Generalized Mahalanobis distance and its application in detecting matrix outliers

Abstract

In this paper, a new distance for matrix observations called generalized Mahalanobis distance is introduced, some of its properties are investigated, and its distribution is obtained for the observations of the matrix variate elliptically contoured distributions. Also, as a significant application, the introduced
distance is used in detecting matrix outliers and two examples are provided for illustrative purposes as well.