Weakly-supervised Federated Graph Learning for Cyber-Physical Systems
PI Jundong Li, Ph.D.
This research is about making computer systems work better together, even with limited and unreliably labeled information, while keeping your data safe.
The project aims to improve the accuracy and efficiency of Cyber-Physical Systems (CPS) by developing a method called Weakly-Supervised Federated Graph Learning (WS-FGL), which helps when there’s not enough labeled data. It focuses on real-world scenarios where labels are scarce or unreliable and offers solutions for different label problems. This research can enhance privacy and collaboration in CPS while also providing educational opportunities about CPS and AI for government, companies, and civil society organizations.
