The B5G/6G mobile telecommunication systems expect to realize Artificial Intelligence of Things (AIoT) applications with the help of edge intelligence/AI by the marriage of edge/fog computing and artificial intelligence (AI). Recently, Federated Learning (FL), a promising and privacy-preserving edge intelligence framework, is the key to bringing the era of AIoT. This talk will show an exemplary AIoT service platform based on the FL framework to support mobile dashcam video analysis. Then, we will demonstrate existing privacy and security threats in the FL system triggered by malicious end devices and the abnormal aggregator. Finally, we will show the integration of semi-supervised learning and knowledge distillation techniques into the current FL framework and discuss future FL research directions.
Te-Chuan Chiu received his B.S. degree in Computer Science from National Tsing Hua University (NTHU), Taiwan, in 2010, M.S. and Ph.D. degrees in Computer Science and Information Engineering from National Taiwan University (NTU), Taiwan, in 2012 and 2018. Since 2018, he has served as a postdoctoral research scholar at the Research Center for Information Technology Innovation (CITI), Academia Sinica until 2022. He has been a research scholar of the department of Electrical and Computer Engineering, University of California, Davis (UCD), USA in 2022. He is currently an assistant professor at the department of Computer Science, National Tsing Hua University (NTHU), Taiwan. His research interests include B5G/6G communications, edge intelligence/AI, fog/edge computing, and AIoT.