Project Grant 2416607

The National Science Foundation (NSF) awarded a $219,332 Project Grant (CFDA 47.070 - Computer and Information Science and Engineering) to Temple University for a collaborative research project titled "Foundations for Trustworthy Decentralized Federated Learning." The 36-month project, beginning August 2024, aims to design and develop a secure and efficient decentralized federated learning (DFL) system that addresses communication, computation, and security issues. The research will focus on developing computational theories, models, and prototype systems to enable trustworthy DFL, considering both high-performance accuracy and security with privacy preservation. Key areas of focus include establishing efficient and robust communication topologies, investigating bilevel optimization algorithms for emerging machine learning models in DFL, and developing security defense strategies against unique threats. The project will also integrate the research into educational curricula and provide research opportunities for students.