Probabilistic PCA on tensors

Yaoming Zhen, Piotr Zwiernik

In probabilistic principal component analysis (PPCA), an observed vector is modeled as a linear transformation of a low-dimensional Gaussian factor plus isotropic noise. We general…