B1103
Title: An efficient tensor regression for high-dimensional data
Authors: Yingying Zhang - East China Normal University (China) [presenting]
Abstract: Most currently used tensor regression models for high-dimensional data are based on Tucker decomposition, which has good properties but loses its efficiency in compressing tensors very quickly as the order of tensors increases, e.g., greater than four or five. However, for the simplest tensor autoregression in handling time series data, its coefficient tensor already has the order of six. The aim is to revise a newly proposed tensor train (TT) decomposition and then apply it to tensor regression to obtain a nice statistical interpretation. The new tensor regression can match the data with hierarchical structures and lead to a better interpretation of the data with factorial structures, which should be better fitted by models with Tucker decomposition. More importantly, the new tensor regression can be easily applied to the case with higher-order tensors since TT decomposition can compress the coefficient tensors much more efficiently. The methodology is also extended to tensor autoregression for time series data, and nonasymptotic properties are derived for the ordinary least squares estimations of both tensor regression and autoregression. A new algorithm is introduced to search for estimators, and its theoretical justification is also discussed. The theoretical and computational properties of the proposed methodology are verified by simulation studies, and the advantages over existing methods are illustrated by two real examples.