Column-Wise Element Selection for Computationally Efficient Nonnegative Coupled Matrix Tensor Factorization

Coupled Matrix Tensor Factorization (CMTF) facilitates the integration and analysis of multiple data sources helps discover meaningful information. Nonnegative CMTF (N-CMTF) has been employed in many applications for identifying latent patterns, prediction, recommendation. However, due to added complexity with coupling between tensor matrix data, existing N-CMTF algorithms exhibit poor computation efficiency. In this paper, a computationally efficient factorization algorithm is presented based on column-wise element selection, preventing frequent gradient updates. Theoretical empirical analyses show that proposed not only more accurate but also than approximating as well underlying nature factors.