Learning and Managing the Complex Transportation Systems with Multi-source Data

报告题目：Learning and Managing the Complex Transportation Systems with Multi-source Data

报告人：马玮，香港理工大学，助理教授，博士生导师

报告时间：2020年12月08日14:30-16:30

腾讯会议：会议 ID：210 387 330

欢迎全校师生参加！

报告摘要：  Transportation systems have become unprecedentedly complicated and inter-connected over the past decade with the involvement of emerging technologies (e.g. autonomous vehicles, shared mobility) and multiple stakeholders (e.g. public agencies, private sectors). However, many planning, operation, and management approaches for transportation systems remain unchanged and face challenges in adapting to the new era. At the same time, the mobility system has also become more informative than ever before. With the emerging advanced sensing and communication technologies, various data resources, including traditional traffic sensors (loops, cameras, etc.) and high-tech sensors (LiDAR, Bluetooth, GPS probe, mobile phone trajectory, parking spot occupancy, social media, etc.) are available and have been archived for decades in many mega-cities. Those data resources are not fully utilized in modeling and decision making for smart cities and communities.

In this talk, Dr. Ma will discuss a general framework to integrate the transportation network models with Artificial Intelligence (AI) and Machine Learning (ML) techniques for addressing one critical issue: statistical inference of spatio-temporal system dynamics and user behaviors in transportation systems. A theoretical framework consisting of theories, models, algorithms and systems that fuse large-scale multi-source data to infer the spatio-temporal network flow in a network will be presented. Optimal system-wise decisions regarding planning and real-time operations can be made to improve system performance. The proposed framework is examined on a real-world large-scale network, and the experiment results are compelling and satisfactory.

专家简介：

马玮博士，香港理工大学土木与环境工程系助理教授，博士生导师。2014年本科毕业于清华大学土木工程系以及数学系（第二学位），之后赴美国卡内基梅隆大学就读，获得土木与环境工程硕士与博士学位，以及机器学习硕士学位。马玮博士的研究主要关注交通网络建模，数据科学和机器学习的交叉与结合，以及相关智慧交通的应用。马玮博士作为主要研究人员参与了多项美国自然科学基金，交通部，能源部的科研及工程项目。他的论文大多发表在一流交通期刊或者机器学习顶级会议上。曾获得茅以升优秀博士论文奖、梁季典奖学金、以及INFORMS数据挖掘与决策最佳论文奖（理论方向）。马玮博士还担任多个国际会议及研讨会的组织委员会委员。