Fast and Robust Sparsity-Aware Block Diagonal Representation

Similarity matrix represents the relationship of n objects and gives us a useful information about these objects. Several models for analyzing this data assume that each object of n objects is embedded as a point or a vector in t dimensional “common space” of n objects in general. However, these models are not appropriate for analyzing a sparse block diagonal similarity matrix as each block diagonal matrix indicates us that each member of the set of objects in a block is represented as a point or a vector in not “common space,” but, “sub-space.” And a model is proposed to analyze this type of a sparse block diagonal similarity matrix. And application to a real data set will be shown.

Robust structure-preservation tensorized representation for multi-view unsupervised feature selection.

Color face image is composed of color information of different channels. Color image provides more information for face recognition task compared with grey scale image. A novel approach coined block diagonal non-negative matrix factorization (BDNMF) is proposed for color face representation and recognition. The approach employs block diagonal matrix to encode color information of different channels. Block diagonal constraint is imposed on the non-negative matrix factorization algorithm to factorize matrices of different channels simultaneously. And block diagonal non-negative matrix factorization algorithm is exploited to extract facial features. Nearest neighborhood classifier is adopted to identify color face samples. Experimental results on CVL and CMU PIE color face databases verify the effectiveness of the proposed approach.

The Drosophila melanogaster RNA-induced silencing complex (RISC) forms a large ribonucleoprotein particle on small interfering RNAs (siRNAs) and catalyzes target mRNA cleavage during RNA interference (RNAi). Dicer-2, R2D2, Loquacious, and Argonaute-2 are examples of RISC-associated factors that are involved in RNAi. Holo-RISC is an approximately 80 S small interfering ribonucleoprotein, which suggests that there are many additional proteins that participate in the RNAi pathway. In this study, we used siRNA affinity capture combined with mass spectrometry to identify novel components of the Drosophila RNAi machinery. Our study identified both established RISC components and novel siRNA-associated factors, many of which contain domains that are consistent with potential roles in RNAi. Functional analysis of these novel siRNA-associated proteins suggests that these factors may play an important role in RNAi.

For the low-rank representation-based subspace clustering, the affinity matrix is block diagonal. In this paper, a novel robust discriminant low-rank representation (RDLRR) algorithm is proposed to enhance the block diagonal property to explore the multiple subspace structures of samples. In order to cluster samples into their corresponding subspace and remove outliers, the proposed RDLRR considers both the within-class and the between-class distance during seeking the lowest-rank representation of samples. RDLRR could well indicate the global structure of samples, when the labeling is available. We conduct experiments on several datasets, including the Extended Yale B, AR and Hopkins 155, to show that our approach outperforms all the other state-of-the-art approaches.

Many vibrating systems involve interactions only between neighbouring parts of the system. Frequently, such systems are analysed by supposing that the mass distribution is lumped at the generalized coordinates. Undamped systems of this type involve a block tri-diagonal stiffness matrix K and a block diagonal inertia matrix M. The inverse problem is to reconstruct K and M from frequency response data. The known reconstruction of a Jacobi matrix using the Forsythe algorithm is generalised so that a block Jacobi matrix can be reconstructed from a certain spectral function. This analysis is used to construct the dynamic stiffness matrix A = L -1 KL -T (M = LL T ) from the frequency response; A is determined to within a block diagonal orthogonal matrix. It is shown that K and M can be separated using special information about their forms. A vibrating lattice composed of rods and masses is used as an example.

1. Introduction In recent years there has been an increased interest in input-output analysis and its multiple applications to regional and interregional analysis. The purpose of this study is to provide a cursory description of the results of estimating trade coefficients and interregional multipliers by means of a statistical implementation of the Chenery--Moses model for Greece. In order to carry out the study in this field, the estimation of interregional trade coefficients by commodities is the fundamental work and is an important part for the compilation of interregional input-output table (Thalassinos 2007). But as most countries in the world, even developed countries, Greece does not have enough statistical data to build such matrix. The author has constructed by own forces an interregional model for the year 2010 of the Greek economy, which fill the gap of official regional statistics. The evaluation of the impacts of an exogenous change on a regional system would require a model that incorporates the industrial interrelations of a region with another region. The impacts were simulated considering three producing regions for the Greek economy, North Greece, Center Greece and Aegean Islands and Crete. The constructed interregional input-output matrix was used considering a set of three regions and ten aggregated sectors. From this matrix interregional-interdependence matrix was computing to show the direct and indirect relationships between producing sectors and regions. Finally, interregional multipliers were computed for each sector and region measured the impact on regions and on the economy as a whole. The above procedure revealed the major producing sector in the nations. In this regard the attempt of the present study to introduce regional input-output tables and regional technical coefficients would be beneficial to interregional model. The rest of the paper is organized as follows: In the second section a short overview of the literature review was provided. In section three we will describe the research methodology of interregional model. Section four is devoted to an empirical illustration. We apply the estimation procedures and data sources in the empirical implementation of the model. A concluding section finalized the paper. 2. Literature Review Isard (1951) was the first writer who developed the interregional input-output model with an analytical interregional trade coefficient matrix. This model requires an extent size of basic data. As a result one of the most powerful model was put forward by Chenery (1953) and then Moses (1955) independently and it is called Chenery-Moses model. The relative advantage of this model due to the fact that the technical input structure of production in each region and interregional trade structure of various products are separately stated. Therefore, the primary data needed to implement such a model, are usually less than the Isard model. Wassily Leontief in collaboration with Alan Strout evaluated a gravity trade model for different commodity shipments. New interregional input-output models have been constructed, Kilkenny and Rose (1995) and Hewings (1995). 3. The Model The Greek economy is divided into three regions: North, Middle and Aegean Islands and Crete. The model contains ten intermediate sectors after aggregation. The solution of the model can be obtained by using two sets of structural constants. The first set describes the structure of production or input requirements in each region. [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] A second category of structural constants is a set of trade coefficients. The trade coefficients may also be presented as a diagonal block matrix. From the above two sets of structural constants a new matrix is obtained named regional input-output coefficient matrix and is given by the matrix A. [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] In the case of Greece with three regions I, II, III the terms [A. …

Clustering analysis, a key step for many data mining problems, can be applied to various fields. However, no matter what kind of clustering method, noise points have always been an important factor affecting the clustering effect. In addition, in spectral clustering, the construction of affinity matrix affects the formation of new samples, which in turn affects the final clustering results. Therefore, this study proposes a noise cutting and natural neighbors spectral clustering method based on coupling P system (NCNNSC-CP) to solve the above problems. The whole algorithm process is carried out in the coupled P system. We propose a natural neighbors searching method without parameters, which can quickly determine the natural neighbors and natural characteristic value of data points. Then, based on it, the critical density and reverse density are obtained, and noise identification and cutting are performed. The affinity matrix constructed using core natural neighbors greatly improve the similarity between data points. Experimental results on nine synthetic data sets and six UCI datasets demonstrate that the proposed algorithm is better than other comparison algorithms.

A flood inundation modelling using v-support vector machine regression model

Introduction: In order to assess how well conventional and hybrid pitch detection techniques perform in speech processing applications, a comparative analysis of the two types of methods is conducted. Method: A proposed hybrid approach, Proposed PEF+CEP, is examined alongside five traditional algorithms, namely Normalized Correlation Function (NCF), Pitch Estimation Filter (PEF), Log- Harmonic Summation (LHS), Summation of Residual Harmonics (SRH) and Cepstrum Pitch Determination (CEP). The effectiveness is evaluated using performance metrics like accuracy, specificity, sensitivity, and Gross Pitch Error (GPE). Results and Discussion: Our findings show that the accuracy and specificity of the traditional methods are impressive; the accuracy and sensitivity of the suggested hybrid method surpass their performance, with an astounding 98.8% accuracy and 99.2% sensitivity. Conclusion: Furthermore, the Proposed PEF+CEP method is a promising solution for accurate and dependable pitch detection in speech processing applications because it strikes a strong balance between computational efficiency and robustness. These results open up new avenues for research in the field of speech processing and demonstrate the potential of hybrid approaches.

Soft sensing technique has been extensively used to predict key quality variables in industrial systems. However, due to the difficulty of quality variable acquisition, only limited labeled data samples are available, and a large number of unlabeled ones are discarded. This raises a big challenge to build a high-quality soft sensor model. In order to furthest exploit information contained in both the labeled and unlabeled data, this paper proposes a principal component-based semi-supervised extreme learning machine (referred to as PCSELM) model. Through this model, extracting latent features and learning nonlinear input-output relationship can be simultaneously performed. In this way, unlabeled samples are utilized efficiently for feature representation and model accuracy improvement. Moreover, mixed regularizations are employed to work in conjunction with the PCSELM to obtain high generality and flexibility. We also derive an efficient parameter learning algorithm with theoretically guaranteed convergence. Comprehensive experiments are conducted via an industrial process. Comparison results illustrate that the proposed PCSELM outperforms other representative semi-supervised algorithms. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —Industrial processes in general incorporate unlabeled samples which are ubiquitous in real world applications. The focus of this paper is to develop a semi-supervised soft sensor model (PCSELM) that is capable to learn the nonlinear features and regression relationship efficiently with both the labeled and unlabeled samples. The proposed model can automatically implement the feature representation and the input-output relationship description. In addition, we introduce mixed norms for the model objective function to improve the final prediction performance and generalization. A feasible model optimization technique with proved convergence is also derived. Experimental results based on a real industrial dataset manifest that PCSELM achieves better prediction accuracy than its peers.

Many real world applications have problems with high dimensionality, which existing algorithms cannot overcome. A critical data preprocessing problem is feature selection, whereby its non-scalability negatively influences both the efficiency and performance of big data applications. In this research, we developed a new algorithm to reduce the dimensionality of a problem using graph-based analysis, which retains the physical meaning of the original high-dimensional feature space. Most existing feature-selection methods are based on a strong assumption that features are independent of each other. However, if the feature-selection algorithm does not take into consideration the interdependencies of the feature space, the selected data fail to correctly represent the original data. We developed a new feature-selection method to address this challenge. Our aim in this research was to examine the dependencies between features and select the optimal feature set with respect to the original data structure. Another important factor in our proposed method is that it can perform even in the absence of class labels. This is a more difficult problem that many feature-selection algorithms fail to address. In this case, they only use wrapper techniques that require a learning algorithm to select features. It is important to note that our experimental results indicates, this proposed simple ranking method performs better than other methods, independent of any particular learning algorithm used.

A semi‐analytical formulation is presented for transient metal‐forming processes which, being axisymmetric in geometry, are subjected to non‐axisymmetric loads and boundary conditions. The problems are described in terms of the flow formulation, and the pseudo‐concentration method (two material based) is used to account for time integration and non‐axisymmetric free surfaces. Proper Fourier series expansion of both the scalar fields defined to identify the free surfaces and of all the variables of the mechanical problem allows the performance of an advective transport–corresponding to time integration–for a non‐axisymmetric velocity field. A block diagonal matrix is obtained for each harmonic component, as it has been achieved for the constant time solution. The new configuration found as a result of this analysis is taken to calculate the velocity field for the incremented time, which in turn is used to perform the following time step.

In this paper, we propose an approximate Bayesian soft symbol detector (SSD) for multi-input multi-output (MIMO) orthogonal frequency division multiplexing underwater acoustic communications over fast time-varying channels. It is enabled by vector approximate message passing (VAMP) algorithm coupled with expectation-maximization (EM) technique, thus is named the EM-VAMP-SSD. The EM-VAMP-SSD consists of an inner soft slicer and an inner soft estimator (ISE), iteratively exchanging extrinsic information to improve symbol estimation performance. The operation of the ISE requires knowledge about channel and noise power, which are estimated via the orthogonal matching pursuit algorithm and the EM algorithm, respectively. To reduce the complexity of ISE, the frequency-domain MIMO channel is approximated by a block diagonal matrix without sacrificing performance much. The proposed EM-VAMP-SSD has been verified by both numerical and experimental data, showing consistent performance advantage over a conventional linear minimum mean square error SSD.

This paper reports simulations of power systems electromechanical transients on a multicomputer, formulated as a nonlinear algebraic problem by using the time parallelization concept. The bi-factorized inversion, which is the most time consuming stage of the simulation, is solved by the "Very Dishonest Newton (VDHN)-Maclaurin" method, a fully parallel indirect method based on the decomposition of the nonupdated Jacobian matrix. This proposal is made to orient the search for the decomposition based on a sufficient condition for the convergence of the Maclaurin series, which is a desirable situation for the design of more robust algorithms for power system simulation. Such condition keeps a close relation with a physical coupling property exhibited by power systems, and the characteristics of the simulation method. Theoretical and numerical results show that a successful implementation of this method can be better reached when the Jacobian matrix is decomposed as a block diagonal matrix plus a matrix with off diagonal blocks elements, the latter representing weak couplings between the diagonal blocks. The /spl epsiv/ Decomposition is used to satisfy the sufficient condition for convergence and the Longest Path Scheduling Method to prevent the uneven loading of processors, permitting to adapt the method in a efficient way on a coarse grain computer. The parallel simulation was written in C language and implemented on a Parsytec PowerXplorer multicomputer. Tests using electromechanical models of the Chilean Central interconnected system and the IEEE300 test system were made to evaluate the advantages and drawbacks of the parallel method.