Machine learning powers many tools we use every day – from navigation apps and online shopping recommendations to medical image analysis. However, these systems rely on large amounts of personal and often sensitive data, raising serious privacy concerns.
In his doctoral dissertation, MSc Tanveer Khan developed new methods that allow machine learning models to be trained and used without exposing sensitive data. His research combines advanced cryptographic techniques, such as homomorphic encryption and secure multiparty computation, with split learning, a collaborative learning approach. These methods make it possible to analyse encrypted data as if it were unencrypted, while aiming to maintain accuracy close to that of unencrypted data. Although some trade-offs remain, his research carefully studies and minimize them.
“Machine learning has great potential in areas like healthcare and finance, but privacy concerns often limit its use. My research shows that it is possible to protect data while keeping model accurate,” Khan explains.
Khan’s findings are particularly for applications such as medical diagnosis and cloud-based AI services, where sensitive data is processed daily. By making privacy-preserving methods more practical, Khan’s work supports the development of safer and more trustworthy AI systems.
Tanveer Khan is originally from Pakistan and is currently working as a doctoral researcher in Network and Information Security Group (NISEC) at Tampere University.
Public defence on Monday 29 September
The doctoral dissertation of MSc Tanveer Khan, in the field of information security, titled Advances in Privacy-Preserving Machine Learning Techniques, Challenges, and Applications will be publicly examined at the Faculty of Information Technology and Communication Sciences at Tampere University at 12:00 on Monday 29 September 2025 at Hervanta campus, Tietotalo building, room TB109 (Korkeakoulunkatu 1, Tampere).
The Opponent will be Associate Professor Kaitai Liang from Delft University of Technology, Netherlands. The Custos will be Professor Antonis Michalas from Tampere University, Finland.