Xu Wangjie, class of 2025 in Electrical Engineering of ZJUI, has earned multiple academic honors, including “Outstanding Academic Model Student” “Innovation and Entrepreneurship Model Student” and “Excellent Student” as well as a Third-Class Scholarship from Zhejiang University. He has also been recognized as an Outstanding Graduate of Zhejiang University. He is a co–first author of Personalized Local Differentially Private Federated Learning with Adaptive Client Sampling, accepted by Institute of Electrical and Electronics Engineers International Conference on Acoustics, Speech, and Signal Processing(IEEE ICASSP), and a co-author of Intelligent Topology Control of Distribution Power Network for Increased Clean Energy Absorption Using Reinforcement Learning, published in Institute of Electrical and Electronics Engineers Asia Conference on Power and Electrical Engineering (IEEE ACPEE). As a member of a 2022 Zhejiang University National SRTP project, he conducted interdisciplinary research on federated learning and differential privacy, successfully completing the project defense and earning the award of Outstanding Performance in 2022 Undergraduate Summer Research Program. In competitions, he was a Finalist Winner in the 2024 Mathematical Contest in Modeling/Interdisciplinary Contest in Modeling (MCM/ICM) and received the “Eaton Cup” Best Design Award in the ECE385 course. He has obtained exemption from examinations for master’s program of ZJUI.
Xu Wangjie recalled that the course offered a wide variety of projects at the time, including an FPGA-based version of Terraria, rhythm game demos, and various FPGA-implemented peripheral driver designs. These projects were both engaging and impressive. However, among all these diverse options, he noticed there were no 3D-related designs—and this sparked his new creative inspiration.
▲ A simplified version of Minecraft implemented on an FPGA platform
"The biggest challenge we faced was figuring out how to integrate 3D rendering technology with FPGAs," Xu remembered. "On one hand, we had no prior experience with 3D rendering; on the other hand, we were still in the process of learning and internalizing FPGA’s new programming methods." To make matters more challenging, all this happened during final exam week. The double pressure of technical difficulties and tight deadlines left them feeling a sudden surge of stress.
Yet Xu and his teammates refused to be defeated by these obstacles. Instead, they made full use of fragmented time to learn extra-curricular knowledge independently and actively explore solutions. "It’s no exaggeration to say that during that period, aside from eating, sleeping, and attending classes, we spent almost all our time developing programs in front of the computer," Xu said with a smile. "I truly felt the power of passion and the unique appeal of interdisciplinary exploration."
▲ ECE385 Course Demo Presentation
They started with the technical challenges of 3D rendering: consulting numerous materials related to computer graphics, and gradually understanding and mastering its core principles through repeated programming and testing. By integrating the SystemVerilog programming language learned in ECE385 and linear algebra knowledge from MATH257, the team finally completed the design and development of FPGA-Minecraft after intensive research and practice—and took home the "Eaton Cup" Best Design Award.
Looking back on this intense yet rewarding experience, he reflected: "Choosing ZJUI was definitely the right decision. It’s precisely because of the Institute’s interdisciplinary curriculum and open, diverse education that my interests were able to truly take root and be put into practice." In this exploratory style of learning, students not only solidify a strong foundation of basic knowledge but also gradually clarify their future development through continuous interdisciplinary research and practice.
When talking about his research training, Xu Wangjie said his first systematic exposure to a research project came in his second semester of freshman year. At that time, he and his teammates signed up to join the "Federated Learning and Differential Privacy" project led by Assistant Prof. Zhang Meng of ZJUI. The project focused on exploring practical applications such as data privacy protection and distributed model training. As a team member, Xu was primarily responsible for theoretical derivation, formulating optimization problems, writing code, and building a federated learning experimental platform based on Raspberry Pi.
▲ Differential Privacy Federated Learning Framework Deployed on a Raspberry Pi Platform
He was particularly grateful for the patient guidance provided by Assist Prof. Zhang Meng. "Assist Prof. Zhang is very kind and approachable. During our first discussion, he patiently walked us through the background knowledge of this field, planned the research path based on our interests, and guided us to think about how to further use multiple Raspberry Pi terminals to build a federated learning experimental platform." With the careful guidance of professors, Xu noted that his ability to read academic literature and write code improved significantly, and he mastered new knowledge such as communication between multiple Raspberry Pi devices, PyTorch, and computer technology development frameworks.
Later, when participating in ZJUI’s Summer Research Training Program as a freshman, Xu once again chose to join Assist Prof. Zhang Meng’s research group. He continued his interdisciplinary research on federated learning and differential privacy, systematically organizing the research results and further clarifying the research’s innovations and core contributions. These two experiences not only helped Xu—who had just entered ZJUI—understand what research entails, but also enabled him to master basic research skills and methods, gradually making the leap from a research "novice" to a beginner.
Building on his previous accumulation in computer technology and combined with his studies in the Electrical Engineering and Automation major, Xu continued to expand his interdisciplinary practical experience and consolidate his theoretical knowledge. By his junior year, he had gradually clarified his research interest: optimizing the topology of distribution networks in power systems using reinforcement learning. "This direction aims to assist power grid dispatching and structural optimization through agent algorithms, enhancing the flexibility and safety of system operation. It is particularly relevant for addressing practical challenges such as new energy integration and power fluctuations," Xu explained. "I hope that through this research, I can contribute my part to the intelligent and green transformation of energy systems."
With this goal in mind, he joined the project "New Multi-Objective Collaborative Optimization Strategy for Multi-Type Resources in Distribution Networks for Optimal Dispatching" led by Associate Prof. Diao Ruisheng at ZJUI. Under Professor Diao’s guidance, Xu mastered how to use multi-objective optimization strategies for power grid dispatching, gained a deep understanding of the design and application of multi-agent systems, and developed a more profound insight into how algorithms can improve power grid operation efficiency through designing and optimizing reinforcement learning agent decisions.
After a period of exploration, Xu and his team achieved phased results in this field. Their related research, titled Intelligent Topology Control of Distribution Power Network for Increased Clean Energy Absorption Using Reinforcement Learning, was published in the IEEE ACPEE Conference. As a representative, he traveled to Beijing for the first time to attend the Asian Conference on Power and Electrical Engineering (IEEE ACPEE) and gave an official presentation.
▲ Xu Wangjie traveled to Beijing to participate in the ACPEE Conference and give an academic report
