Prof.Jie Wu

China Telecom and Temple University

Abstract

As Deep Neural Networks (DNNs) are increasingly used in various applications, including computer vision for image segmentation and recognition, reducing the makespan of DNN computation—particularly when running on IoT devices—has become essential. Offloading offers a viable solution by transferring computation from a slower IoT device to a faster, but remote, cloud server. Since DNN computation involves a multi-stage processing pipeline, it is critical to determine the optimal stage at which offloading should occur to minimize the total makespan. Our observations show that local computation time on a mobile device increases linearly with the number of processed layers, while the offloading time decreases monotonically and follows a convex curve as more DNN layers are computed locally. Based on this observation, we first study the optimal partitioning and scheduling for a single line-structured DNN and then extend our results to multiple line-structured DNNs. Heuristic approaches for general DNN structures, represented by Directed Acyclic Graphs (DAGs), are also discussed using a path-based scheduling policy. The effectiveness of our proposed solutions is validated through real-system implementation.

Biography

Jie Wu is the Laura H. Carnell Professor at Temple University and the Director of the Center for Networked Computing (CNC). He served as Chair of the Department of Computer and Information Sciences from 2009 to 2016 and as Associate Vice Provost for International Affairs from 2015 to 2017. Before joining Temple University, he was a Program Director at the National Science Foundation and a Distinguished Professor at Florida Atlantic University. His current research interests include mobile computing and wireless networks, routing protocols, network trust and security, distributed algorithms, applied machine learning, and cloud computing. Dr. Wu has published extensively in scholarly journals, conference proceedings, and books, and serves on several editorial boards, including IEEE/ACM Transactions on Networking. He has served as General Chair or Co-Chair for IEEE IPDPS 2023, ACM MobiHoc 2023, and IEEE CCGrid 2024, and as Program Chair or Co-Chair for IEEE INFOCOM 2011, CCF CNCC 2013, and ICCCN 2020. He also chaired the IEEE Technical Committee on Distributed Processing (TCDP). Dr. Wu is a Fellow of the AAAS and IEEE and a Member of Academia Europaea (MAE). He is currently the Chief Scientist and Director of the Cloud Computing Research Institute at China Telecom.