This paper presents a method for learning an overcomplete, nonnegative dictionary and for obtaining the corresponding coefficients so that a group of nonnegative signals can be sparsely represented by them. This is accomplished by posing the learning as a problem of nonnegative matrix factorization (NMF) with maximization of the incoherence of the dictionary and of the sparsity of coefficients. By incorporating a dictionary-incoherence penalty and a sparsity penalty in the NMF formulation and then adopting a hierarchically alternating optimization strategy, we show that the problem can be cast as two sequential optimal problems of quadratic functions. Each optimal problem can be solved explicitly so that the whole problem can be efficiently solved, which leads to the proposed algorithm, i.e., sparse hierarchical alternating least squares (SHALS). The SHALS algorithm is structured by iteratively solving the two optimal problems, corresponding to the learning process of the dictionary and to the estimating process of the coefficients for reconstructing the signals. Numerical experiments demonstrate that the new algorithm performs better than the nonnegative K-SVD (NN-KSVD) algorithm and several other famous algorithms, and its computational cost is remarkably lower than the compared algorithms.

With the successful adoption of link analysis techniques such as PageRank and web spam filtering, current web search engines well support “navigational search”. However, due to the use of a simple conjunctive Boolean filter in addition to the inappropriateness of user queries, such an engine does not necessarily well support “informational search”. Informational search would be better handled by a web search engine using an informational retrieval model combined with enhancement techniques such as query expansion and relevance feedback. Moreover, the realization of such an engine requires a method to prosess the model efficiently. In this paper we propose a novel extension of an existing top-k query processing technique to improve search efficiency. We add to it the technique utilizing a simple data structure called a “term-document binary matrix,” resulting in more efficient evaluation of top-k queries even when the queries have been expanded. We show on the basis of experimental evaluation using the TREC GOV2 data set and expanded versions of the evaluation queries attached to this data set that the proposed method can speed up evaluation considerably compared with existing techniques especially when the number of query terms gets larger.

This paper describes a computer-aided diagnosis (CAD) method to classify pneumoconiosis on HRCT images. In Japan, the pneumoconiosis is divided into 4 types according to the density of nodules: Type 1 (no nodules), Type 2 (few small nodules), Type 3-a (numerous small nodules) and Type 3-b (numerous small nodules and presence of large nodules). Because most pneumoconiotic nodules are small-sized and irregular-shape, only few nodules can be detected by conventional nodule extraction methods, which would affect the classification of pneumoconiosis. To improve the performance of nodule extraction, we proposed a filter based on analysis the eigenvalues of Hessian matrix. The classification of pneumoconiosis is performed in the following steps: Firstly the large-sized nodules were extracted and cases of type 3-b were recognized. Secondly, for the rest cases, the small nodules were detected and false positives were eliminated. Thirdly we adopted a bag-of-features-based method to generate input vectors for a support vector machine (SVM) classifier. Finally cases of type 1,2 and 3-a were classified. The proposed method was evaluated on 175 HRCT scans of 112 subjects. The average accuracy of classification is 90.6%. Experimental result shows that our method would be helpful to classify pneumoconiosis on HRCT.

We show a phase transition of the first eigenvalue of random (c,d)-regular graphs, whose instance of them consists of one vertex with degree c and the other vertices with degree d for c > d. We investigate a reduction from the first eigenvalue analysis of a general (c,d)-regular graph to that of a tree, and prove that, for any fixed c and d, and for a graph G chosen from the set of all (c,d)-regular graphs with n vertices uniformly at random, the first eigenvalue of G is approximately with high probability.

We propose a novel network traffic matrix decomposition method named Stable Principal Component Pursuit with Frequency-Domain Regularization (SPCP-FDR), which improves the Stable Principal Component Pursuit (SPCP) method by using a frequency-domain noise regularization function. An experiment demonstrates the feasibility of this new decomposition method.

In our previous study, we proposed the waveform interpolation (WI) approach to model the excitation signals for hidden Markov model (HMM)-based speech synthesis. This letter presents several techniques to improve excitation modeling within the WI framework. We propose both the time domain and frequency domain zero padding techniques to reduce the spectral distortion inherent in the synthesized excitation signal. Furthermore, we apply non-negative matrix factorization (NMF) to obtain a low-dimensional representation of the excitation signals. From a number of experiments, including a subjective listening test, the proposed method has been found to enhance the performance of the conventional excitation modeling techniques.

A semi-analytical method for a planar periodic array formed by a finite number of magnetized ferrite circular cylinders is presented using a model of layered cylindrical structures. The method uses the T-matrix approach and the extraction of the reflection and transmission matrices based on the cylindrical harmonic mode expansion. Based on the proposed method, plane wave scattering by the finite number of magnetized ferrite circular cylinders is numerically studied from the viewpoint of realization the electronic switching and electronic scanning effects by varying the applied magnetic field.

Variable-weight optical orthogonal codes (OOCs) have application in multimedia optical code division multiple access (OCDMA) systems supporting multiple quality of services (QoS). In this paper, several combinatorial constructions for optimal variable-weight OOCs are presented explicitly. A useful recursive construction for optimal variable-weight OOCs is proposed as well. Based on these results, two new infinite classes of optimal variable-weight OOCs with Hamming weights 3 and 4 are obtained.

Microwave imaging techniques, in particular synthetic aperture radar (SAR), are able to obtain useful images even in adverse weather or darkness, which makes them suitable for target position or feature estimation. However, typical SAR imagery is not informative for the operator, because it is synthesized using complex radio signals with greater than 1.0 m wavelength. To deal with the target identification issue for imaging radar, various automatic target recognition (ATR) techniques have been developed. One of the most promising ATR approaches is based on neural network classification. However, in the case of SAR images heavily contaminated by random or speckle noises, the classification accuracy is severely degraded because it only compares the outputs of neurons in the final layer. To overcome this problem, this paper proposes a self organized map (SOM) based ATR method, where the binary SAR image is classified using the unified distance matrix (U-matrix) metric given by the SOM. Our numerical analyses and experiments on 5 types of civilian airplanes, demonstrate that the proposed method remarkably enhances the classification accuracy, particular in lower S/N situations, and holds a significant robustness to the angular variations of the observation.

We apply the Hadamard equivalence to all the binary matrices of the size mn and study various properties of this equivalence relation and its classes. We propose to use HR-minimal as a representative of each equivalence class, and count and/or estimate the number of HR-minimals of size mn. Some properties and constructions of HR-minimals are investigated. Especially, we figure that the weight on an HR-minimal's second row plays an important role, and introduce the concept of Quasi-Hadamard matrices (QH matrices). We show that the row vectors of mn QH matrices form a set of m binary vectors of length n whose maximum pairwise absolute correlation is minimized over all such sets. Some properties, existence, and constructions of Quasi-orthogonal sequences are also discussed. We also give a relation of these with cyclic difference sets. We report lots of exhaustive search results and open problems, one of which is equivalent to the Hadamard conjecture.

Fast and accurate estimation of missing relations, e.g., similarity, distance and kernel, among objects is now one of the most important techniques required by major data mining tasks, because the missing information of the relations is needed in many applications such as economics, psychology, and social network communities. Though some approaches have been proposed in the last several years, the practical balance between their required computation amount and obtained accuracy are insufficient for some class of the relation estimation. The objective of this paper is to formalize a problem to quickly and efficiently estimate missing relations among objects from the other known relations among the objects and to propose techniques called “PSD Estimation” and “Row Reduction” for the estimation problem. This technique uses a characteristic of the relations named “Positive Semi-Definiteness (PSD)” and a special assumption for known relations in a matrix. The superior performance of our approach in both efficiency and accuracy is demonstrated through an evaluation based on artificial and real-world data sets.

Recently a simple algorithm was presented by the first author which enables one to successively compute the transformation matrix of various order for the general 1-D to 1-D polynomial transformation. This letter extends the result to the general 1-D to 2-D polynomial transformation. It is also shown that the matrix obtained can be used for the 2-D to 2-D polynomial transformation as well.

This paper proposes an easy-to-design, theory-consistent compact feeding circuit, with a single input and four outputs, being comprised of two hybrid circuits that are capable of switching a beam in three directions. The circuits that determine the phase differences between the antennas are present on the same single layer, and thus there is no effect of vias and the design agrees well with the underlying theory. In addition, the vertically and horizontally symmetrical circuit pattern contributes to a substantial reduction in design time. The circuit is designed for use in the ISM band and its properties are evaluated using an RF circuit simulator. A prototype is fabricated and evaluated. The results of the simulation and measurement agree well with the theoretical values. The dimensions of the feeding circuit are 75 (H)55 (W)3.0 (T) mm.

Random beamforming(RBF) is a simple and practical method that can realize multi-user multi-input multi-output (MU-MIMO) systems. In this letter, we analyze the average sum rate of RBF with minimum mean squared error (MMSE) receive beamforming. To this end, we exploit the empirical eigenvalue distribution [5] and extreme value theory. The numerical verification shows that the proposed analysis provides a good approximation of the average sum rate of RBF even for the small number of antennas.

Non-negative matrix factorization (NMF) is widely used for music transcription because of its efficiency. However, the conventional NMF-based music transcription algorithm often causes harmonic confusion errors or time split-up errors, because the NMF decomposes the time-frequency data according to the activated frequency in its time. To solve these problems, we proposed an NMF with temporal continuity and harmonicity constraints. The temporal continuity constraint prevented the time split-up of the continuous time components, and the harmonicity constraint helped to bind the fundamental with harmonic frequencies by reducing the additional octave errors. The transcription performance of the proposed algorithm was compared with that of the conventional algorithms, which showed that the proposed method helped to reduce additional false errors and increased the overall transcription performance.

This letter proposes a new image colorization algorithm based on the sparse optimization. Introducing some assumptions, a problem of recovering a color image from a grayscale image with the small number of known color pixels is formulated as a mixed l0/l1 norm minimization, and an iterative reweighted least squares (IRLS) algorithm is proposed. Numerical examples show that the proposed algorithm colorizes the grayscale image efficiently.

This paper presents a novel algorithm to generate homogeneous superpixels from Markov random walks. We exploit Markov clustering (MCL) as the methodology, a generic graph clustering method based on stochastic flow circulation. In particular, we introduce a graph pruning strategy called compact pruning in order to capture intrinsic local image structure. The resulting superpixels are homogeneous, i.e. uniform in size and compact in shape. The original MCL algorithm does not scale well to a graph of an image due to the square computation of the Markov matrix which is necessary for circulating the flow. The proposed pruning scheme has the advantages of faster computation, smaller memory footprint, and straightforward parallel implementation. Through comparisons with other recent techniques, we show that the proposed algorithm achieves state-of-the-art performance.

It is well known that Boolean functions used in stream and block ciphers should have high algebraic immunity to resist algebraic attacks. Up to now, there have been many constructions of Boolean functions achieving the maximum algebraic immunity. In this paper, we present several constructions of rotation symmetric Boolean functions with maximum algebraic immunity on an odd number of variables which are not symmetric, via a study of invertible cyclic matrices over the binary field. In particular, we generalize the existing results and introduce a new method to construct all the rotation symmetric Boolean functions that differ from the majority function on two orbits. Moreover, we prove that their nonlinearities are upper bounded by .

In this paper, we derive a simple formula to generate a wide-sense systematic generator matrix(we call it quasi-systematic) B for a Reed-Solomon code. This formula can be utilized to construct an efficient interpolation based erasure-only decoder with time complexity O(n2) and space complexity O(n). Specifically, the decoding algorithm requires 3kr + r2 - 2r field additions, kr + r2 + r field negations, 2kr + r2 - r + k field multiplications and kr + r field inversions. Compared to another interpolation based erasure-only decoding algorithm derived by D.J.J. Versfeld et al., our algorithm is much more efficient for high-rate Reed-Solomon codes.

Visually saliency detection provides an alternative methodology to image description in many applications such as adaptive content delivery and image retrieval. One of the main aims of visual attention in computer vision is to detect and segment the salient regions in an image. In this paper, we employ matrix decomposition to detect salient object in nature images. To efficiently eliminate high contrast noise regions in the background, we integrate global context information into saliency detection. Therefore, the most salient region can be easily selected as the one which is globally most isolated. The proposed approach intrinsically provides an alternative methodology to model attention with low implementation complexity. Experiments show that our approach achieves much better performance than that from the existing state-of-art methods.