Database Cleanroom ML for Privacy-Preserving Data Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing database systems face challenges in securely and scalably sharing data between different users while preserving privacy, particularly in the context of machine learning model training and scoring, where malicious entities can exploit unrestricted access to extract sensitive information and perform attacks.
Innovation Solution
The data clean room system enables secure data sharing by allowing parties to train machine learning models on combined data sets using privacy-preserving methods, with capabilities to revoke models unilaterally and share pre-trained models securely without revealing details, while implementing row access policies and approved query statements to restrict data exposure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If data is shared between different users in a database, then data accessibility and collaboration are improved, but security and privacy protection deteriorate
Solution Approach 1:
The patent introduces a data clean room as an intermediary environment that enables secure data sharing between users. The clean room acts as a mediator that allows multiple parties to access and analyze combined datasets without exposing raw data, thus maintaining security while improving accessibility. The system uses controlled query execution and result aggregation to achieve this balance.
2Adaptability or versatility
If unrestricted access to data is allowed for machine learning training, then model training flexibility is improved, but vulnerability to malicious attacks increases
Solution Approach 1:
The patent applies preliminary anti-action by implementing security controls and access restrictions before machine learning training begins. The system pre-configures the clean room environment with approved query statements and access policies that prevent malicious attacks while allowing legitimate training operations. This proactive security measure blocks potential attacks before they can exploit unrestricted access.
3Measurement precision
If detailed data is exposed for analysis, then analytical accuracy is improved, but privacy preservation deteriorates
Solution Approach 1:
The patent extracts only the necessary aggregated results from the data analysis process while leaving the detailed raw data protected within the clean room. The system allows analytical operations to be performed on combined datasets but extracts only high-level statistics and insights, not the underlying detailed records. This enables analytical accuracy while preserving privacy by taking out only what is needed.
Data Source
AI summary
Embodiments of the present disclosure may provide a data sharing system implemented as a local application in a consumer database of a distributed database. The local application can include a training function and a scoring function to train a machine learning model on provider and consumer data, and generate output data by applying the trained machine learning model on input data. The input data can include data portions from a consumer database and a provider database that are joined to create a joined dataset for scoring.