ECE Seminar - Solving the Fusion Problem: Model-Based Refinement and Deep Learning Synthesis
Supporting the below United Nations Sustainable Development Goals:支持以下聯合國可持續發展目標：支持以下联合国可持续发展目标：
Abstract: Effective data fusion requires solving two types of problems: refining the models we have, and creating new synthesis for problems we can't easily define. This talk delves into both through two specific research advancements. The first paper, "Supervisory Measurement-Guided Noise Covariance Estimation," addresses the refinement problem. It introduces a computationally efficient, bilevel optimization grounded in a novel likelihood factorization. This factorization converts a nested Bayesian network into a chain structure, enabling parallel State and Derivative Filters to precisely estimate sensor noise covariances and make a known model-based estimator perform optimally. The second paper, "SonarSweep," addresses the synthesis problem. In the visually degraded underwater environment, traditional geometric models break down. SonarSweep uses a deep learning pipeline to synthesize a fusion function from sonar and vision data, enabling 3D reconstruction where no explicit model exists. Together, these papers showcase the spectrum of modern data fusion, from making principled algorithms smarter to creating entirely new capabilities through learned perception.
Junfeng Wu received the B.Eng. degree from the Department of Automatic Control, Zhejiang University, Hangzhou, China, and the Ph.D. degree in electrical and computer engineering from Hong Kong University of Science and Technology, Hong Kong, in 2009, and 2013, respectively. From September to December 2013, he was a Research Associate in the Department of Electronic and Computer Engineering, Hong Kong University of Science and Technology. From January 2014 to June 2017, he was a Postdoctoral Researcher in the ACCESS (Autonomic Complex Communication nEtworks, Signals and Systems) Linnaeus Center, School of Electrical Engineering, KTH Royal Institute of Technology, Stockholm, Sweden. From 2017 to 2021, he was with the College of Control Science and Engineering, Zhejiang University, Hangzhou, China. He is currently an Associate Professor at the School of Data Science, the Chinese University of Hong Kong, Shenzhen. His research interests include networked control systems, state estimation, and wireless sensor networks, multi-agent systems. Dr. Wu received the Guan Zhao-Zhi Best Paper Award at the 34th Chinese Control Conference in 2015. He is a senior member of IEEE. He has been serving as an associate editor for IEEE Transactions on Control of Network Systems from 2023.