2014년 제6회 통계세미나 개최 안내
- BK 세미나 시리즈 -
통계연구소에서는 다음과 같이 통계 세미나를 개최하오니 많은 참여 바랍니다.
일시: 2014년 4월 22일 (화) 오후 12시
장소: 고려대학교 정경관 618호
연사: Professor Xinyuan Song (Chinese University of Hong Kong)
Recent Development of Nonparametric
Latent Variable Models
<Abstract>
In the behavioral, social, psychological, and medical sciences, latent variables represent unobservable traits that are measured by multiple observed variables. Latent variable models are useful tools for assessing the interrelationships among latent and observed variables. Due to their wide applications, latent variable models have attracted significant attention from various fields. However, the majority of the existing works in the latent variable modeling are parametric. In this talk I will introduce several recent works related to nonparametric latent variable models, including nonparametric structural equation models, transformation latent variable models, and varying-coefficient latent variable models. These models can be used to reveal the true relationships among latent and observed variables, analyze non-normally distributed multidimensional data, and examine time-varying eects of explanatory latent and observed variables on outcome latent variables. The Bayesian P-splines approach, together with Markov chain Monte Carlo algorithms, is developed to estimate unknown smooth functions, unknown parameters, and latent variables in the models. The methodologies developed were applied to the medical studies related to prevention of osteoporosis fracture and intervention treatment of ploydrug use.
고려대학교 통계연구소