Prof. Celimuge Wu
Celimuge Wu received his PhD degree from the University of Electro-Communications, Tokyo, Japan, in 2010. He is a professor at the University of Electro-Communications, Japan. His research interests include Vehicular Networks, Internet-of-Things, Edge Computing, and Application of Machine Learning in Wireless Networking and Computing. He serves as an associate editor of IEEE Transactions on Network Science and Engineering, IEEE Transactions on Green Communications and Networking, and IEEE Open Journal of the Computer Society. He also has been a guest editor of IEEE Transaction on Intelligent Transportation Systems, IEEE Transactions on Emerging Topics in Computational Intelligence, IEEE Computational Intelligence Magazine etc. He is a recipient of the 2021 IEEE Communications Society Outstanding Paper Award, 2021 IEEE Internet of Things Journal Best Paper Award, IEEE Computer Society 2020 Best Paper Award, and IEEE Computer Society 2019 Best Paper Award Runner-Up. He is/has been a symposium Co-Chair IEEE ICC 2023, a TPC Co-Chair of IEEE International Smart Cities Conference 2021, a TPC Co-chair of the 2021 IEEE Autonomous Driving AI Test Challenge, a General Chair of ICT-DM 2021, a TPC Co-chair of Wireless Days 2021, a Track Co-Chair of IEEE VTC-Spring 2020, a Track Co-Chair of ICCCN 2019, and a Track Chair of IEEE PIMRC 2016. He is the chair of IEEE TCBD Special Interest Group on Big Data with Computational Intelligence, and IEEE TCGCC Special Interest Group on Green Internet of Vehicles. He is a senior member of IEEE.
Title: “Efficient Federated Learning for Internet of Vehicles”
Federated learning (FL) is a promising paradigm for achieving distributed intelligence by protecting user privacy in Internet-of-Vehicles (IoV). Considering limited computing and communication resources, it is important to compress learning models and select appropriate clients from a huge number of users to participate in the training process. This talk discusses two approaches to empower federated learning in IoV. First approach uses a knowledge distillation-based scheme to compress local models to reduce the communication overhead between FL client and the central server. Second approach uses a fuzzy logic based client selection scheme to improve the learning efficiency