반복영역 바로가기
주메뉴로 바로가기
좌측메뉴로 바로가기
본문으로 바로가기

Home News&ActivityNotices

Notices

프린트페이스북

[대학원생 마일리지 적용] 학과 세미나 안내 (5/14(월) 펜실베니아 주립대 송은혜 교수/ May 14th at 4pm, Prof. Eunhye Song at Penn State University)

2018.05.15 09:13

5월 14일(월)에 ISysE 학사/석사 졸업생이면서 현재 펜실베니아 주립대학 산업공학과에

재직하고 있는 송은혜 교수님이 세마나를 하실 예정입니다.

 

송은혜 교수님은 시뮬레이션 관련한 연구를 수행해왔으며, 관련한 구체적인 내용들은

첨부한 CV를 참조해 주십시오. 많은 참여를 부탁드립니다.

 

On 14th May (Mon), Prof. Eunhye Song is going to give a seminar on simulation optimization.   
   She received BS and MS from our department. She is currently an assistant professor at Penn State University, IE.   
   Please see the attached cv for her research projects and others.   
      
      
Large-scale Discrete Simulation Optimization    
using Multi-resolution Gaussian Markov Random Fields     
    
5.14 (월) 산업경영동(E2-2) 멀티미디어실(2501) , 4:00 PM ~ 5:30 PM   
        
In this talk, we discuss solving a discrete simulation optimization problem with    
a combinatorially large solution space. Gaussian Markov Improvement Algorithm (GMIA)    
is an adaptive random search based on a Gaussian Markov random field (GMRF)    
metamodel. At each iteration, GMIA updates the conditional distribution of the GMRF    
based on the simulated solutions and selects the next solution to simulate based on    
the inference provided by the conditional distribution. GMIA is globally convergent    
to the optimal solution when the simulation budget increases to infinity and shows    
excellent empirical finite-sample performance as it 1) simulates only a small fraction    
of the solution space, and 2) stops correctly when the desired optimality gap is achieved.    
When the dimension of the solution space is large, there is a computational limitation    
to the GMIA, which led us to devise multi-resolution (MR-GMIA) based on multi-   
resolution GMRFs. The solution space is divided into subregions, where each subregion   
 is modeled by a solution-level GMRF and the subregions become “solutions” to the    
region-level GMRF. The MR-GMIA inherits nice statistical properties of GMIA including    
global convergence and increases the efficiency of the search by saving computational    
effort. Some ongoing research topics related to the hyperparameter estimation    
and parallelization MR-GMIA will be discussed briefly.     
        
        
ㅇ 연사:     
    Penn State University, 송은혜 교수 (Eunhye Song)       
    https://sites.google.com/view/eunhyesongphd/home                    
       
ㅇ 약력:    
    첨부파일 참조 (please see the attached file.) 
 
 
 
 

List