연구세미나 안내(25.3.5(수) 10:30~12:00. 112동 2층 화상회의실/Research Seminar Invitation
2025.02.28- Date
- 2025-03-05 10:30:00
- Lecturer
- 김태호
- Venue
- 112동 2층 화상회의실
📅 세미나 정보
🔸일시 : 2025년 3월 5일(수) 10:30
🔸장소 : 112동 2층 화상회의실
🔸발표자 : Taeho Kim (HKUST)
🔸강연주제: Optimizing Input Data Collection for Ranking and Selection
– This seminar will be conducted in English
🔸Zoom: https://unist-ac-kr.zoom.us/j/85808732261?pwd=XYLebYOwP7lDdQpSCoLH8yJt56xHJ5.1
ID: 858 0873 2261
PW: unist1234
📑Abstract:
This work considers a class of simulation optimization problems, called ranking and selection, under input uncertainty when all solutions share a common Bayesian input distribution.
Given the data sampling budget, we aim to propose a sequential sampling procedure to find the best solution as efficiently as possible by balancing two types of sampling decisions:
gather additional input data from each input data source to update the Bayesian input model or simulate to learn each solution’s performance.
The most probable best (MPB) is adopted as an estimator of the best solution. Assuming a finite support of input parameters, we study the optimal sampling ratios for input collection and simulation.
Furthermore, an extension to a continuous-valued input parameter case is discussed by exploiting the kernel ridge regression.
Numerical studies are provided to support our findings.
👤Pesenter Bio:
Taeho Kim is a postdoctoral researcher in the HKUST Business School at the Hong Kong University of Science and Technology.
He earned my BS and Ph.D. from the Department of Mathematical Sciences at Korea Advanced Institute of Science and Technology (KAIST) in August 2015 and August 2021, respectively.
After that, we spent 3 years as a postdoc in the United States. His research interests lie in stochastic operations research, particularly in simulation optimization,
sequential decision making under uncertainty, and risk quantification in stochastic systems.