主讲嘉宾：韩国科学技术院Changhyun Kwon副教授
时 间：2024年5月17日16:00
地 点：信息楼510会议室
主 题：
城市交通优化与人工智能
摘 要：
In this talk, we will discuss how optimization and machine learning algorithms are used for urban transportation systems. We first discuss free-floating electric vehicle sharing systems, which require a rebalancing operation at night to move and charge vehicles to meet the next day's demand. Rebalancing requires not only drivers to move the vehicles but also additional service vehicles to assist with driver transportation. Determining where to move the electric vehicles and selecting the routes for the service vehicles involves a complex optimization process. In this seminar, we introduce two methods: reinforcement learning using artificial neural networks and an adaptive large neighborhood search algorithm. We also discuss the Traveling Salesman Problem with Drone, wherein we coordinate a truck and a drone for parcel delivery. A deep reinforcement learning algorithm for the problem is introduced.
主讲人简介：
Dr. Changhyun Kwon是韩国科学技术院(KAIST)工业与系统工程系副教授。2008年获得宾夕法尼亚州立大学工业工程博士学位，2000年获得韩国科学技术院机械工程学士学位。2023年加入韩国科学技术院，在此之前在University at Buffalo和University of South Florida任职。目前担任Transportation Research Part B: Methodological、Socio-Economic Planning Sciences和Transportation Network Modeling Committee of TRB期刊的编委会成员。目前的研究方向是推进用于高效运输和物流系统的计算优化方法，重点是通过使用机器学习方法来改进启发式和精确算法的效率，以解决大规模的车辆路径问题和移动服务运营问题。其他研究方向包括大规模优化、鲁棒优化、双层优化、深度强化学习、神经组合优化、计算博弈论、市场设计、组合拍卖、稳定匹配等。目前已在国际重要学术期刊发表60余篇，发表的期刊包括：Operations Research, Transportation Science, Transportation Research Part B, INFORMS Journal on Computing等。出版一部书籍Julia Programming for Operations Research，并在Julia中开发数学优化工具。