Generalized Singular Value Decomposition Research Articles

Regularization of soft-X-ray imaging in the DIII-D tokamak

An image inversion scheme for the soft-X-ray imaging system (SXRIS) diagnostic at the DIII-D tokamak is developed to obtain the local soft-X-ray emission at a poloidal cross-section from the spatially line-integrated image taken by the SXRIS camera. The scheme uses the Tikhonov regularization method since the inversion problem is generally ill-posed. The regularization technique uses the generalized singular value decomposition (GSVD) to determine a solution that depends on a free regularization parameter. The latter has to be chosen carefully, and the so-called L-curve method to find the optimum regularization parameter is outlined. A representative test image is used to study the properties of the inversion scheme with respect to inversion accuracy, amount/strength of regularization, image noise and image resolution. The optimum inversion parameters are identified, while the L-curve method successfully computes the optimum regularization parameter. Noise is found to be the most limiting issue, but sufficient regularization is still possible at noise to signal ratios up to 10%–15%. Finally, the inversion scheme is applied to measured SXRIS data and the line-integrated SXRIS image is successfully inverted.

Maximum Likelihood Estimation of GEVD: Applications in Bioinformatics.

We propose a method, maximum likelihood estimation of generalized eigenvalue decomposition (MLGEVD) that employs a well known technique relying on the generalization of singular value decomposition (SVD). The main aim of the work is to show the tight equivalence between MLGEVD and generalized ridge regression. This relationship reveals an important mathematical property of GEVD in which the second argument act as prior information in the model. Thus we show that MLGEVD allows the incorporation of external knowledge about the quantities of interest into the estimation problem. We illustrate the importance of prior knowledge in clinical decision making/identifying differentially expressed genes with case studies for which microarray data sets with corresponding clinical/literature information are available. On all of these three case studies, MLGEVD outperformed GEVD on prediction in terms of test area under the ROC curve (test AUC). MLGEVD results in significantly improved diagnosis, prognosis and prediction of therapy response.

Rescaling the GSVD with application to ill-posed problems

The generalized singular value decomposition (GSVD) of a pair of matrices expresses each matrix as a product of an orthogonal, a diagonal, and a nonsingular matrix. The nonsingular matrix, which we denote by X T , is the same in both products. Available software for computing the GSVD scales the diagonal matrices and X T so that the squares of corresponding diagonal entries sum to one. This paper proposes a scaling that seeks to minimize the condition number of X T . The rescaled GSVD gives rise to new truncated GSVD methods, one of which is well suited for the solution of linear discrete ill-posed problems. Numerical examples show this new truncated GSVD method to be competitive with the standard truncated GSVD method as well as with Tikhonov regularization with regard to the quality of the computed approximate solution.

Simplified GSVD computations for the solution of linear discrete ill-posed problems

The generalized singular value decomposition (GSVD) often is used to solve Tikhonov regularization problems with a regularization matrix without exploitable structure. This paper describes how the ...

GSVD Beamforming for Two-User MIMO Downlink Channel

This paper introduces the fundamentals of generalized singular value decomposition (GSVD) beamforming: a noniterative beamforming technique based on GSVD for the two-user multiple-input-multiple-output (MIMO) downlink. The numbers of private/common channels produced by GSVD beamforming and transmit power normalization are investigated; the channel gains under Rayleigh fading are characterized exactly for configurations that involve only common channels. Moreover, symbol error rates (SERs) are presented for elementary multicasting and two-way relaying applications; the performance distinctions of the private/common channels and the effect of channel-estimation errors and asymmetries in channel fading are highlighted thereby.

Simplified GSVD computations for the solution of linear discrete ill-posed problems

The generalized singular value decomposition (GSVD) often is used to solve Tikhonov regularization problems with a regularization matrix without exploitable structure. This paper describes how the standard methods for the computation of the GSVD of a matrix pair can be simplified in the context of Tikhonov regularization. Also, other regularization methods, including truncated GSVD, are considered. We compare the computational efforts required by the simplified GSVD method and the A-weighted generalized inverse introduced by Eldén.

Divide and Conquer the CS Decomposition

We develop a divide-and-conquer algorithm for the bidiagonal CS decomposition (CSD). This complements an earlier algorithm based on simultaneous QR iteration. The new algorithm is designed to provide the efficiency gains of familiar divide-and-conquer algorithms on both serial and parallel architectures. The solution uses many components of existing algorithms, particularly the bidiagonal SVD algorithm of Gu and Eisenstat, but extra steps and reparameterizations are required to maintain orthogonality and consistent singular vectors, especially when the singular vectors are ill conditioned. The algorithm supports the stable computation of the generalized singular value decomposition (GSVD) in addition to the CSD.

Joint SVD-GSVD precoding technique and secrecy capacity lower bound for the MIMO relay wire-tap channel

We consider a problem of secure communications for the communication system consisting of multiple inputs for a source and a relay and multiple outputs for the relay, a destination and an eavesdropper. For the above-mentioned communication system, we establish a lower bound on the secrecy capacity at which secure communications between the source and the destination are attainable. We make use of the singular value decomposition (SVD) and its generalization to decompose the whole system into parallel independent channels. At the source, the generalized singular value decomposition (GSVD) is performed to simultaneously diagonalize the channel matrices of the relay and the destination and independently code across the resulting parallel channels. At the relay, the SVD is performed to beamform the signal towards the destination. The scalar case of what we are considering in this article has been investigated in previous literature, to prove that the introduction of a fourth party, the relay, in the wire-tap channel facilitates secure wireless communications. Our simulation results are in line with the scalar case’s and prove to be successful in achieving secrecy capacity where the conventional model failed, i.e., when no relay is introduced and the eavesdropper’s channel incurs as little noise as the legitimate receiver.

A Closed-Form Power Allocation and Signal Alignment for a Diagonalized MIMO Two-Way Relay Channel With Linear Receivers

A novel channel diagonalization scheme for an amplify-and-forward, multiple-input multiple-output (MIMO), two-way relay channel (TWRC) is proposed using generalized singular value decomposition (GSVD). Diagonalization of the MIMO TWRC is a sub-optimal approach that achieves two main purposes: reducing the computational complexity for optimizing the linear precoders at each node; and reducing the detection complexity at the source nodes by separating the multiple data streams. For the given diagonalized structure, we first align the entries of the diagonalized channels using a permutation to maximize a lower bound of average achievable sum rate (ASR), and a joint source-relay power allocation is then performed to maximize the ASR of the aligned TWRC; the overall problem is divided into two convex subproblems, the solutions to which are provided in closed-form. Our analysis for the proposed scheme underlines the benefits of acquiring channel state information. Simulation results demonstrate that the proposed GSVD-based relaying scheme, with the signal alignment and closed-form power allocation, significantly improves the ASR while retaining the diagonalized channel structure. In addition, the proposed scheme achieves the same level of ASR with much less computational complexity as compared to the iterative schemes.

General tensor decomposition, moment matrices and applications

The tensor decomposition addressed in this paper may be seen as a generalization of Singular Value Decomposition of matrices. We consider general multilinear and multihomogeneous tensors. We show how to reduce the problem to a truncated moment matrix problem and give a new criterion for flat extension of Quasi-Hankel matrices. We connect this criterion to the commutation characterization of border bases. A new algorithm is described. It applies for general multihomogeneous tensors, extending the approach on binary forms by J.J. Sylvester. An example illustrates the algebraic operations involved in this approach and how the decomposition can be recovered from eigenvector computation.

SLNR-Based Transmit Beamforming for MIMO Wiretap Channel

In this paper, a novel closed-form transmit beamforming scheme is proposed for multiple-input multiple-output (MIMO) wiretap channel. The proposed method follows the idea of maximizing signal-to-leakage-plus-noise ratio and does not need additional power allocation. An efficient approach is further developed to implement the generalized eigenvalue decomposition required by our method when n r < n t (n t , n r and n e denote the antenna numbers at the transmitter, the intended receiver and the eavesdropper, respectively), which has lower complexity than the existing approaches. Moreover, we identify some shortcomings of the generalized singular value decomposition (GSVD) based beamforming method, a common beamforming scheme for MIMO wiretap channel. Finally, simulation results show that the proposed method significantly outperforms GSVD-based method when n e < n t and has almost the same performance as GSVD-based method when n e ? n t .

DISCO-SCA and Properly Applied GSVD as Swinging Methods to Find Common and Distinctive Processes

BackgroundIn systems biology it is common to obtain for the same set of biological entities information from multiple sources. Examples include expression data for the same set of orthologous genes screened in different organisms and data on the same set of culture samples obtained with different high-throughput techniques. A major challenge is to find the important biological processes underlying the data and to disentangle therein processes common to all data sources and processes distinctive for a specific source. Recently, two promising simultaneous data integration methods have been proposed to attain this goal, namely generalized singular value decomposition (GSVD) and simultaneous component analysis with rotation to common and distinctive components (DISCO-SCA).ResultsBoth theoretical analyses and applications to biologically relevant data show that: (1) straightforward applications of GSVD yield unsatisfactory results, (2) DISCO-SCA performs well, (3) provided proper pre-processing and algorithmic adaptations, GSVD reaches a performance level similar to that of DISCO-SCA, and (4) DISCO-SCA is directly generalizable to more than two data sources. The biological relevance of DISCO-SCA is illustrated with two applications. First, in a setting of comparative genomics, it is shown that DISCO-SCA recovers a common theme of cell cycle progression and a yeast-specific response to pheromones. The biological annotation was obtained by applying Gene Set Enrichment Analysis in an appropriate way. Second, in an application of DISCO-SCA to metabolomics data for Escherichia coli obtained with two different chemical analysis platforms, it is illustrated that the metabolites involved in some of the biological processes underlying the data are detected by one of the two platforms only; therefore, platforms for microbial metabolomics should be tailored to the biological question.ConclusionsBoth DISCO-SCA and properly applied GSVD are promising integrative methods for finding common and distinctive processes in multisource data. Open source code for both methods is provided.

Palmprint and Face Multi-Modal Biometric Recognition Based on SDA-GSVD and Its Kernelization

When extracting discriminative features from multimodal data, current methods rarely concern themselves with the data distribution. In this paper, we present an assumption that is consistent with the viewpoint of discrimination, that is, a person's overall biometric data should be regarded as one class in the input space, and his different biometric data can form different Gaussians distributions, i.e., different subclasses. Hence, we propose a novel multimodal feature extraction and recognition approach based on subclass discriminant analysis (SDA). Specifically, one person's different bio-data are treated as different subclasses of one class, and a transformed space is calculated, where the difference among subclasses belonging to different persons is maximized, and the difference within each subclass is minimized. Then, the obtained multimodal features are used for classification. Two solutions are presented to overcome the singularity problem encountered in calculation, which are using PCA preprocessing, and employing the generalized singular value decomposition (GSVD) technique, respectively. Further, we provide nonlinear extensions of SDA based multimodal feature extraction, that is, the feature fusion based on KPCA-SDA and KSDA-GSVD. In KPCA-SDA, we first apply Kernel PCA on each single modal before performing SDA. While in KSDA-GSVD, we directly perform Kernel SDA to fuse multimodal data by applying GSVD to avoid the singular problem. For simplicity two typical types of biometric data are considered in this paper, i.e., palmprint data and face data. Compared with several representative multimodal biometrics recognition methods, experimental results show that our approaches outperform related multimodal recognition methods and KSDA-GSVD achieves the best recognition performance.

Feature extraction using fuzzy maximum margin criterion

In pattern recognition, feature extraction techniques are widely employed to reduce the dimensionality of date. In this paper, a novel feature extraction criterion, fuzzy maximum margin criterion (FMMC), is proposed by means of the maximum margin criterion (MMC) and fuzzy set theory. More specifically, the between-class and within-class fuzzy scatter matrices are redefined by incorporating the membership degrees of samples which relates the samples distribution information; then the feature extraction criterion maximized the average margin between classes after dimensionality reduction is applied. Furthermore, we utilize the generalized singular value decomposition (GSVD) to the criterion, which make the algorithm more effective; for nonlinear separated problems, we extend the kernel extension of FMMC with positive definite kernels. The effective of the novel criterion for linear and nonlinear separated problems is illustrated by experiments.

Structured penalties for functional linear models—partially empirical eigenvectors for regression

One of the challenges with functional data is incorporating geometric structure, or local correlation, into the analysis. This structure is inherent in the output from an increasing number of biomedical technologies, and a functional linear model is often used to estimate the relationship between the predictor functions and scalar responses. Common approaches to the problem of estimating a coefficient function typically involve two stages: regularization and estimation. Regularization is usually done via dimension reduction, projecting onto a predefined span of basis functions or a reduced set of eigenvectors (principal components). In contrast, we present a unified approach that directly incorporates geometric structure into the estimation process by exploiting the joint eigenproperties of the predictors and a linear penalty operator. In this sense, the components in the regression are 'partially empirical' and the framework is provided by the generalized singular value decomposition (GSVD). The form of the penalized estimation is not new, but the GSVD clarifies the process and informs the choice of penalty by making explicit the joint influence of the penalty and predictors on the bias, variance and performance of the estimated coefficient function. Laboratory spectroscopy data and simulations are used to illustrate the concepts.

A Higher-Order Generalized Singular Value Decomposition for Comparison of Global mRNA Expression from Multiple Organisms

The number of high-dimensional datasets recording multiple aspects of a single phenomenon is increasing in many areas of science, accompanied by a need for mathematical frameworks that can compare multiple large-scale matrices with different row dimensions. The only such framework to date, the generalized singular value decomposition (GSVD), is limited to two matrices. We mathematically define a higher-order GSVD (HO GSVD) for N≥2 matrices , each with full column rank. Each matrix is exactly factored as Di = UiΣiVT, where V, identical in all factorizations, is obtained from the eigensystem SV = VΛ of the arithmetic mean S of all pairwise quotients of the matrices , i≠j. We prove that this decomposition extends to higher orders almost all of the mathematical properties of the GSVD. The matrix S is nondefective with V and Λ real. Its eigenvalues satisfy λk≥1. Equality holds if and only if the corresponding eigenvector vk is a right basis vector of equal significance in all matrices Di and Dj, that is σi,k/σj,k = 1 for all i and j, and the corresponding left basis vector ui,k is orthogonal to all other vectors in Ui for all i. The eigenvalues λk = 1, therefore, define the “common HO GSVD subspace.” We illustrate the HO GSVD with a comparison of genome-scale cell-cycle mRNA expression from S. pombe, S. cerevisiae and human. Unlike existing algorithms, a mapping among the genes of these disparate organisms is not required. We find that the approximately common HO GSVD subspace represents the cell-cycle mRNA expression oscillations, which are similar among the datasets. Simultaneous reconstruction in the common subspace, therefore, removes the experimental artifacts, which are dissimilar, from the datasets. In the simultaneous sequence-independent classification of the genes of the three organisms in this common subspace, genes of highly conserved sequences but significantly different cell-cycle peak times are correctly classified.

SSHI methods for solving general linear matrix equations

PurposeThe purpose of this paper is to find the efficient iterative methods for solving the general matrix equation A1X+ XA2+A3XH+XHA4=B (including Lyapunov and Sylvester matrix equations as special cases) with the unknown complex (reflexive) matrix X.Design/methodology/approachBy applying the principle of hierarchical identification and the Hermitian/skew‐Hermitian splitting of the coefficient matrix quadruplet A1; A2; A3; A4 the authors propose a shift‐splitting hierarchical identification (SSHI) method to solve the general linear matrix equation A1X+XA2+A3XH+XHA4=B. Also, the proposed algorithm is extended for finding the reflexive solution to this matrix equation.FindingsThe authors propose two iterative methods for finding the solution and reflexive solution of the general linear matrix equation, respectively. The proposed algorithms have a simple, neat and elegant structure. The convergence analysis of the methods is also discussed. Some numerical results are given which illustrate the power and effectiveness of the proposed algorithms.Originality/valueSo far, several methods have been presented and used for solving the matrix equations by using vec operator and Kronecker product, generalized inverse, generalized singular value decomposition (GSVD) and canonical correlation decomposition (CCD) of matrices. In several cases, it is difficult to find the solutions by using matrix decomposition and generalized inverse. Also vec operator and Kronecker product enlarge the size of the matrix greatly therefore the computations are very expensive in the process of finding solutions. To overcome these complications and drawbacks, by using the hierarchical identification principle and the Hermitian=skew‐Hermitian splitting of the coefficient matrix quadruplet (A1; A2; A3; A4), the authors propose SSHI methods for solving the general matrix equation.

Integrated Analysis of Gene Expression and Copy Number Data on Gene Shaving Using Independent Component Analysis

DNA microarray gene expression and microarray-based comparative genomic hybridization (aCGH) have been widely used for biomedical discovery. Because of the large number of genes and the complex nature of biological networks, various analysis methods have been proposed. One such method is "gene shaving," a procedure which identifies subsets of the genes with coherent expression patterns and large variation across samples. Since combining genomic information from multiple sources can improve classification and prediction of diseases, in this paper we proposed a new method, "ICA gene shaving" (ICA, independent component analysis), for jointly analyzing gene expression and copy number data. First we used ICA to analyze joint measurements, gene expression and copy number, of a biological system and project the data onto statistically independent biological processes. Next, we used these results to identify patterns of variation in the data and then applied an iterative shaving method. We investigated the properties of our proposed method by analyzing both simulated and real data. We demonstrated that the robustness of our method to noise using simulated data. Using breast cancer data, we showed that our method is superior to the Generalized Singular Value Decomposition (GSVD) gene shaving method for identifying genes associated with breast cancer.

Efficient multi-user MIMO precoding based on GSVD and vector perturbation

We present an efficient precoding method by using generalized singular value decomposition (GSVD) and vector perturbation in the downlink of multi-user MIMO. Linear precoding/decoding matrices are explicitly computed through the GSVD algorithm, avoiding any matrix inversion or squaring operation. Accordingly, the computational complexity can be drastically reduced compared to the conventional iterative algorithms. In addition, by considering the inter-user interference, we derive a new perturbation vector that minimizes system-wide mean-squared-error of detected symbols. We also show that the proposed precoding scheme is suitable to update the precoding weights efficiently when the channel is newly updated, rather than recomputing the precoding weights for all users from scratch.

Cooperative Jamming for Secure Communications in MIMO Relay Networks

Secure communications can be impeded by eavesdroppers in conventional relay systems. This paper proposes cooperative jamming strategies for two-hop relay networks where the eavesdropper can wiretap the relay channels in both hops. In these approaches, the normally inactive nodes in the relay network can be used as cooperative jamming sources to confuse the eavesdropper. Linear precoding schemes are investigated for two scenarios where single or multiple data streams are transmitted via a decode-and-forward (DF) relay, under the assumption that global channel state information (CSI) is available. For the case of single data stream transmission, we derive closed-form jamming beamformers and the corresponding optimal power allocation. Generalized singular value decomposition (GSVD)-based secure relaying schemes are proposed for the transmission of multiple data streams. The optimal power allocation is found for the GSVD relaying scheme via geometric programming. Based on this result, a GSVD-based cooperative jamming scheme is proposed that shows significant improvement in terms of secrecy rate compared to the approach without jamming. Furthermore, the case involving an eavesdropper with unknown CSI is also investigated in this paper. Simulation results show that the secrecy rate is dramatically increased when inactive nodes in the relay network participate in cooperative jamming.