LOGIN ENGLISH

제목 [학술세미나] [학과세미나] 9월 18일(금) 학과세미나 안내
작성일 2020-09-11 08:12:41
내용 [학과세미나] 9월 18일(금) 학과세미나 안내

▪제목 : Statistical Inference for Models with Doubly-intractable Normalizing Constants
▪연사 : 진익훈 (연세대 응용통계학과)
▪일시 : 2020년 9월 18일(금) AM 11:00 – 12:00
▪진행 : ZOOM 활용 비대면 세미나
        (접속 링크 : https://snu-ac-kr.zoom.us/j/99462617399)
 

초 록
When a normalizing constant of the likelihood function does not have a close form and depends on the parameters, then it poses a major computational challenge because the normalizing constant cannot be cancelled in a Metropolis-Hasting algorithm. This computational problem is called as the doubly-intractable normalizing constant problem. Doubly-intractable distributions commonly arise in many complex statistical models in physics, epidemiology, ecology, social science, among other disciplines. To address the computational difficulties of the doubly-intractable normalizing constants, two approaches have been used: approximation-based method and auxiliary variable Markov chain Monte Carlo (MCMC) method. In this presentation, we first introduce several computational algorithms for the doubly-intractable normalizing constants with their advantages and disadvantages. If possible, we will illustrate some applications of the model with doubly-intractable normalizing constants; one is non-Gaussian Markov random field with applications to the dental caries researches with and the other is exponential random graph model in a network data analysis.
파일 세미나 안내_200918_진익훈.hwp [14.5KB]