Four Skewed Tensor Distributions

Abstract

With the rise of the "big data" phenomenon in recent years, data is coming in many different complex forms. One example of this is multi-way data that come in the form of higher-order tensors such as coloured images and movie clips. Although there has been a recent rise in models for looking at the simple case of three-way data in the form of matrices, there is a relative paucity of higher-order tensor variate methods. The most common tensor distribution in the literature is the tensor variate normal distribution; however, its use can be problematic if the data exhibit skewness or outliers. Herein, we develop four skewed tensor variate distributions which to our knowledge are the first skewed tensor distributions to be proposed in the literature, and are able to parameterize both skewness and tail weight. Properties and parameter estimation are discussed, and real and simulated data are used for illustration.

References in corpus (5)

• A Mixture of Generalized Hyperbolic Distributions
• Mixtures of Shifted Asymmetric Laplace Distributions
• Mixtures of Skew-t Factor Analyzers
• Covariance pattern mixture models for the analysis of multivariate heterogeneous longitudinal data
• Clustering Higher Order Data: An Application to Pediatric Multi-variable Longitudinal Data

Cited by in corpus (2)

• Matrix variate and tensor variate Laplace distributions
• Tensor distributions with covariance tensor or correlation tensor