Siheng Chen is an associate professor at Shanghai Jiao Tong University. Before that, he was a research scientist at Mitsubishi Electric Research Laboratories (MERL) and an autonomy engineer at Uber Advanced Technologies Group, working on the perception and prediction systems of self-driving cars. Before joining the industry, he was a postdoctoral research associate at Carnegie Mellon University. He received the doctorate in Electrical and Computer Engineering from Carnegie Mellon University(CMU) in 2016, where he also received two masters degrees in Electrical and Computer Engineering and Machine Learning, respectively. He received his bachelor’s degree in Electronics Engineering in 2011 from Beijing Institute of Technology, China. His paper “Discrete signal processing on graphs: Sampling theory” won the 2018 IEEE Signal Processing Society Young Author Best Paper Award. His co-authored paper received the Best Student Paper Award at IEEE GlobalSIP 2018.
His research mainly focuses on graph-structured data science, whose goal is to develop theories and algorithms to analyze large-scale data associated with complex and irregular structures. His research is conducted from three aspects: theory (graph signal processing), algorithms (graph neural networks), and applications (autonomous systems, human behavior analysis, 3D point cloud processing, and smart infrastructure). Please see more information in his CV and Google Scholar page.
Graph-structured data science
Data today is often generated from a diverse sources, including social, citation, biological, and physical infrastructure. Unlike time-series signals or images, such signals possess complex and irregular structures, which can be modeled as graphs. Analyzing graph signals requires dealing with the underlying irregular relationships. Graph signal processing generalizes the classical signal processing toolbox to the graph domain and provides a series of techniques to process graph signals, including graph-based transformations, sampling and recovery of graph signals, and graph topology learning. Graph neural networks provide a powerful framework to learn from graph signals with graphs as induced biases. Permeating the benefits of deep learning to the graph domain, graph convolutional networks and variants have attained remarkable success in social network analysis, 3D point cloud processing, quantum chemistry and computer vision.
In our research team, we promote the interpretability of graph signal processing as well as the learning power of graph neural networks. Our paper on sampling theory of graph signals is one of TOP 20 most cited papers among 13,816 publications in IEEE signal processing society over the past 10 years and also won IEEE signal processing society best young author paper award (selected from 7 journals over past 5 years). We also propose two frameworks, including graph scattering networks and graph unrolling networks, which provide two bridges to connect and upgrade graph signal processing and graph neural networks to achieve theoretical interpretability and learning ability at the same time; see more details in our latest T-PAMI paper: https://ieeexplore.ieee.org/abstract/document/9200568/ and NeurIPS 2021 paper: https://arxiv.org/pdf/2110.09807.pdf
Group intelligence occurs when intelligent agents come together and collaborate. In a recent report of European commission, group intelligence is listed as one of the 100 Radical Innovation Breakthroughs for the future. Scientists at the MIT Center for Collective Intelligence are exploring “how people and computers can be connected so that—collectively—they act more intelligently than any person, group, or computer has ever done before.”
In our research team, we study a new perspective from graph-structure data science for group intelligence. We propose a collaboration graph to model the internal relationship among the agents and convert the optimization of collaboration strategy to the task of graph learning. We improve the efficiency of information refinement in collaboration by over 192 times; see more details in our latest NeurIPS 2021 paper: https://github.com/ai4ce/DiscoNet
As one of the most exciting engineering projects of the modern world, autonomous driving is an aspiration for many researchers and engineers across generations. It is a goal that might fundamentally redefine the future of human society and everyone’s daily life. Once autonomous driving becomes mature, we will witness a transformation of public transportation, infrastructure and the appearance of our cities. The world is looking forward to exploiting autonomous driving to reduce traffic accidents caused by driver errors, to save drivers’ time and liberate the workforce, as well as to save parking spaces, especially in the urban area.
In our research team, we study a new generation of perception and prediction system for autonomous driving. Perception is the task of perceiving the surrounding environment and extracting information that is related to navigation. By introducing V2X-based communication system, we could fundamentally resolve the occlusion and long-range issues encountered in traditional perception. Prediction is the task of predicting the future potential trajectories of each object in the 3D scene. We leverage graph neural networks to explicitly model underlying interactions among the traffic actors and provide more precise and interpretable predictions; see more details in our latest NeurIPS 2021 paper: https://arxiv.org/pdf/2110.13947.pdf
陈老师在CMU获得博士学位，博士后完成后，在Uber Advanced Technologies Group、三菱电机实验室MERL担任Research Scientist，至今在TPAMI、TIP、NeurIPS（oral）、CVPR （oral）、AAAI （oral）、ICLR上发表了50余篇论文，Google Scholar引用2000余次，获得过IEEE信号处理协会最佳年轻作者论文奖！学界+业界影响力Max！
如果你想做【一流的学术】工作、在国际顶级期刊和会议发表【高水平 & 有学术影响力】的论文；
303A, Cooperative Medianet Innovation Center, Building 5
School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University
800 Dongchuan Road, Minhang District, Shanghai 200240, China