Time: 31 December 2021 (Friday) 15:00-17:00 pm

Venue: 418, Main Building

Speaker: Song Jie, Long-tenured Professor, Peking University

Speaker: Song Jie, Associate Dean of the School of Engineering, Peking University, PhD supervisor of the Department of Industrial Engineering and Management, and Distinguished Professor under the "Changjiang Scholar Award Scheme" of the Ministry of Education. D. degree in Industrial Engineering from Tsinghua University in 2010. He has conducted postdoctoral and visiting research at Georgia Tech, University of Wisconsin-Madison and Columbia University.

His research interests include stochastic optimization modelling and algorithm design, and its application to the optimal allocation of resources and operation management of complex service systems such as healthcare, energy management and national strategic reserves. He serves on the editorial boards of three international SCI journals, including IEEE Automation Science and Engineering, and chairs the Management Committee of the IEEE RAS International Society for Medical Automation. Winner of the IEEE Robotics and Automation Society Best Paper Award 2013. 2020 IEEE TASE Journal of the Year Best Paper Award. 2021 Society of Industrial and Systems Engineers Journal IISE Transactions Best Paper Honorable Mention Peking University 2016, 2017 Teaching Excellence Award

Introduction to the report：We formulate a novel class of online matching problems with learning. In these problems, randomly arriving customers must be matched to perishable resources so as to maximize a total expected reward. The matching accounts for variations in rewards among different customer-resource pairings. It also accounts for the perishability of the resources. For concreteness, we focus on healthcare platforms, but our work can be easily extended to other service applications. Our work belongs to the online resource allocation streams in service system. We propose the first online algorithm for contextual learning and resource allocation with perishable resources. Our algorithm explores and exploits in distinct interweaving phases. We prove that our algorithm achieve an expected regret per period of O(K^−1/3 ),where K is the number of planning cycles. We propose a pioneer algorithm that helps service system to optimize resource allocation decisions while learns the uncertain reward of matching customer-resource pairings.

(Organised by: Department of Management Engineering, Centre for Research and Academic Communication)