A tensor completion method via Strassen-Ottaviani flattening and truncated nuclear norm
In this paper,a tensor completion method is proposed based on Strassen-Ottaviani flattening,which can reveal the underlying tensor rank intrinsically.The resulting tensor completion optimization problem is formulated by the nonconvex truncated nuclear norm as surrogate for the rank function.In order to solve this nonconvex optimization problem,the inexact augmented Lagrange method(IALM)is proposed and the global convergence is guaranteed.Numerical experiments on color images show the effectiveness of the proposed method.
NETL
NSTL
万方数据
