Kanishka Ranaweera
Kanishka Ranaweera
PhD Student,
ISP Group
CyberPhysical Systems Research Program
Data61, CSIRO
Contacts
email:
phone:
Short Bio
Kanishka is currently reading his PhD in Privacy Preserving Machine Learning at Deakin University and Data61|CSIRO – Australia, under the supervision of Dr. David Smith and Prof. Pubudu Pathirana. He is currently working on communication efficient utility optimization for Differentially Private Federated Learning in Industrial IOT. More specifically he’s work is based on optimizing the differentially private stochastic gradient descent algorithms to improve utility while preserving the privacy aspects in Federated Learning applications in Industrial IOT. His current research interests include:
- Federated Learning with user level differential privacy
- Differentially private optimization methods
- Application of privacy preserved federated learning in IIOT domains.
- Utility optimization for DP-SGD
- Adaptive Gradient Clipping Techniques