期刊
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
卷 34, 期 5, 页码 2451-2465出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TNNLS.2021.3106654
关键词
Tensors; Matrix decomposition; Merging; Automation; Minimization; Image color analysis; Visualization; Image; video recovery; low-rank; sparsity; tensor completion; tensor-ring (TR) decomposition
This article proposes a new low-rank sparse TR completion method by introducing Frobenius norm regularization on the latent space. The method is capable of exploiting the low rankness and sparsity of high-order tensors using the Frobenius norm of latent TR-cores. Experimental results demonstrate that the proposed method achieves better results compared to conventional TR-based completion methods and is robust even with increasing TR-rank.
Tensor-ring (TR) decomposition was recently studied and applied for low-rank tensor completion due to its powerful representation ability of high-order tensors. However, most of the existing TR-based methods tend to suffer from deterioration when the selected rank is larger than the true one. To address this issue, this article proposes a new low-rank sparse TR completion method by imposing the Frobenius norm regularization on its latent space. Specifically, we theoretically establish that the proposed method is capable of exploiting the low rankness and Kronecker-basis-representation (KBR)-based sparsity of the target tensor using the Frobenius norm of latent TR-cores. We optimize the proposed TR completion by block coordinate descent (BCD) algorithm and design a modified TR decomposition for the initialization of this algorithm. Extensive experimental results on synthetic data and visual data have demonstrated that the proposed method is able to achieve better results compared to the conventional TR-based completion methods and other state-of-the-art methods and, meanwhile, is quite robust even if the selected TR-rank increases.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据