A novel nonconvex approach to recover the low-tubal-rank tensor data: when t-SVD meets PSSV
share
In this paper we fix attention on a recently developed tensor decomposition scheme named tensor SVD (t-SVD). the t-SVD not only provides similar properties as the matrix case, but also convert the tensor tubal-rank minimization into matrix rank minimization in the Fourier domain. Generally, minimizing the tensor nuclear norm (TNN) may cause some bias. In this paper, to alleviate these bias phenomenon, we consider to minimize the proposed partial sum of the tensor nuclear norm (PSTNN) in place of the tensor nuclear norm. The novel PSTNN is used for the problems of tensor completion (TC) and tensor principal component analysis (TRPCA). The effectiveness of the proposed methods are conducted on the synthetic data and real world data, and experimental results reveal that the algorithm outperforms TNN based methods.
READ FULL TEXT