An Alternating Approach of Maximum Likelihood Estimation for Mixture of Multivariate Skew t-Distribution
치우친 다변량 t-분포 혼합모형에 대한 최우추정

Seung-Gu Kim
2014 Korean Journal of Applied Statistics  
The Exact-EM algorithm can conventionally fit a mixture of multivariate skew distribution. However, it suffers from highly expensive computational costs to calculate the moments of multivariate truncated tdistribution in E-step. This paper proposes a new SPU-EM method that adopts the AECM algorithm principle proposed by Meng and van Dyk (1997)'s to circumvent the multi-dimensionality of the moments. This method offers a shorter execution time than a conventional Exact-EM algorithm. Some experments are provided to show its effectiveness.
doi:10.5351/kjas.2014.27.5.819 fatcat:527keswasveq5jwmdd32rnt3by