Tech Hub,
技术站是我们为数据驱动的研究与实践提供的技术支持。我们致力于帮助学生及从业者掌握前沿数据及算法技术在各自专业领域的融合应用。技术站与PassionULab的科研与实践项目联动，为学员提供带薪实践和积累学术成果及实践经验的宝贵机会。
The Tech Hub serves as our technical support center for data-driven research and practice. We are dedicated to helping students and professionals master the integration and application of cutting-edge data and algorithmic technologies within their respective fields. In collaboration with PassionULab’s research and practice initiatives, the Tech Hub provides participants with valuable opportunities for paid practice, academic achievement, and hands-on professional experience.
技术站主任 | Hub Director
科研机构AI算法研究员
清华大学电子工程系访问博士
擅长领域：
深度学习，时空图神经网络，城市计算，交通大数据，计算社会科学。
研究经历：
Guangyin Jin是一位聚焦于“AI+社会/城市科学”的研究者，他本科和博士均毕业于国内双一流大学，曾是清华大学电子工程系的访问博士和罗马第一大学建筑规划设计与技术系访的访问学者，目前在国内某部属研究院担任研究员，主攻方向是社会/城市复杂系统研究。他目前在相关的研究领域发表了40余篇的SCI/EI论文，其中多篇发表在例如TKDE、TITS、TRC、AAAI、CIKM、SIGSPATIAL等行业顶尖期刊会议上，目前在谷歌学术上已经获得超过2000次引用量。同时，他还多次受邀担任顶级国际期刊和会议的审稿人，担任多本SCI/EI期刊的编委和副主编。
他在过往的研究中主要致力于采用深度学习方法解决城市的时空动态预测问题，例如城市中的突发事件、交通流量、出行轨迹等。技术路线上主要通过时空图神经网络、动态图学习、神经微分方程以及自动机器学习等方法实现各类城市动态的可预测性，其中他所提出的基于对偶图卷积的轨迹行程时间预测算法曾在高德地图数据库中部署并达到最优效果，基于自注意力的轨迹嵌入算法在2021年的ACM SIGSPATIAL与滴滴出行联合主办的数据竞赛中取得世界前10名的成绩。Jin在未来的研究中将更加聚焦于大模型智能体赋能的社会复杂系统研究、轻量化和具备可解释性的时空预测模型以及面向复杂混沌系统的预测与模拟。
AI Algorithm Researcher at a National Research Institute; Visiting Ph.D. Scholar, Department of Electronic Engineering, Tsinghua University
Expertise:
Deep Learning · Spatiotemporal Graph Neural Networks · Urban Computing · Transportation Big Data · Computational Social Science
Research Experience:
Guangyin Jin is a researcher specializing in AI-driven social and urban science, currently serving as a Research Fellow at a national research institute. He earned both his bachelor’s and Ph.D. degrees from top-tier universities in China and has been a Visiting Ph.D. Scholar at Tsinghua University’s Department of Electronic Engineering and a Visiting Researcher at Sapienza University of Rome. His work bridges artificial intelligence, urban computing, and social complexity science.
Jin has published over 40 SCI/EI-indexed papers in premier venues such as TKDE, TITS, TRC, AAAI, CIKM, and SIGSPATIAL, accumulating more than 2,000 citations on Google Scholar. He also serves as a reviewer and associate editor for multiple leading international journals.
His research applies deep learning and graph-based modeling to understand and predict urban spatiotemporal dynamics, including traffic flow, mobility trajectories, and urban anomalies. His innovations—such as dual graph convolutional models and self-attention-based trajectory embeddings—have achieved state-of-the-art results, with one algorithm deployed in Amap’s production system and another ranking Top 10 globally in the 2021 ACM SIGSPATIAL–Didi Data Competition. Moving forward, Jin’s research focuses on large-model intelligent agents for complex social systems, lightweight and interpretable spatiotemporal models, and simulation frameworks for chaotic urban dynamics.
W
技术站课程分享 | Tech Hub Courses