Bio
Dan Ristea is a PhD candidate at the UCL Cybersecurity Centre for Doctoral Training with a background in software development. He graduated from the University of Edinburgh with a BSc (Hons) in Computer Science and, just prior to starting his PhD, he gained an MSc in Information Security at UCL. Working for start-ups at various stages of growth, Dan developed an interest in security and privacy, and saw first-hand how privacy and data collection are handled in practice. These experiences led to his current interest in techniques that promise to minimise harms to individuals from data collection while still providing useful aggregate data. His research into how these techniques are verified is supervised by Professor Steven Murdoch and Dr Enrico Mariconti.
Research interests
Dan’s focus at the Turing is to explore how techniques meant to preserve the privacy of individuals, most importantly differential privacy, can be verified, especially when applied to machine learning. Attacks on machine learning models that can extract private information from the data on which the model was trained have motivated the integration of privacy protection with machine learning methods. The success of this approach requires correct proofs and implementations but there is no guarantee of either. Dan is looking to develop ways to ensure that theoretical guarantees of privacy are preserved when translated into practical machine learning applications.