Sharif Digital Repository

Kernel learning is one of the most important and recent approaches to constrained clustering. Until now many kernel learning methods have been introduced for clustering when side information in the form of pairwise constraints is available. However, almost all of the existing methods either learn a whole kernel matrix or learn a limited number of parameters. Although the non-parametric methods that learn whole kernel matrix can provide capability of finding clusters of arbitrary structures, they are very computationally expensive and these methods are feasible only on small data sets. In this paper, we propose a kernel learning method that shows flexibility in the number of variables between... 

The so-called enriched weight functions (EWFs) are utilized in mesh-free methods (MMs) to solve linear elastic fracture mechanics (LEFM) problems; the following issues are of concern: convergence behavior; sufficiency of EWFs to capture singular fields around the crack-tip; and the preservation of the J-integral path-independency. EWFs prove useful in conjunction with the moving least square reproducing kernel method (MLSRKM); for this purpose, both EWFs and MLSRKM are modified. Since EWFs are not truly representative of the near-tip solution, fully EWFs (FEWFs) are introduced. Finally, some descriptive examples address the aforementioned concerns and the accuracy and efficacy of the... 

In this study, we explored the growth kinetics of the Al-Fe intermetallic (IM) layer at the joint interface of the St-12/Al-5083 friction stir lap welds during post-weld annealing treatment at 350, 400 and 450 C for 30 to 180 min. Optical microscope (OM), field emission gun scanning electron microscope (FEG-SEM) and transmission electron microscope (TEM) were employed to investigate the structure of the weld zone. The thickness and composition of the IM layers were evaluated using image analysis system and electron back-scatter diffraction (EBSD), respectively. Moreover, kernel average misorientation (KAM) analysis was performed to evaluate the level of stored energy in the as-welded state.... 

This paper presents vibration and buckling analysis of functionally graded beams with different boundary conditions, using reproducing kernel particle method (RKPM). Vibration of simple Euler-Bernoulli beam using RKPM is already developed and reported in the literature. Modeling of FGM beams using theoretical method or finite element technique is not evolved with accurate results for power law form of FGM with large power of "n" value so far. Accuracy of the RKPM results is very good and is not sensitive to n value. System of equations of motion is derived using Lagrange's method under the assumption of Euler-Bernoulli beam theory. Boundary conditions of the beam are taken into account using... 

Segmentation of a scene to the part made is a challenging work. In this paper a graphical model is used for this task. The methods based on geometrical derivatives such as curvature and normal often haven't good result in segmentation of geometrically-complex architecture and lead to over-segmentation and even failure. Proposed method for segmentation contains two steps. At first region growing based on curvature, normal and color is used for growing region. This segmented cloud is used for unary potential in graphical model. Fully connected graph for Conditional Random Field with Gaussian kernel for pair wise potentials is used for correcting this segmentation. Gaussian kernels are based on... 

This paper introduces the first pattern-based Persian Temporal Relation Classifier (PTRC) that finds the type of temporal relations between pairs of events in the Persian texts. The proposed system uses support vector machines (SVMs) equipped by combinations of simple, convolution tree, and string subsequence kernels (SSK). In order to evaluate the algorithm, we have developed a Persian TimeBank (PTB) corpus. PTRC not only increases the performance of the classification by applying new features and SSK, but also alleviates the probable adverse effects of the Free Word Orderness (FWO) of Persian on temporal relation classification. We have also applied our proposed algorithm to two standard... 

Distance metric learning is a powerful approach to deal with the clustering problem with side information. For semi-supervised clustering, usually a set of pairwise similarity and dissimilarity constraints is provided as supervisory information. Although some of the existing methods can use both equivalence (similarity) and inequivalence (dissimilarity) constraints, they are usually limited to learning a global Mahalanobis metric (i.e., finding a linear transformation). Moreover, they find metrics only according to the data points appearing in constraints, and cannot utilize information of other data points. In this paper, we propose a probabilistic metric learning algorithm which uses... 

In this paper, we introduce a model-based Bayesian denoising framework for phonocardiogram (PCG) signals. The denoising framework is founded on a new dynamical model for PCG, which is capable of generating realistic synthetic PCG signals. The introduced dynamical model is based on PCG morphology and is inspired by electrocardiogram (ECG) dynamical model proposed by McSharry et al. and can represent various morphologies of normal PCG signals. The extended Kalman smoother (EKS) is the Bayesian filter that is used in this study. In order to facilitate the adaptation of the denoising framework to each input PCG signal, the parameters are selected automatically from the input signal itself. This... 

In this paper, we address the problem of data description using a Bayesian framework. The goal of data description is to draw a boundary around objects of a certain class of interest to discriminate that class from the rest of the feature space. Data description is also known as one-class learning and has a wide range of applications. The proposed approach uses a Bayesian framework to precisely compute the class boundary and therefore can utilize domain information in form of prior knowledge in the framework. It can also operate in the kernel space and therefore recognize arbitrary boundary shapes. Moreover, the proposed method can utilize unlabeled data in order to improve accuracy of... 

In this letter, a new adaptive Gaussian process (GP) frame work for signal extrapolation is proposed. Signal extrapolation is an essential task in many applications such as concealment of corrupted data in image and video communications. While possessing many interesting properties, Gaussian process priors with inappropriate stationary kernels may create extremely blurred edges in concealed areas of the image. To address this problem, we propose adaptive non-stationary kernels in a Gaussian process framework. The proposed adaptive kernel functions are defined based on the hypothesized edges of the missing areas. Experimental results verify the effectiveness of the proposed method compared to... 

Sparse representation is a powerful tool for image processing, including noise removal. It is an effective method for Gaussian noise removal by taking advantage of a fixed and learned dictionary. In this study, the variable distribution of Rician noise is reduced in magnetic resonance (MR) images by sparse representation based on reconstruction error sets. Standard deviation of Gaussian noise is used to find these errors locally. The proposed method represents two formulas for local error calculation using standard deviation of noise. The acquired results from the real and simulated images are comparable, and in some cases, better than the best Rician noise removal method due to the... 

The effects of temperature, seed mass and agitation rate on agglomeration kinetics of aluminum hydroxide in Bayer process have been studied in a batch system. Collected raw data were analyzed and the kinetics data of agglomeration were derived through simulation of the process using a pre-developed software. The results showed that agglomeration kinetics constant (agglomeration kernel) increases with increase in temperature and agitation rate. Moreover a maximum value of agglomeration rate versus added seed mass was observed. Furthermore the magnitude of calculated activation energy of agglomeration was close to that of growth  

A classifier based on Support Vector Data Description (SVDD) is proposed for spoken digit recognition. We use the Mel Frequency Discrete Wavelet Coefficients (MFDWC) and the Mel Frequency cepstral Coefficients (MFCC) as the feature vectors. The proposed classifier is compared to the HMM and results are promising and we show the HMM and SVDD classifiers have equal accuracy rates. The performance of the proposed features and SVDD classifier with several kernel functions are evaluated and compared in clean and noisy speech. Because of multi resolution and localization of the Wavelet Transform (WT) and using SVDD, experiments on the spoken digit recognition systems based on MFDWC features and... 

In this paper, we investigate the problem of designing compact-support interpolation kernels for a given class of signals. By using calculus of variations, we simplify the optimization problem from an nonlinear infinite dimensional problem to a linear finite dimensional case, and then find the optimum compact-support function that best approximates a given filter in the least square sense (ℓ 2 norm). The benefit of compact-support interpolants is the low computational complexity in the interpolation process while the optimum compact-support interpolant guarantees the highest achievable signal-to-noise ratio (SNR). Our simulation results confirm the superior performance of the proposed kernel... 

Investigation of response of reinforced concrete (RC) structures due to axisymmetric macrocell corrosion of rebars is of concern after propagation of microcracks within the concrete medium. The geometry, boundary and interfaces conditions of the present problem are identical to those stated in part I. As seen in the companion paper, the exact solution to the boundary value problem corresponding to the uncracked steel-rust-concrete composite was possible. After appearance of the microcracks the concrete behavior becomes nonlinear anisotropic with post-cracking softening, and the associated problem is analytically intractable. Therefore, it is proposed to employ a novel meshless method, namely... 

The main purpose of this article is to generalize a characterization of constant functions to the context of metric-measure spaces. In fact, we approximate a measurable function, in terms of a certain integrability condition, by Lipschitz functions. Then, similar to Brezis (2002) [2], we establish a necessary and sufficient condition in order that any measurable function which satisfies an integrability condition to be constant a.e. Also, we provide a different proof for the main result of Pietruska-Pałuba (2004) [7] in the setting of Dirichlet forms  

Kernel based methods are significantly important in the pattern classification problem, especially when different classes are not linearly separable. In this paper, we propose a new kernel, which is the modified version of the polynomial kernel. The free parameter (d) of the proposed kernel considerably affects the error rate of the classifier. Thus, we present a new algorithm based on the Fisher criterion to find the optimum value of d. Simulation results show that using the proposed kernel for classification leads to satisfactory results. In our simulation in most cases the proposed method outperforms the classification using the polynomial kernel  

Active learning has been a popular area of research in recent years. It can be used to improve the performance of learning tasks by asking the labels of unlabeled data from the user. In these methods, the goal is to achieve the highest possible accuracy gain while posing minimum queries to the user. The existing approaches for active learning have been mostly applicable to the traditional binary or multi-class classification problems. However, in many real-world situations, we encounter problems in which we have access only to samples of one class. These problems are known as one-class learning or outlier detection problems and the User relevance feedback in image retrieval systems is an... 

This paper presents free vibration analysis of functionally graded material (FGM) beams with different boundary conditions, using RKPM (Reproducing Kernel Particle Method), which is a meshless method. System of equations of motion is derived by using Lagrange's method under the assumption of Euler-Bernoulli beam theory. Boundary conditions of beam are taken into account by using Lagrange multipliers. It is assumed that material properties of the beam vary continuously in the thickness direction according to the power-law form. RKPM is applied to obtain eigenvalue equation of vibration and natural frequencies are obtained. It should be noted that for special cases where the beam is uniform,... 

A meshless method called reproducing kernel particle method (RKPM) is exploited to cope with elastic-plastic fracture mechanics (EPFM) problems. The idea of arithmetic progression is assumed to place particles within the refinement zone in the vicinity of the crack tip. A comparison between two conventional treatments, visibility and diffraction, to crack discontinuity is conducted. Also, a tracking to find the appropriate diffraction parameter is performed. To assess the suggestions made, two mode I numerical simulations, pure tension and pure bending tests, are executed. Results including J integral, crack mouth opening displacement (CMOD), and plastic zone size and shape are compared with...