A tensor network low rank completion method

This paper proposes a tensor completion technique for recovering multidimensional network data. By combining tensor completion methods with side information, we are able to address multidimensional network data completion challenges. We formulate the missing data inference task as a tensor completion problem. Then, we employ a regularized non-negative tensor ring low-rank algorithm and the application of the Alternating Direction Method of Multipliers scheme in conjunction with the side information about the network. By exploiting side information, we improved the recovery accuracy and enabled more effective data inference. Some numerical results demonstrate the effectiveness of this approach.