Abstract:
With the rapid advancement of the IoT and social networking applications, billions of smart devices with enormous amounts of data are generated at the network edge. These data provide valuable information for prediction, classification, and other intelligent applications for our daily lives. However, to enable such intelligent services, traditional data-driven AI technologies require collecting and centralizing the training data in the data center or cloud, raising severe data privacy concerns. To tackle this challenge, Federated learning (FL) has emerged as an attractive distributed AI paradigm, which enables many clients to collaboratively train a machine learning model under the coordination of a central server while keeping their raw data private.      

In this talk, I will first introduce FL’s unique characteristics and technical challenges, i.e., system and statistical heterogeneity. Then, I will highlight some of our research works on designing and deploying effective and efficient FL at the resource-constraint network edge, including IoT/Mobile prototype development, adaptive parameter control, and optimal client sampling. Finally, I will talk about some promising FL applications in the medical and finance sectors, as well as some interdisciplinary FL research directions, such as fairness and incentive issues from social and economic perspectives.

Speaker Bio:
Before joining DKU, Dr. Bing Luo was a Postdoctoral Research Fellow at CUHK(SZ) and Yale University. He received his Ph.D. degree from the University of Melbourne, Australia, and B.E. and M.E. degrees from Beijing University of Posts and Telecommunications (BUPT), China. Before pursuing his Ph.D., he was a project manager at China Mobile Corporation, Beijing, China. He was a visiting scholar at Yale University, IBM T. J. Watson Research Center, USA, Friedrich-Alexander-University Erlangen-Nuremberg, Germany, and Aalto University, Finland. 

His research focuses on federated learning (FL), optimization and game theory, embedded AI for IoT and mobile systems, and wireless communications and energy harvesting systems. He has published more than 15 first-author papers in leading journals and conferences in the network and communications community, including IEEE Journal on Selected Areas in Communications (JSAC), IEEE Transactions on Communications (TCOM), and IEEE International Conference on Computer Communications (INFOCOM). He served as TPC Member at IEEE ICC and GLOBECOM, and PC Member at FL-NeurIPS’22, FL-ICML’21, FL-IJCAI’22, and FL-AAAI’22.

The talk will be in English and is open to all members of DKU community.
Register at: https://duke.qualtrics.com/jfe/form/SV_a9UeihXgWgsU3DU

Last registration by Oct. 20, Thursday.