SURE
SURE offers a unified evaluation system for synthetic data, simultaneously measuring privacy and utility. This integrated approach contrasts with conventional methods, which evaluate these factors in isolation, and allows users to effectively balance them during data synthesis.
- Motivation for the project: To address AI's privacy risks, including potential identification in anonymized data, by developing SURE, an open-source library focusing on evaluating Synthetic Data to preserve utility, privacy, and ensure regulatory compliance.
- Generic use case description: Enables users to evaluate synthetic datasets' privacy and utility, providing a tool for individuals with limited AI knowledge to test and preserve privacy without sacrificing data utility.
- Essential functionalities: Open source library with: User-Centric Privacy and Utility Controls and Insights: -Compliance Auditing and Transparency features.
- How these functionalities can be integrated within the software ecosystem: Through its Python SDK and dockerized format, SURE's library ensures interoperability, easy integration into various infrastructures, and supports transparent data governance in line with TRUSTCHAIN’s objectives.
- Gap being addressed: The privacy-utility conundrum in AI model training, where existing anonymization techniques compromise either privacy or data utility, necessitating a balanced approach through PETs and synthetic data.
- Expected benefits achieved with the novel technology building blocks: By enabling privacy-preserving data utility, SURE democratizes access to secure data solutions, supporting responsible AI adoption across sectors and enhancing TRUSTCHAIN’s data privacy goals.
- Potential demonstration scenario: Focus groups with fintech, financial organizations, and healthcare providers to validate SURE's library, showcasing its adaptability and effectiveness in ensuring data privacy and utility in AI applications.
Team
Dr. Shalini Kurapati
Co-founder at Clearbox AI, combining expertise in Technology, Policy, and Management, with a strong background and global professional experience in AI, Data privacy, Ethics, and Stewardship across multiple sectors.
Dr. Luca Gilli
Chief architect behind Clearbox's AI technology, with a PhD in computational mathematics from Delft University of Technology, and with a strong international expertise in Generative AI, Uncertainty Quantification, and Statistical Validation techniques for Trustworthy AI.
Entities
CLEARBOX AI SOLUTIONS SRL
Synthetic Data Solutions provider for privacy, data augmentation, and fairness.
Website: https://www.clearbox.ai
