Imagine you want to find patterns in a giant pile of secret documents without actually reading them or letting anyone else see them. This project creates a digital 'blindfold' that lets computers analyze data and find answers while the information stays encrypted. It also strips away unfair biases, like race or gender, so the results are fair and neutral.
Organizations cannot analyze large volumes of sensitive data because of privacy risks and the presence of unfair biases (like race or gender) in AI models.
A set of cryptographic tools (Functional and Hybrid Homomorphic Encryption) and PPML models that analyze encrypted data without decrypting it.
If you are a medical provider dealing with sensitive patient records across borders — this project developed a sleep medicine demonstrator that allows high-accuracy health analysis without exposing private patient identities.
If you are a local authority dealing with security threats across different jurisdictions — this project developed a threat intelligence demonstrator that identifies risks using encrypted data to maintain security and privacy.
If you are a security firm dealing with the need to share threat data without leaking proprietary secrets — this project developed privacy-preserving machine learning models that classify encrypted data directly.
Based on available project data, specific pricing for the resulting software is not mentioned, though the project received an EU contribution of EUR 4,015,550 for development.
The project focused on creating scalable solutions and demonstrated them through two real-world cross-border scenarios in health and threat intelligence.
Based on available project data, specific licensing terms are not provided, but the project contributes to Open Science and Reproducible Research.
Yes, it uses Differential Privacy and cryptographic schemes to ensure data is processed in a privacy-preserving way, reducing the risk of privacy loss.
The project developed a way to combine cryptography with Differential Privacy specifically to secure and privatise existing databases.
The consortium is well-balanced for technology transfer, featuring 13 partners across 9 countries. With an industry ratio of 38% (including 5 industry partners and 3 SMEs), there is significant commercial involvement to ensure the cryptographic tools meet market needs, supported by 6 universities and 1 research center.
Contact TAMPEREEN KORKEAKOULUSAATIO SR in Finland
Contact us to explore licensing the PPML models for your data privacy needs.
