B0415
Title: Graphical ridge with sparsity in high-dimension
Authors: Azam Kheyri - University of Pretoria (South Africa) [presenting]
Andriette Bekker - University of Pretoria (South Africa)
Mohammad Arashi - Ferdowsi University of Mashhad (Iran)
Abstract: The focus is on the estimation of the precision matrix in a high-dimensional Gaussian graphical model. Since the accuracy of the maximum likelihood approach can be improved by penalization, we consider the elastic net type penalty, which combines the L1 and L2 penalties while taking the target matrix into estimation consideration. We suggest a novel estimator that combines alternative ridge and graphical lasso estimators to improve precision estimation. Numerical results support our findings and demonstrate the superior efficiency of our proposal compared to the alternatives.
