Summer Getaway: A Glance at Computer Science,Peking University


The School of Computer Science at Peking University covers the most comprehensive research fields in Computer Science. We conduct a wide range of research in various areas, including software engineering, information security, human-computer interaction and multimedia, search engine and information mining, database and information systems, cloud computing, mobile computing and wireless autonomous networks, computational linguistics, microprocessor research, digital audio and video coding, multimedia communications, digital media processing platform, computer reconfigurable logic CECARL, non-volatile memory technology, 3D chip technology, hardware architecture for energy efficient wireless LAN, and energy-efficient applications.

This summer, we have invited 10 professors from the School of Computer Science to share their research which will mainly cover 4 major research areas, including Artificial Intelligence, Systems, Software Engineering and Theory of Computing.




He Wang

Dr. He Wang is a tenure-track assistant   professor in the Center on Frontiers of Computing Studies (CFCS) at Peking   University, where he founds and leads Embodied Perception and InteraCtion (EPIC)   Lab. Prior to joining Peking University, he received his Ph.D. degree from   Stanford University in 2021 under the advisory of Prof. Leonidas J. Guibas   and his Bachelor's degree in 2014 from Tsinghua University. His research   interests span 3D vision, robotics, and machine learning, with a special   focus on embodied AI. His research objective is to endow robots working in   complex real-world scenes with generalizable 3D vision and interaction   policies. He has published more than 20 papers on top vision and learning   conferences (CVPR/ICCV/ECCV/NeurIPS) with 8 of his works receiving CVPR/ICCV   orals and one work receiving Eurographics 2019 best paper honorable mention.   He serves as an area chair in CVPR 2022 and WACV 2022.

Shanghang Zhang

Shanghang Zhang is an assistant professor from the School of   Computer Science, Peking University. Before joining PKU, she has been the   postdoc research fellow at Berkeley AI Research Lab (BAIR), EECS, UC   Berkeley. She received her Ph.D. from Carnegie Mellon University in 2018, and   her Master from Peking University. Her research covers machine learning and   computer vision, with around 40 papers published on top-tier AI journals and   conference proceedings. She has also been the author and editor of the book   “Deep Reinforcement Learning” (Springer Nature), which is selected to Annual   High-Impact Publications in Computer Science by Chinese researchers. She   is the recipient of AAAI 2021 Best Paper Award,   “Rising Stars in EECS, USA”, and Qualcomm Innovation Fellowship (QInF) Finalist   Award. She has been the chief organizer of several workshops on ICML/NeurIPS,   organizer of the special issue on ICMR, and senior program committee of AAAI.

Zongqing Lu


Zongqing Lu is currently a Boya assistant   professor in the School of Computer Science at Peking University. He received   PhD degree in computer science from Nanyang Technological University in 2014,   master and bachelor degrees from Southeast University. His research focuses   on reinforcement learning.

Shiliang Zhang


Shiliang Zhang received the Ph.D. degree in   Computer Science from the Institute of Computing Technology, Chinese Academy   of Sciences. He was a Post-Doctoral Scientist with NEC Laboratories America   and a Post-Doctoral Research Fellow with the University of Texas at San Antonio.   He is currently a tenure-track Associate Professor in the School of Computer   Science, Peking University. His research interests include large-scale   fine-grained image retrieval and recognition. He has authored or co-authored   over 80 papers in journals and conferences, including IJCV, IEEE T-PAMI,   T-IP, T-NNLS, ACM Multimedia, NeurIPS, ICCV, CVPR, AAAI, etc., with total   citations 6500+. Several of his works are most cited papers in the person   ReID community. He was a recipient of the Outstanding Doctoral Dissertation   Awards from the Chinese Academy of Sciences (CAS) and Chinese Computer   Federation, the President Scholarship from the CAS, the NEC Laboratories   America Spot Recognition Award, the NVidia Pioneering Research Award, and the   Microsoft Research Fellowship. He was a recipient of the Top 10% Paper Award   at the IEEE MMSP, and best paper candidate of IEEE Trans. on Circuits and   Systems for Video Technology. He served as the Associate Editor of CVIU and   IET Computer Vision, Guest Editor of ACM TOMM, Area Chair of CVPR, AAAI,   ICPR, and VCIP, and Co-Chair or TPC Co-chair of a series of workshops and   tutorials on fine-grained visual recognition in ICME, CVPR, FG, ICPR, etc.

Guojie Luo

Guojie Luo is currently an associate professor   at the School of Computer Science. He received his B.S. degree (with honors)   in Computer Science from Peking University, Beijing, China in 2005, and his   Ph.D. degree in Computer Science from the University of California, Los   Angeles in 2011, respectively. He has won the 2013 ACM SIGDA Outstanding   Ph.D. Dissertation Award in Electronic Design Automation and the 2017 ASP-DAC   Ten-Year Retrospective Most Influential Paper Award. His current research   interests include design automation for emerging computer architectures.

Wenfei Wu

Wenfei Wu is an assistant professor at Peking   University. Dr. Wu got his Ph.D. from the University of Wisconsin-Madison in   2015, and his research interest is in networked systems. Dr. Wu has published   43 papers, and some of them are in top conferences, including NSDI, INFOCOM,   SIGKDD, etc. His Ph.D. work virtual network diagnosis was awarded the best   student paper in SoCC13. His work on 5G transport layer design was also   awarded the best paper runner-up in IPCCC19. And his work on programmable   switch accelerated machine learning systems got the best paper award in   NSDI21.

Kaigui Bian


Dr. Kaigui Bian is an associate professor in   Computer Science at Peking University, Beijing. He was a visiting researcher   at Microsoft Research Asia in 2013. His research interests include wireless   networking, mobile computing, and network security. He received six best   paper awards of international conferences. He was awarded the Peking   University Teaching Excellence Award in 2014, and the Tsang Hin-chi Teaching   Excellence Award in 2017. He was the recipient of IEEE Communication Society   Asia-Pacific Board (APB) Outstanding Young Researcher Award in 2018. He is an   IEEE Communication Society Distinguished Lecturer for 2020-2021. He serves as   an Editor for IEEE Transactions on Vehicular Technology.

Xin Zhang

Xin Zhang is an assistant professor in the   School of Computer Science at Peking University. His research areas are   programming languages and software engineering, with a focus on the interplay   between programming systems and machine learning. On one hand, he leverages   machine learning ideas to improve the usability of programming systems. On   the other hand, he develops new analyses and languages to ensure the quality   of machine learning programs. His work has received Distinguished Paper   Awards from PLDI'14 and FSE'15. Xin was a postdoctoral associate at MIT CSAIL   from 2017 to 2020 and received his Ph.D. from Georgia Tech in 2017 which was   partly supported by a Facebook Fellowship.

Leye Wang


Leye WANG is an assistant professor at Peking   University. He obtained Ph.D. at Telecom SudParis and Université Pierre et   Marie CURIE (UPMC), Paris, in 2016. Before joining Peking University, he was   a postdoc research associate at the Hong Kong University of Science and   Technology. His research interests include crowdsensing, urban computing,   data privacy, and federated learning. He has published more than 70 papers   with 3000+ citations according to Google Scholar. His work received the   Honorable Mention award at UbiComp (2016) and the news highlight at   Artificial Intelligence journal (2019).

Tongyang Li


Dr. Tongyang Li is currently an assistant   professor at Center on Frontiers of Computing Studies, School of Computer   Science, Peking University. Previously he was a postdoctoral associate at the   Center for Theoretical Physics, Massachusetts Institute of Technology. He   received Master and Ph.D. degrees from the Department of Computer Science,   University of Maryland in 2018 and 2020, respectively. He received Bachelor   of Engineering from Institute for Interdisciplinary Information Sciences,   Tsinghua University and Bachelor of Science from Department of Mathematical   Sciences, Tsinghua University, both in 2015. Dr. Tongyang Li’s research   focuses on designing quantum algorithms for machine learning and   optimization. In general, he is interested in better understanding about the   power of quantum algorithms, including topics such as quantum query   complexity, quantum simulation, and quantum walks. He was a recipient of the   IBM Ph.D. Fellowship, the NSF QISE-NET Triplet Award, and the Lanczos   Fellowship.


Our summer school will last for 3 days, from July 12-14. There will be two sessions each day, one morning session from 9:00-11:00 AM (Beijing time) and one evening session from 8:30-10:30 PM (Beijing time).

If you would like to participate and learn more about the School of Computer Science or just Computer Science in general, please join us for the 3-day summer school.







9:00-10:30 AM

Zhenjiang Hu

Bin Cui

Yao Guo

Minghui Zhou

Opening: Introduction   of School of Computer Science at Peking University

Through this opening session, you will gain an   insight of School of Computer Science, such as its history, disciplines,   global impact, etc.

This session will also share with you the   programs that the School of Computer Science offers for international   students.

Last but not the least, you will hear from   current international students at the School of Computer Science. 

8:30-9:30 PM

He Wang

Robot Vision and   Learning

The research and development of robotic and   unmanned systems, e.g.  home robots and autonomous vehicles, is a   frontier field in computer science and artificial intelligence leading a way   to artificial general intelligence (AGI). In recent years, deep learning   based 3D vision systems and reinforcement learning algorithms have achieved a   number of breakthroughs, spawning the emerging field -- embodied artificial   intelligence, and generating many new directions and topics worthy of in-depth   investigation. Therefore, we offer this advanced graduate-level course for   students with backgrounds in deep learning and computer vision to further   their study in 3D vision and robot learning. The course will cover various   tasks and problems ranging from the construction of robot vision systems to   vision-based robot control and interaction, and aims to offer deep and broad   discussion of this cutting-edge field.

9:30-10:30 PM

Shanghang Zhang

Towards Machine   Learning Generalization in the Open World

Even   though a great deal of existing work has been devoted to the field of machine   learning, it still suffers from severe challenges: 1) Domain shift and novel   categories of objects often arise dynamically in nature, which fundamentally   limits the scalability and applicability of deep learning models to handle   this dynamic scenario when labeled examples are not available. 2) Since   real-world data usually varies over different environments and has a   long-tailed distribution, it is prohibitively expensive to annotate enough   data to cover all these variances. However, existing deep learning models   usually lack generalization capability and fails to generalize to the   out-of-distribution data with limited labels. In this talk, I will introduce   my research on how to address these challenges by building machine learning   systems that can automatically adapt to new domains, tasks, and dynamic   environments with limited training data. Specifically, I will talk about a   series of my research on both theoretical study and algorithm design from   three aspects: 1) Generalize to new domains; 2) Generalize to new categories;   3) Generalized and efficient machine learning for IoT applications, including   intelligent transportation and healthcare, which promotes the landing of AI   in the real world. Especially, I will discuss the exploration of brain   cognition mechanism to develop generalized machine learning that can adapt to   new domains and modalities with limited labels.


9:00-10:00 AM

Zongqing Lu

Multi-Agent   Reinforcement Learning

Multi-agent reinforcement learning (MARL) is a   well-abstracted model for many real-world problems. In this talk, I will   focus on the MARL algorithms to solve cooperative multi-agent tasks, covering   value decomposition, multi-agent actor-critic, and more recent advances in   this research field.

10:00-11:00 AM

Shiliang Zhang

An Overview to Person   Re-Identification

PERSON Re-Identification (ReID) is a task that   retrieves and identifies a query person from non-overlapping camera networks.   It is commonly tackled as a fine-grained image retrieval task and faces many   challenging issues. For example, lots of persons share similar appearance,   and the appearance of each person can be affected by lots of factors like   cloth change, viewpoint and illumination variance, occlusions, etc. Moreover,   it is very difficult to manually identify the same person across different   cameras, making the data annotation very time consuming and expensive. Due to   its important applications in surveillance and public security, person ReID   has become a popular topic in computer vision and image retrieval community.   Many efforts have been made to promote its performance. This talk gives an   overview to person ReID, its challenges, as well as recent efforts on   supervised, semi-supervised, and fully unsupervised methods for person ReID.

8:30-9:10 PM

Guojie Luo

Application mapping on   Reconfigurable and Tiled Processors

Reconfigurable and tiled processors provide an   extra trade-off point of programmability and efficiency among CPU, GPU, and   ASIC. Coarse-grained reconfigurable architecture (CGRA) is one of the   representative computing devices. The CGRA compilation problem is to map an   application onto a 3D time-space model of the CGRA. In this lecture, we will   give a survey of application mapping problems, as well as an example of   optimization modulo theories (OMT) formulation for an efficient solution.

9:10-9:50 PM

Wenfei Wu

An Efficient   Infrastructure for Distributed Modeling Training

In Deep Neural Network (DNN), the size of the   model and dataset is increasing, and the DNN training tends to be implemented   in a distributed architecture. The PS-worker architecture for DNN systems   suffers from the traffic incast problem, where many workers exchange traffic   with the PS, causing the PS to be the bottleneck. Inspired by the recent   progress in programmable switches, we propose an Aggregation Transmission   Protocol (ATP), which supports multi-tenant and multi-rack in-network   aggregation for DNN training. ATP consists of the networking stack on end   hosts and the aggregation service on switches. The switch allocates its   computation resources to jobs in a decentralized manner. The end host   networking stack has a fallback to complement the switch’s corner-case incapability(e.g.,   overflow, packet loss) and congestion control to share network resources.   Finally, we made a bunch of engineering optimizations to make ATP saturate   the high-bandwidth network (100Gbps). We wrap up ATP as a primitive in the   transport layer and integrate it with ML systems, and show that ATP can   provide both performance gain and correctness to typical DNN training (e.g.,   AlexNet, VGG, ResNet).

9:50-10:30 PM

Kaigui Bian

Improving Quality of   Experience for Video Streaming with AI at Network Edge

Over Internet, video content has consumed more   than 80% bandwidth. In many countries like China, the number of users   watching long- or short-form videos has exceeded 600 millions. However, the   high-speed mobile access network, congested backboned network, and   under-construction edge networks cannot fulfill the demands in video   streaming from Internet users. Hence, it is still challenging to improve the   quality of experience of watching a video online. To address the problem, it   is promising to have artificial intelligence (AI) techniques for enhancing   the video streaming services, e.g., to predict the popularity of video   content in future, to characterize the dynamics of network bandwidth, and to   analyze the user behaviors. Key enabling techniques includes video content   caching, dynamic bit rate selection, super-resolution, object detection,   which support better quality of experience for video content consumers in the   era of 5G and beyond.


9:00-11:00 AM

Xin Zhang

Probabilistic   Programming and Its Applications in Software Analyses

Probabilistic programming has emerged as a new   approach to program artificial intelligence systems. On one hand, it is a new   programming model/language that has built-in support for random variables. On   the other hand, it is a new machine learning model that allows expressing   highly-complex probabilistic models using a general-purpose programming   language. In this talk, I will use representative probabilistic programming   languages as examples to introduce the theories, algorithms, and applications   of probabilistic programming. Then, I will talk about how software analyses   can leverage probabilistic programming to gain new capabilities. These   capabilities enables us building smarter software engineering tools.

8:30-9:15 PM

Leye Wang

Principle of Least   Sensing & Computing: Building an Intelligent System with Minimum Data

With the worldwide emergence of data protection   regulations, how to conduct law-regulated big data analytics becomes a   challenging and fundamental problem. This talk introduces the principle of   least sensing & computing, a promising paradigm toward law-regulated big   data analytics. Under the guidance of this principle, various techniques   including sparse sensing, differential privacy, and federated learning can be   integrated to build an intelligent system with the minimum data.

9:15-10:30 PM

Tongyang Li

Algorithm Design and   Analysis: From Classical to Quantum

Algorithm design and analysis is one of the   most fundamental directions in computer science. Classical algorithms have   been extensively studied since the start of computer science research, but in   the current trend of quantum computing, the design of quantum algorithms is   much less understood. In this talk, I will introduce my research that bridges   the gap between the fields of quantum computing and theoretical computer   science. To be more specific, I will briefly introduce some of my recent   developments on quantum algorithms for machine learning and optimization, and   introduce their connections to the general study of computer science.

While you are interested, please fill out this form (or scan the QR code) before June 30, so that we can send you the link for participation!

Should you have any questions, please contact us at