In tensor completion tasks, the traditional low-rank decomposition models suffer from laborious model selection problem due to their high sensitivity. particular, for ring (TR) decomposition, number of possibilities grows exponentially with order, which makes it rather challenging find optimal TR decomposition. this paper, by exploiting structure latent space, we propose a novel method is robust selection. contrast imposing constraint on data introduce nuclear norm regularization factors,...

Time-series remote sensing (RS) images are often corrupted by various types of missing information such as dead pixels, clouds, and cloud shadows that significantly influence the subsequent applications. In this paper, we introduce a new low-rank tensor decomposition model, termed ring (TR) decomposition, to analysis RS data sets propose TR completion method for reconstruction. The proposed model has ability utilize property time-series from different dimensions. To further explore...

Tensor train decomposition is a powerful representation for high-order tensors, which has been successfully applied to various machine learning tasks in recent years. In this paper, we study more generalized tensor with ring-structured network by employing circular multilinear products over sequence of lower-order core tensors. We refer such as ring (TR) representation. Our goal introduce algorithms including sequential singular value decompositions and blockwise alternating least squares...

The problem of incomplete data is common in signal processing and machine learning. Tensor completion algorithms aim to recover the from its partially observed entries. In this paper, taking advantages high compressibility flexibility recently proposed tensor ring (TR) decomposition, we propose a new approach named weighted optimization (TR-WOPT). It finds latent factors by gradient descent algorithm, then are employed predict missing entries tensor. We conduct various experiments on...

Genome-wide identification of the transcriptomic responses human cell lines to drug treatments is a challenging issue in medical and pharmaceutical research. However, drug-induced gene expression profiles are largely unknown unobserved for all combinations drugs lines, which serious obstacle practical applications.Here, we developed novel computational method predict parts various new therapeutic indications wide range diseases. We proposed tensor-train weighted optimization (TT-WOPT)...

In this paper, we aim at the problem of tensor data completion. Tensor-train decomposition is adopted because its powerful representation ability and linear scalability to order. We propose an algorithm named Sparse Optimization (STTO) which considers incomplete as sparse uses first-order optimization method find factors tensor-train decomposition. Our shown perform well in simulation experiments both low-order cases high-order cases. also employ a ten-sorization transform higher-order form...

Dimensionality reduction is an essential technique for multi-way large-scale data, i.e., tensor. Tensor ring (TR) decomposition has become popular due to its high representation ability and flexibility. However, the traditional TR algorithms suffer from computational cost when facing data. In this paper, taking advantages of recently proposed tensor random projection method, we propose two algorithms. By employing on every mode tensor, can be processed at a much smaller scale. The simulation...

The problem of incomplete data is common in signal processing and machine learning. Tensor completion algorithms aim to recover the from its partially observed entries. In this paper, taking advantages high compressibility flexibility recently proposed tensor ring (TR) decomposition, we propose a new approach named weighted optimization (TR-WOPT). It finds latent factors by gradient descent algorithm, then are employed predict missing entries tensor. We conduct various experiments on...

Low-rank matrix completion (LRMC) is a classical model in both computer vision (CV) and machine learning, has been successfully applied to various real applications. In the recent CV tasks, usually employed on variants of data, such as "non-local" or filtered, rather than their original forms. This fact makes that theoretical analysis conventional LRMC no longer suitable these To tackle this problem, we propose more general framework for LRMC, which linear transformations data are taken into...

In this paper, a tensor-based interpolation method for spatio-temporal field data is proposed. The observed are organized as sparse tensor, which then represented the decomposition of canonical polyadic (CP) model. latent factors, abstracting underlying structure data, exploited by introducing weighted optimization solution. With reconstruction these values missing can progressively be recovered. simulation experiments suggest that our interpolate with various levels sparseness. even...

In low-rank tensor completion tasks, due to the underlying multiple large-scale singular value decomposition (SVD) operations and rank selection problem of traditional methods, they suffer from high computational cost sensitivity model complexity. this paper, taking advantages compressibility recently proposed ring (TR) decomposition, we propose a new for problem. This is achieved through introducing convex surrogates assumption on latent factors, which makes it possible Schatten norm...

In recent studies, tensor ring (TR) decomposition has shown to be effective in data compression and representation. However, the existing TR-based completion methods only exploit global low-rank property of visual data. When applying them remote sensing (RS) image processing, spatial information RS is ignored. this paper, we introduce TR processing propose a method for reconstruction. We incorporate total-variation regularization into model continuity simultaneously. The proposed algorithm...

In tensor completion tasks, the traditional low-rank decomposition models suffer from laborious model selection problem due to their high sensitivity. particular, for ring (TR) decomposition, number of possibilities grows exponentially with order, which makes it rather challenging find optimal TR decomposition. this paper, by exploiting structure latent space, we propose a novel method is robust selection. contrast imposing constraint on data introduce nuclear norm regularization factors,...

Brain Computer Interface (BCI) aims to translate the brain signals, reflecting neural activities of evoked by external stimuli or mental tasks, into corresponding commands, which thus provides a direct communication between human and machine. P300 based BCI has demonstrated be one most reliable subject independent paradigm. However, existing only uses single modality, i. e., visual potential. In this paper, further improve reliability system, we develop hybrid using auditory stimulus...

In this paper, we aim at the completion problem of high order tensor data with missing entries. The existing factorization and methods suffer from curse dimensionality when N>>3. To overcome problem, propose an efficient algorithm called TT-WOPT (Tensor-train Weighted OPTimization) to find latent core tensors recover Tensor-train decomposition, which has powerful representation ability linear scalability order, is employed in our algorithm. experimental results on synthetic natural image...

In order to solve the problem of modelling, simulation and optimisation magnetorheological brake (MR brake) which is a multi-domain coupling system, braking performance control parameters MR were investigated under quarter-car model. Firstly, based on Modelica/MWorks platform, using unified modelling method, model with anti-lock system (ABS) was built. Then by response surface method (RSM), function express relation between distance three formulated, solved at MWorks. Finally, according...

In this paper, we aim at the problem of tensor data completion. Tensor-train decomposition is adopted because its powerful representation ability and linear scalability to order. We propose an algorithm named Sparse Optimization (STTO) which considers incomplete as sparse uses first-order optimization method find factors tensor-train decomposition. Our shown perform well in simulation experiments both low-order cases high-order cases. also employ a tensorization transform higher-order form...