Xinyuan Zhang | UBC Sauder School of Business

Welcome to my website!

I am a PhD candidate in Management Science at Sauder School of Business, University of British Columbia. My advisor is Prof. Michael Jong Kim. As a researcher, I aim to utilize statistical learning and stochastic control methods to address challenges inherent in data-driven decision-making, especially within service operations and revenue management applications. My recent research has focused on improving decision-making in complex information-sharing environments. (e.g., misinformation, strategic behaviors). Details can be found here.

Prior to my Ph.D., I obtained my Master of Applied Science degree in Operations Research from the University of Toronto and my Bachelor of Science degree in Physics and Mathematics from McGill University. You can find my CV here.

I am on the 2024-2025 academic job market. Below, you’ll discover information about my recent talks and what’s upcoming.

news

Oct 23, 2024	Diversified Learning @ INFORMS Annual Meeting, Seattle, USA. Room: Summit-336, Time: 8:30 AM.
Jun 30, 2024	Diversified Learning @ INFORMS MSOM 2024 Conference, Minneapolis, USA.
Jun 3, 2024	Diversified Learning and Service Allocation with Returns @ CORS Annual Conference, London, Canada
Apr 25, 2024	Diversified Learning @ POMS Annual Conference, Minneapolis, USA.

selected publications

2022

OR
Learning Manipulation Through Information Dissemination

J. Keppo, M. Kim, and X. Zhang

Operations Research, 2022

Abs Bib HTML

We study the optimal manipulation of a Bayesian learner through adaptive information provisioning, motivated by firms employing false advertising on e-commerce platforms and social networks. Specifically, we consider a firm that disseminates biased information to manipulate public beliefs about a parameter related to its payoffs. In our sequential optimization model, the firm dynamically decides on both the quantity and content of information sent to the public to maximize its expected total profits. By solving the Bayesian dynamic programming equation, we explicitly characterize the optimal manipulation strategy and also the evolution of public belief under such manipulation over time. We also consider the public as partially Bayesian social learners who utilize public reviews to mitigate manipulation. We show that the public eventually learns the truth from misinformation using social learning.
@article{keppoetal2023learnmanip, title = {Learning Manipulation Through Information Dissemination}, author = {Keppo, J. and Kim, M. and Zhang, X.}, journal = {Operations Research}, volume = {70}, number = {8}, pages = {3490-3510}, year = {2022} }