Title of the Report:Federated Training Technology and Resource-Constrained Edge Inference Techniques for Edge Intelligence Applications
Presenter:Professor Junchao Du
Affiliation:Xidian University
Date of the Report:June 7, 2024 (Friday) 10:00 AM
Location of the Report:Third Conference Room, First Floor, Block A, Feicui Science and Education Building
Report Abstract:With the rapid development of edge intelligence applications, especially embodiment intelligence applications enhanced by large models, federated training technology of multimodal perception large models in edge intelligence applications and inference technology in resource-constrained edge nodes have attracted attention from academia and industry. This report introduces the team's research on federated training technology of multimodal perception models, two-way knowledge distillation technology, and federated semi-supervised learning technology in the face of challenges such as data and computation heterogeneity and lack of labeled data. For inference technology in resource-constrained edge nodes, research on model compression frameworks, cross-domain adaptation technology, and multi-person command recognition technology in noisy environments is conducted.
Biography of the Presenter:Junchao Du, a leading professor of "Huashan Scholar" at Xidian University, a second-level professor of Xidian University, a specially appointed professor of "Yangtze River Scholar" by the Ministry of Education, the executive deputy director of the Blockchain Technology Application and Evaluation Engineering Research Center under the Ministry of Education, a backbone member of Shaanxi Provincial Key Laboratory of Intelligent Human-Computer Interaction and Wearable Technology, a leader of Shaanxi Provincial Scientific Innovation Team, a chief scientist of Shaanxi Qinyuan Scientist + Engineer Team, a standing director of ACM Xi'an Chapter, and an organizing committee member of ACM China Turing Congress. He won the Second Prize of the Ministry of Education's Technology Invention in 2022 and the First Prize/Special Prize of the Science and Technology Award of Shaanxi Colleges and Universities in 2021/2019 as the first completer. He has published the national key textbook "ZigBee Technology Principle and Practice". He has undertaken projects and topics such as the National Natural Science Foundation's key project, major national special projects, and national key R&D projects. He has published many papers in international flagship conferences such as ACM Ubicomp, ACM Mobisys, and IEEE/ACM journals and won the 2017 ACM Ubicomp Outstanding Paper Award (CCF Class A). He presided over the formulation of the China Electronics Society standard "Computer System Virtualization Technology and General Testing Requirements" (T/CIE106-2021).