- This event has passed.
Graduate Student Seminar- Prof. Lee (UNC SDSS) and Prof. Chen (UNC STOR)
3 Nov @ 3:30 pm - 4:30 pm
Graduate Student Seminar- Prof. Lee (UNC SDSS) and Prof. Chen (UNC STOR)
3 Nov @ 3:30 pm – 4:30 pm
The Department of Statistics and Operations Research The University of North Carolina at Chapel Hill
Graduate Student Seminar
Friday, November 3 rd, 2023
125 Hanes Hall 3:30-4:30pm
or via Zoom:
https://unc.zoom.us/j/92731259806?pwd=eTc3Ylo0SWNFOHV4eERvSjRrUmlKQT09
Meeting ID: 927 3125 9806
Passcode: 100210
Harlin Lee (3.30-4 pm) UNC School of Data Science and Society
Blob Method for Optimal Transport
Abstract: Optimal transport (OT) aims to find the most efficient way of moving mass from one distribution to another with minimum cost. In this joint work with Katy Craig (UCSB) and Karthik Elamvazhuthi (UC Riverside), we apply the classical vortex blob method to dynamic OT, which results in a regularized and discretized optimization problem. This is shown to converge to the original problem formulation, with the added benefit of being able to flexibly handle state and control constraints. I will demonstrate a few numerical experiments on control theory, as well as potential application to sampling.
Guanting Chen (4-4.30 pm) UNC Chapel Hill – Statistics & Operations Research
Learning to Make Adherence-Aware Advice
Abstract: As artificial intelligence (AI) systems play an increasingly prominent role in human decision-making, challenges surface in the realm of human-AI interactions. One challenge arises from the suboptimal AI policies due to the inadequate consideration of humans disregarding AI recommendations, as well as the need for AI to provide advice selectively when it is most pertinent. This paper presents a sequential decision-making model that (i) takes into account the human’s adherence level (the probability that the human follows/rejects machine advice) and (ii) incorporates a defer option so that the machine can temporarily refrain from making advice. We provide learning algorithms that learn the optimal advice policy and make advice only at critical time stamps. Compared to problem-agnostic reinforcement learning algorithms, our specialized learning algorithms not only enjoy better theoretical convergence properties but also show strong empirical performance.