On Privacy and Ethical AI: The AI Ethicon by HCAIM
The two days dedicated to privacy and ethical artificial intelligence have ended. On June 16-17, students from Hungary, Ireland, Italy and the Netherlands were challenged to a hackathon on the topic of Privacy-Preserving Machine Learning (PPML).
The initiative was organised as part of the activities under the EU-funded project HCAIM and involved four universities, three research centres, three SMEs, and professionals from the ICT world. On the other hand, the attendees were students interested in attending the Master’s Programme in Human-Centred Artificial Intelligence, which will be launched simultaneously in 4 European universities in the next academic year.
During the hackathon, students used a synthetic data generation library to create a differentially private synthetic dataset based on the private dataset. It was then used to train and evaluate a Machine Learning model aimed to be of some value in a real-world setting.
Supported by the professors and professionals, the students worked by comparing themselves several times with the other teams working simultaneously at Technological University Dublin, University of Naples Federico II, HU University of Applied Sciences Utrecht and Budapest University of Technology and Economics.
The final solution was explained by illustrating all the privacy guarantees applied, and the discussion in the presence of companies, experts, and students earned the Unina team the victory of the challenge. “But the most significant challenge still has to be won – commented the University of Naples Federico II students – and is to maintain the balance between the need to have increasingly intelligent systems and the need to protect individual privacy and commercially sensitive data”.