Welcome to the AI-SDC family of tools#
Our tools are designed to help researchers assess the privacy disclosure risks of their outputs, including tables, plots, statistical models, and trained machine learning models
Statistical Disclosure Control for Python
Tools for the Semi-Automatic Checking of Research Outputs. Drop-in replacements for common analysis commands with built-in privacy protection.
Machine Learning Privacy Tools
Collection of tools and resources for managing the statistical disclosure control of trained machine learning models.
R Package Integration
R-language interface for the Python ACRO library, providing familiar R syntax for statistical disclosure control.
Graphical User Interface
A graphical user interface for fast, secure and effective output checking, which can work in any TRE (Trusted Research Environment).
SACRO-ML: Machine Learning Privacy Tools#
SACRO-ML is a free and open source collection of tools and resources for managing the statistical disclosure control (SDC) of trained machine learning models. It provides both ante-hoc and post-hoc privacy assessment capabilities for researchers working with ML models in secure data environments.
Note
New in v1.4.0: Enhanced support for PyTorch models and improved structural attack capabilities.
Getting Started#
Get SACRO-ML installed and configured in your environment
Explore comprehensive examples for all frameworks and use cases
Complete API documentation and attack reference
Community and Support#
Indices and tables#
Acknowledgement#
This work was supported by UK Research and Innovation as part of the Data and Analytics Research Environments UK (DARE UK) programme, delivered in partnership with Health Data Research UK (HDR UK) and Administrative Data Research UK (ADR UK). The specific projects were Semi-Automated Checking of Research Outputs (SACRO; MC_PC_23006), Guidelines and Resources for AI Model Access from TrusTEd Research environments (GRAIMATTER; MC_PC_21033), and TREvolution (MC_PC_24038). This project has also been supported by MRC and EPSRC (PICTURES; MR/S010351/1).
