Prof. Tian Wang Beijing Normal University, China | Brief Introduction: Wang Tian is a Distinguished Professor of Beijing Normal University, Director of the Engineering Research Center of “Big Data, Cloud and Intelligent Collaboration” of the Ministry of Education of China, doctoral supervisor, national young top-notch talent, moderator of key R&D program of the Ministry of Science and Technology, and leader of innovation team of Guangdong ordinary universities. D. from the City University of Hong Kong, selected as one of the top 2% of global top scientists for life, and selected as one of the leading talents cultivation program of Beijing Normal University. He is engaged in research work in the field of Internet of Things and Edge Intelligence, and has published more than 50 papers in CCF Class A and IEEE/ACM Transactions series journals. His papers have been cited more than 15,000 times, with H-index of 69, 10 ESI highly cited papers (including 3 ESI hot papers), 30 authorized invention patents (1 transferred), presided over 1 National Key Research and Development Program of the Ministry of Science and Technology, 5 National Natural Science Foundation of China, and was awarded the Second Prize of Natural Science in Hunan Province, Second Prize of Scientific and Technological Advancement in Fujian Province, and Third Prize of Natural Science in Fujian Province. 汇报题目：合作式边缘智能的研究与应用 摘要：随着物联网（IoT）设备的爆炸性增长和5G网络的普及，数据量激增，对实时处理能力和低延迟的需求也日益增加。传统的云计算模型由于数据传输的延迟和带宽限制，在处理大量边缘生成的数据时面临挑战。为了解决这些问题，合作式边缘智能作为一种新兴范式，正在成为研究的热点。本报告旨在概述其关键概念、任务卸载策略、资源分配算法、技术挑战及其应用案例。 |
Assoc. Prof. Xiaolong Zheng Beijing University of Posts and Telecommunications, China | Brief Introduction: Xiaolong Zheng is an associate professor and doctoral supervisor at Beijing University of Posts and Telecommunications (BUPT), and a national young talent. He has long been engaged in research work related to intelligent IoT, presided over 2 top-level projects of National Natural Science Foundation of China (NSFC) and other important scientific research projects. He has published more than 30 academic papers in Class A journals and conferences such as IEEE/ACM TON, TMC, MobiCom, etc., and has been awarded the Best Paper Award/Best Paper Nominee Award of renowned international academic conferences for 7 times including the Best Paper Award of IEEE SECON, ACM SenSys 2023, and Best Paper Nominee Award of IEEE SECON 2022, a CCF Class B conference. 2022 Best Paper Award, ACM SenSys 2023 Best Paper Nomination Award, etc. He has been selected for the 5th China Association for Science and Technology (CAST) Young Talent Support Project, and has been awarded the ACM SIGMOBILE China Rising Star Award, the Second Prize of the Natural Science of the Chinese Computer Society, and other awards. Speech Title: AI Inference on IoT End Devices Abstract: In recent years, AI applications on IoT end devices (such as smartphones and smartwatches) have developed rapidly, with neural network-driven mobile applications becoming an indispensable part of daily life. The emergence of LLMs like ChatGPT has further elevated natural language theory and generation capabilities to new heights. However, due to the limited memory and computing resources of IoT devices, the performance of localized AI inference on the end devices is often unstable, and the fine-grained behavior during inference is difficult to measure. This talk introduces a measurement tool called nnPerf. This tool enables low-overhead measurement of kernel-level behavior during model inference on edge devices, including CPU and GPU states as well as kernel-level latency during inference. By utilizing nnPerf, it is possible to gain deep insights into fine-grained model inference on end devices, accurately identify and promptly address runtime efficiency bottlenecks, and further tackle the challenges of running complex models and improving inference efficiency on memory-and-computation-constrained end devices. |
