In this digital era, where the number of people using various digital services and tools are higher than ever before, opportunities abound to collect large amounts of data for statistical purposes and identifying behavioral patterns. This data can be used for further analysis and decision-making by stakeholders from different sectors includding retail, transportation, healthcare, insurance, media and entertainment or public sectors such as medical research, statistics on demographics, etc.
Use of available large volumes of user data is very limited due to privacy concerns, which is the reason why the data are kept isolated in islands of the system, not available for secondary use and processing. Furthermore, practice showed that in many scenarios data are unrightfully accessed and shared with third-parties, and even, when the consent for the data processing exists, the learning models incorporate proxies that are inexact, biased and often unfair.
Project HARPOCRATES, focuses on setting the foundations of digitally blind evaluation systems that will, by design, eliminate proxies such as geography, gender, race, and others and eventually have a tangible impact on building fairer, democratic and unbiased societies. To do so, we plan to design several practical cryptographic schemes (Functional Encryption and Hybrid Homomorphic Encryption) for analysing data in a privacy-preserving way.
Besides processing statistical data in a privacy-preserving way, we also aim to enable a richer, more balanced and comprehensive approach where data analytics and cryptography go hand in hand with a shift towards increased privacy.
In HARPOCRATES we will first show how to effectively combine cryptography with the principles of differential privacy to secure and privatise databases.
Next, we will build privacy-preserving machine learning models able to classify encrypted data by performing high accuracy predictions directly on ciphertexts across federated data spaces.
Finally, to demonstrate how these solutions respond to users’ needs, we will implement two real-world cross-border data sharing scenarios related to health data analysis for sleep medicine and threat intelligence for local authorities.
Furthermore, Harpocrates will build an ecosystem by leveraging existing ecosystems and communities, and thus minimizing the amount of work and resources which are required to contribute to strengthening the EU’s cybersecurity capacities and sovereignty in digital technologies.
Availability of Big Data combined with advancements in Artificial Intelligence (AI) enable broad capabilities for both private and public actors. However, cross-organisation and cross-border data sharing in-line with GDPR is increasingly difficult, as collection of granular, multi-dimensional personal data meets improved capabilities to cross-link data sets.
HARPOCRATES leverages novel cryptographic schemes to advance the capabilities of Privacy Preserving Machine Learning (PPML) and Federated Learning (FL), thus enabling decentralised training, validation, and prediction on encrypted data. Such privacy-preserving services and secure computation enable users to both benefit from cloud-based machine intelligence and maintain control over data.
HARPOCRATES will enable digitally blind evaluation systems demonstrated in practical application scenarios, helping build fairer, democratic, and unbiased societies.
Design efficient function-hiding Functional Encryption schemes.
Combine Functional Encryption and differential privacy for private encrypted databases.
Design a practical multi-client Hybrid Homomorphic Encryption scheme.
Build a Privacy-Preserving machine learning framework by combining Functional Encryption and Hybrid Homomorphic Encryption.
Byzantine-robust Federated Learning scheme with data privacy guarantees.
Real-world case studies and contribution to Open Science and Reproducible Research.
Contribute to Scalable Automated GDPR Compliance.