Hsin-Hsiung Huang

A Bayesian Sparse Kronecker Product Decomposition Framework for Tensor Predictors With Mixed‐Type Responses: Applications to Neuroimaging Data Mining

The Bayesian sparse Kronecker product decomposition (BSKPD) is proposed, which represents a regression or classification coefficient tensor as a low‐rank sum of Kronecker products of sparse component tensors, and establishes identifiability and posterior consistency in both classical and high‐dimensional regimes.