary
A first computing system includes a data store with a sensitive dataset. The first computing system uses a feature extraction tool to perform a statistical analysis of the dataset to generate feature description data to describe a set of features within the dataset. A second computing system is coupled to the first computing system and does not have access to the dataset. The second computing system uses a data synthesizer to receive the feature description data and generate a synthetic dataset that models the dataset and includes the set of features. The second computing system trains a machine learning model with the synthetic data set and provides the trained machine learning model to the first computing system for use with data from the data store as an input.