We consider a synthetic aperture imaging configuration, such as synthetic aperture radar (SAR), where we want to first separate reflections from moving targets from those coming from a stationary background, and then to image separately the moving and the stationary reflectors. For this purpose, we introduce a representation of the data as a third order tensor formed from data coming from partially overlapping sub-apertures. We then apply a tensor robust principal component analysis (TRPCA) to the tensor data which separates them into the parts coming from the stationary and moving reflectors. Images are formed with the separated data sets. Our analysis shows a distinctly improved performance of TRPCA, compared to the usual matrix case. In particular, the tensor decomposition can identify motion features that are undetectable when using the conventional motion estimation methods, including matrix RPCA. We illustrate the performance of the method with numerical simulations in the X-band radar regime.
Spectrum cartography aims at estimating power propagation patterns over a geographical region across multiple frequency bands (i.e., a radio map)---from limited samples taken sparsely over the region. Classic cartography methods are mostly concerned with recovering the aggregate radio frequency (RF) information while ignoring the constituents of the radio map---but fine-grained emitter-level RF information is of great interest. In addition, many existing cartography methods work explicitly or implicitly assume random spatial sampling schemes that may be difficult to implement, due to legal/privacy/security issues. The theoretical aspects (e.g., identifiability of the radio map) of many existing methods are also unclear. In this work, we propose a joint radio map recovery and disaggregation method that is based on coupled block-term tensor decomposition. Our method guarantees identifiability of the individual radio map of \textit{each emitter} (thereby that of the aggregate radio map as well), under realistic conditions. The identifiability result holds under a large variety of geographical sampling patterns, including a number of pragmatic systematic sampling strategies. We also propose effective optimization algorithms to carry out the formulated radio map disaggregation problems. Extensive simulations are employed to showcase the effectiveness of the proposed approach.
Hyperspectral super-resolution (HSR) aims at fusing a hyperspectral image (HSI) and a multispectral image (MSI) to produce a super-resolution image (SRI). Recently, a coupled tensor factorization approach was proposed to handle this challenging problem, which admits a series of advantages over the classic matrix factorization-based methods. In particular, modeling the HSI and MSI as low-rank tensors following the {\it canonical polyadic decomposition} (CPD) model, the approach is able to provably identify the SRI, under some mild conditions. However, the latent factors in the CPD model have no physical meaning, which makes utilizing prior information of spectral images as constraints or regularizations difficult---but using such information is often important in practice, especially when the data is noisy. In this work, we propose an alternative coupled tensor decomposition approach, where the HSI and MSI are assumed to follow the {\it block-term decomposition (BTD)} model. Notably, the new method also entails identifiability of the SRI under realistic conditions. More importantly, when modeling a spectral image as a BTD tensor, the latent factors have clear physical meaning, and thus prior knowledge about spectral images can be naturally incorporated. Simulations using real hyperspectral images are employed to showcase the effectiveness of the proposed approach with nonnegativity constraints.
Tensor decomposition (TD) is widely used in hyperspectral image (HSI) compression. The initialization of factor matrix in tensor decomposition can determine the HSI compression performance. It is worth noting that HSI is highly correlated in bands. However, this phenomenon is ignored by the previous TD method. Aiming at improving the HSI compression performance, we propose a method called correlation-based TD initialization algorithm. As HSI is well approximated by means of a reference band. In accordance with the SVD result of the reference band, the initialized factor matrices of TD are produced. We compare our methods with random and SVD-based initialization methods. The experimental results reveal that our correlation-based TD initialization method is capable of significantly reducing the computational cost of TD while keeping the initialization quality and compression performance.
A new deep learning-based electroencephalography (EEG) signal analysis framework is proposed. While deep neural networks, specifically convolutional neural networks (CNNs), have gained remarkable attention recently, they still suffer from high dimensionality of the training data. Two-dimensional input images of CNNs are more vulnerable to be redundant versus one-dimensional input time-series of conventional neural networks. In this study, we propose a new dimensionality reduction framework for reducing the dimension of CNN inputs based on the tensor decomposition of the time-frequency representation of EEG signals. The proposed tensor decomposition-based dimensionality reduction algorithm transforms a large set of slices of the input tensor to a concise set of slices which are called super-slices. Employing super-slices not only handles the artifacts and redundancies of the EEG data but also reduces the dimension of the CNNs training inputs. We also consider different time-frequency representation methods for EEG image generation and provide a comprehensive comparison among them. We test our proposed framework on HCB-MIT data and as results show our approach outperforms other previous studies.
Spatial channel covariance information can replace full instantaneous channel state information for the analog precoder design in hybrid analog/digital architectures. Obtaining spatial channel covariance estimation, however, is challenging in the hybrid structure due to the use of fewer radio frequency (RF) chains than the number of antennas. In this paper, we propose a spatial channel covariance estimation method based on higher-order tensor decomposition for spatially sparse time-varying frequency-selective channels. The proposed method leverages the fact that the channel can be represented as a low-rank higher-order tensor. We also derive the Cram\'er-Rao lower bound on the estimation accuracy of the proposed method. Numerical results and theoretical analysis show that the proposed tensor-based approach achieves higher estimation accuracy in comparison with prior compressive-sensing-based approaches or conventional angle-of-arrival estimation approaches. Simulation results reveal that the proposed approach becomes more beneficial at low signal-to-noise (SNR) region.
Muscle synergies have recently been utilised in myoelectric control systems. Thus far, all proposed synergy-based systems rely on matrix factorisation methods. However, this is limited in terms of task-dimensionality. Here, the potential application of higher-order tensor decomposition as a framework for proportional myoelectric control is demonstrated. A novel constrained Tucker decomposition (consTD) technique of synergy extraction is proposed for synergy-based myoelectric control model and compared with state-of-the-art matrix factorisation models. The extracted synergies were used to estimate control signals for the wrist's Degree of Freedom (DoF) through direct projection. The consTD model was able to estimate the control signals for each DoF by utilising all data in one 3rd-order tensor. This is contrast with matrix factorisation models where data are segmented for each DoF and then the synergies often have to be realigned. Moreover, the consTD method offers more information by providing additional shared synergies, unlike matrix factorisation methods. The extracted control signals were fed to a ridge regression to estimate the wrist's kinematics based on real glove data. The Coefficient of Determination (R2) for the reconstructed wrist position showed that the proposed consTD was higher than matrix factorisation methods. In sum, this study provides the first proof of concept for the use of higher-order tensor decomposition in proportional myoelectric control and it highlights the potential of tensors to provide an objective and direct approach to identify synergies.
Complex networks have now become integral parts of modern information infrastructures. This paper proposes a user-centric method for detecting anomalies in heterogeneous information networks, in which nodes and/or edges might be from different types. In the proposed anomaly detection method, users interact directly with the system and anomalous entities can be detected through queries. Our approach is based on tensor decomposition and clustering methods. We also propose a network generation model to construct synthetic heterogeneous information network to test the performance of the proposed method. The proposed anomaly detection method is compared with state-of-the-art methods in both synthetic and real-world networks. Experimental results show that the proposed tensor-based method considerably outperforms the existing anomaly detection methods.
Prior studies have proposed methods to recover multi-channel electroencephalography (EEG) signal ensembles from their partially sampled entries. These methods depend on spatial scenarios, yet few approaches aiming to a temporal reconstruction with lower loss. The goal of this study is to retrieve the temporal EEG signals independently which was overlooked in data pre-processing. We considered EEG signals are impinging on tensor-based approach, named nonlinear Canonical Polyadic Decomposition (CPD). In this study, we collected EEG signals during a resting-state task. Then, we defined that the source signals are original EEG signals and the generated tensor is perturbed by Gaussian noise with a signal-to-noise ratio of 0 dB. The sources are separated using a basic non-negative CPD and the relative errors on the estimates of the factor matrices. Comparing the similarities between the source signals and their recovered versions, the results showed significantly high correlation over 95%. Our findings reveal the possibility of recoverable temporal signals in EEG applications.
Higher-order tensor decompositions have hardly been used in muscle activity analysis despite multichannel electromyography (EMG) datasets naturally occurring as multi-way structures. Here, we seek to demonstrate and discuss the potential of tensor decompositions as a framework to estimate muscle synergies from $3^{rd}$-order EMG tensors built by stacking repetitions of multi-channel EMG for several tasks. We compare the two most widespread tensor decomposition models -- Parallel Factor Analysis (PARAFAC) and Tucker -- in muscle synergy analysis of the wrist's three main Degree of Freedoms (DoFs) using the public first Ninapro database. Furthermore, we proposed a constrained Tucker decomposition (consTD) method for efficient synergy extraction building on the power of tensor decompositions. This method is proposed as a direct novel approach for shared and task-specific synergy estimation from two biomechanically related tasks. Our approach is compared with the current standard approach of repetitively applying non-negative matrix factorisation (NMF) to a series of movements. The results show that the consTD method is suitable for synergy extraction compared to PARAFAC and Tucker. Moreover, exploiting the multi-way structure of muscle activity, the proposed methods successfully identified shared and task-specific synergies for all three DoFs tensors. These were found to be robust to disarrangement with regard to task-repetition information, unlike the commonly used NMF. In summary, we demonstrate how to use tensors to characterise muscle activity and develop a new consTD method for muscle synergy extraction that could be used for shared and task-specific synergies identification. We expect that this study will pave the way for the development of novel muscle activity analysis methods based on higher-order techniques.
The 3GPP suggests to combine dual polarized (DP) antenna arrays with the double directional (DD) channel model for downlink channel estimation. This combination strikes a good balance between high-capacity communications and parsimonious channel modeling, and also brings limited feedback schemes for downlink channel state information within reach---since such channel can be fully characterized by several key parameters. However, most existing channel estimation work under the DD model has not yet considered DP arrays, perhaps because of the complex array manifold and the resulting difficulty in algorithm design. In this paper, we first reveal that the DD channel with DP arrays at the transmitter and receiver can be naturally modeled as a low-rank tensor, and thus the key parameters of the channel can be effectively estimated via tensor decomposition algorithms. On the theory side, we show that the DD-DP parameters are identifiable under very mild conditions, by leveraging identifiability of low-rank tensors. Furthermore, a compressed tensor decomposition algorithm is developed for alleviating the downlink training overhead. We show that, by using judiciously designed pilot structure, the channel parameters are still guaranteed to be identified via the compressed tensor decomposition formulation even when the size of the pilot sequence is much smaller than what is needed for conventional channel identification methods, such as linear least squares and matched filtering. Numerical simulations are presented to showcase the effectiveness of the proposed methods.
The widespread use of multisensor technology and the emergence of big data sets have brought the necessity to develop more versatile tools to represent higher-order data with multiple aspects and high dimensionality. Data in the form of multidimensional arrays, also referred to as tensors, arises in a variety of applications including chemometrics, hyperspectral imaging, high resolution videos, neuroimaging, biometrics, and social network analysis. Early multiway data analysis approaches reformatted such tensor data as large vectors or matrices and then resorted to dimensionality reduction methods developed for classical two-way analysis such as PCA. However, one cannot discover hidden components within multiway data using conventional PCA. To this end, tensor decomposition methods which are flexible in the choice of the constraints and that extract more general latent components have been proposed. In this paper, we review the major tensor decomposition methods with a focus on problems targeted by classical PCA. In particular, we present tensor methods that aim to solve three important challenges typically addressed by PCA: dimensionality reduction, i.e. low-rank tensor approximation, supervised learning, i.e. learning linear subspaces for feature extraction, and robust low-rank tensor recovery. We also provide experimental results to compare different tensor models for both dimensionality reduction and supervised learning applications.
The most unique advantage of multipass SAR interferometry (InSAR) is the retrieval of long term geophysical parameters, e.g. linear deformation rates, over large areas. Recently, an object-based multipass InSAR framework has been proposed in [1], as an alternative to the typical single-pixel methods, e.g. Persistent Scatterer Interferometry (PSI), or pixel-cluster-based methods, e.g. SqueeSAR. This enables the exploitation of inherent properties of InSAR phase stacks on an object level. As a followon, this paper investigates the inherent low rank property of such phase tensors, and proposes a Robust Multipass InSAR technique via Object-based low rank tensor decomposition (RoMIO). We demonstrate that the filtered InSAR phase stacks can improve the accuracy of geophysical parameters estimated via conventional multipass InSAR techniques, e.g. PSI, by a factor of ten to thirty in typical settings. The proposed method is particularly effective against outliers, such as pixels with unmodeled phases. These merits in turn can effectively reduce the number of images required for a reliable estimation. The promising performance of the proposed method is demonstrated using high-resolution TerraSAR-X image stacks.
We provide a high fidelity deep learning algorithm (HyperSeg) for interactive video segmentation tasks using a convolutional network with context-aware skip connections, and compressed, hypercolumn image features combined with a convolutional tessellation procedure. In order to maintain high output fidelity, our model crucially processes and renders all image features in high resolution, without utilizing downsampling or pooling procedures. We maintain this consistent, high grade fidelity efficiently in our model chiefly through two means: (1) We use a statistically-principled tensor decomposition procedure to modulate the number of hypercolumn features and (2) We render these features in their native resolution using a convolutional tessellation technique. For improved pixel level segmentation results, we introduce a boundary loss function; for improved temporal coherence in video data, we include temporal image information in our model. Through experiments, we demonstrate the improved accuracy of our model against baseline models for interactive segmentation tasks using high resolution video data. We also introduce a benchmark video segmentation dataset, the VFX Segmentation Dataset, which contains over 27,046 high resolution video frames, including greenscreen and various composited scenes with corresponding, hand crafted, pixel level segmentations. Our work presents an extension to improvement to state of the art segmentation fidelity with high resolution data and can be used across a broad range of application domains, including VFX pipelines and medical imaging disciplines.
In this paper we show that simple semidefinite programs inspired by degree $4$ SOS can exactly solve the tensor nuclear norm, tensor decomposition, and tensor completion problems on tensors with random asymmetric components. More precisely, for tensor nuclear norm and tensor decomposition, we show that w.h.p. these semidefinite programs can exactly find the nuclear norm and components of an $(n\times n\times n)$-tensor $\mathcal{T}$ with $m\leq n^{3/2}/polylog(n)$ random asymmetric components. Unlike most of the previous algorithms, our algorithm provides a certificate for the decomposition, does not require knowledge about the number of components in the decomposition and does not make any assumptions on the sizes of the coefficients in the decomposition. As a byproduct, we show that w.h.p. the nuclear norm decomposition exactly coincides with the minimum rank decomposition for tensors with $m\leq n^{3/2}/polylog(n)$ random asymmetric components. For tensor completion, we show that w.h.p. the semidefinite program, introduced by Potechin & Steurer (2017) for tensors with orthogonal components, can exactly recover an $(n\times n\times n)$-tensor $\mathcal{T}$ with $m$ random asymmetric components from only $n^{3/2}m polylog(n)$ randomly observed entries. For non-orthogonal tensors, this improves the dependence on $m$ of the number of entries needed for exact recovery over all previously known algorithms and provides the first theoretical guarantees for exact tensor completion in the overcomplete regime.
How can we expand the tensor decomposition to reveal a hierarchical structure of the multi-modal data in a self-adaptive way? Current tensor decomposition provides only a single layer of clusters. We argue that with the abundance of multimodal data and time-evolving networks nowadays, the ability to identify emerging hierarchies is important. To this effect, we propose RecTen, a recursive hierarchical soft clustering approach based on tensor decomposition. Our approach enables us to: (a) recursively decompose clusters identified in the previous step, and (b) identify the right conditions for terminating this process. In the absence of proper ground truth, we evaluate our approach with synthetic data and test its sensitivity to different parameters. We also apply RecTen on five real datasets which involve the activities of users in online discussion platforms, such as security forums. This analysis helps us reveal clusters of users with interesting behaviors, including but not limited to early detection of some real events like ransomware outbreaks, the emergence of a blackmarket of decryption tools, and romance scamming. To maximize the usefulness of our approach, we develop a tool which can help the data analysts and overall research community by identifying hierarchical structures. RecTen is an unsupervised approach which can be used to take the pulse of the large multi-modal data and let the data discover its own hidden structures by itself.
Knowledge graphs have been demonstrated to be an effective tool for numerous intelligent applications. However, a large amount of valuable knowledge still exists implicitly in the knowledge graphs. To enrich the existing knowledge graphs, recent years witness that many algorithms for link prediction and knowledge graphs embedding have been designed to infer new facts. But most of these studies focus on the static knowledge graphs and ignore the temporal information that reflects the validity of knowledge. Developing the model for temporal knowledge graphs completion is an increasingly important task. In this paper, we build a new tensor decomposition model for temporal knowledge graphs completion inspired by the Tucker decomposition of order 4 tensor. We demonstrate that the proposed model is fully expressive and report state-of-the-art results for several public benchmarks. Additionally, we present several regularization schemes to improve the strategy and study their impact on the proposed model. Experimental studies on three temporal datasets (i.e. ICEWS2014, ICEWS2005-15, GDELT) justify our design and demonstrate that our model outperforms baselines with an explicit margin on link prediction task.
Processing sentence constituency trees in binarised form is a common and popular approach in literature. However, constituency trees are non-binary by nature. The binarisation procedure changes deeply the structure, furthering constituents that instead are close. In this work, we introduce a new approach to deal with non-binary constituency trees which leverages tensor-based models. In particular, we show how a powerful composition function based on the canonical tensor decomposition can exploit such a rich structure. A key point of our approach is the weight sharing constraint imposed on the factor matrices, which allows limiting the number of model parameters. Finally, we introduce a Tree-LSTM model which takes advantage of this composition function and we experimentally assess its performance on different NLP tasks.
Feature selection by maximizing high-order mutual information between the selected feature vector and a target variable is the gold standard in terms of selecting the best subset of relevant features that maximizes the performance of prediction models. However, such an approach typically requires knowledge of the multivariate probability distribution of all features and the target, and involves a challenging combinatorial optimization problem. Recent work has shown that any joint Probability Mass Function (PMF) can be represented as a naive Bayes model, via Canonical Polyadic (tensor rank) Decomposition. In this paper, we introduce a low-rank tensor model of the joint PMF of all variables and indirect targeting as a way of mitigating complexity and maximizing the classification performance for a given number of features. Through low-rank modeling of the joint PMF, it is possible to circumvent the curse of dimensionality by learning principal components of the joint distribution. By indirectly aiming to predict the latent variable of the naive Bayes model instead of the original target variable, it is possible to formulate the feature selection problem as maximization of a monotone submodular function subject to a cardinality constraint - which can be tackled using a greedy algorithm that comes with performance guarantees. Numerical experiments with several standard datasets suggest that the proposed approach compares favorably to the state-of-art for this important problem.
Anomaly detection in spatiotemporal data is a challenging problem encountered in a variety of applications including hyperspectral imaging, video surveillance, and urban traffic monitoring. Existing anomaly detection methods are most suited for point anomalies in sequence data and cannot deal with temporal and spatial dependencies that arise in spatiotemporal data. In recent years, tensor-based methods have been proposed for anomaly detection to address this problem. These methods rely on conventional tensor decomposition models, not taking the structure of the anomalies into account, and are supervised or semi-supervised. We introduce an unsupervised tensor-based anomaly detection method that takes the sparse and temporally continuous nature of anomalies into account. In particular, the anomaly detection problem is formulated as a robust lowrank + sparse tensor decomposition with a regularization term that minimizes the temporal variation of the sparse part, so that the extracted anomalies are temporally persistent. We also approximate rank minimization with graph total variation minimization to reduce the complexity of the optimization algorithm. The resulting optimization problem is convex, scalable, and is shown to be robust against missing data and noise. The proposed framework is evaluated on both synthetic and real spatiotemporal urban traffic data and compared with baseline methods.