Search

Distributed Denial of Service (DDoS) is one of the key cyber-attack nowadays. DDoS attacks disrupt the target systems to unavailable to the legitimate users. Lightweight Directory Access Protocol (LDAP) is a reflection DDoS attack, it makes target server inaccessible to the legitimate users by sending large number of LDAP requests to the target server. Inaccessibility of digital service creates many negative consequences nowadays because everything is digitalized these days. Attack detection is very essential to reduce losses in all aspects. This study proposed detection of LDAP DDoS using Support Vector Machine (SVM) classifier linear, sigmoid, RBF and poly kernels using network flow features. The LDAP_DrDoS dataset was used in this study to conduct experiments on it, which was collected from the CIC-DDoS2019 evolution datasets. This study evaluates the proposed model with four different uncorrelated features subsets. Among these four subsets, three are collected by Pearson, Kendall and Spearman correlation methods and fourth one which is called PSK formed with common features of three uncorrelated features subsets. Classification results are evaluated with accuracy, log-loss value, ROC-AUC score and K-fold cross validation evaluation metrics. Among kernel functions, poly and RBF kernel functions give better results for LDAP DDoS attack detection. Among feature subsets, Pearson produce the better classification results. Overall, The SVM poly kernel is most effective in detecting LDAP DDoS attacks with Pearson features subsets. [ABSTRACT FROM AUTHOR]

The accuracy of the work done by several semi-parametric regression models, which are used to address issues in various fields such as economics, medicine, etc., is determined by a set of criteria and conditions such as the speed of access to results and their accuracy. Several evolutionary algorithms have been used which have good results, in this manuscript, the bat algorithm (BA) is used in a separate space to determine the smoothing coefficient used in these models. Where the approach finds the optimal values of the smoothing coefficient by first filling the initial population with random values and then using the kernel function to analyze the results of this approach, which showed more accurate results and took less time than the rest of the compared methods such as CV and GCV. [ABSTRACT FROM AUTHOR]

The inverse problem of finding a one-dimensional memory kernel of a time convolution integral depending on a time variable t and one-dimensional spatial variable x in the two-dimensional heat equation with a time-dependent coefficient of thermal conductivity is studied. The direct problem is the Cauchy problem for this equation. The integral term has the time convolution form of kernel and an elliptic operator of direct problem solution. As additional information, the solution of the direct problem on the x=0 is given. The problem reduces to an auxiliary problem which is more convenient for further consideration. Then the auxiliary problem is replaced by an equivalent system of Volterra-type integral equations with respect to unknown functions. Applying the method of contraction mappings to this system in the Hölder class of functions, it is proved the main result of the paper representing a local existence and uniqueness theorem. [ABSTRACT FROM AUTHOR]

Maximum-entropy moment closures for describing non-equilibrium rare fied gaseous flow shave previously been shown to provide accurate and computationally efficient descriptions of transition-regime flows. Unfortunately, for high-order variants of these closures above second order in velocity space, there are no analytical closures for the systems of hyperbolic partial differential equations (PDEs) which govern the transport ofthe macroscopic moment quantities and instead approximate closures have been sought. In this study, abi-Gaussian approximation for then umber density function (NDF) is considered both for approximating the NDF and closing moment fluxes of the resulting fourth-order 14-moment maximum-entropy closure associated with fully three-dimensional kinetic theory. Prior investigations of the bi-Gaussian approximation applied to simplified one-dimensional univariate kinetic theory has yielded excellent results when compared to the actual maximum-entropy solutions as well a similar interpolative-based maximum-entropy-based (IBME) closure. In the one-dimensional univariate case, the bi-Gaussian closure is equivalent to the so-called extended quadrature method of moments (EQMOM) with anormal or Gaussian kernel basis function. A potential benefit of the bi-Gaussian approach proposed herein is that an essentially closed-form analytical expression results for the NDF. In this study, the extension of the bi-Gaussian closure to the multi-dimensional case is considered and compared to the equivalent multi-dimensional IBME closure. The approximate form for the NDF and closing fluxes interms of the relevant moments are derived and the validity and hyperbolicity of the closure for the space of realizable predicted moments are all explored and compared to those of the IBME closure. It is shown that the bi-Gaussian closure in the multi-dimensional case unfortunately suffers from several de ficiencies: firstly, the valid region of realizable moment space for the bi-Gaussian closure is a small subset of the full realizable 14-moment space; and secondly, the closure and moment equation eigen structure for solutions associated with zero heat fluxbe come undefined. The findings here in suggest that the proposed bi-Gaussian closure may not be agood choice for practical multi-dimensional rare fied flow predictions despite the promising results exhibited in the one-dimensional case. [ABSTRACT FROM AUTHOR]

Aim: To predict the accuracy percentage of misinformation about COVID-19 using SVM algorithm. Materials and methods: Support Vector Machine (SVM) with sample size=20 and Logistic Regression with sample size=20 was iterated at different times for predicting the accuracy percentage of misinformation about COVID19. The Novel Poly kernel function used in SVM maps the dataset into higher dimensional space which helps to improve accuracy percentage. Results and Discussion: SVM has significantly better accuracy (94.48%) compared to Logistic Regression accuracy (91.07%). SVM performs significantly better than the Logistic Regression with (p=0.024) (p<0.05). Conclusion: SVM with Novel Poly kernel helps in predicting with more accuracy the percentage of misinformation about COVID-19 [ABSTRACT FROM AUTHOR]

According to the expert, Twitter is a microblogging site that provides facilities for users to send a text message with a maximum length of 280 characters, and the text that we type on Twitter called 'tweet' we can write down what happened in various places followed by the author's emotions. One of the information obtained from Twitter is related to politics. Through the sentiment analysis, it can find emotions and correlation between the results of the statement analysis electability statistic of the candidate Paris for governor and deputy governor using sentiment analysis. The purpose of this paper is to get the best accuracy result by comparing the three kernels (Gaussian RBF Kernel, Linear Kernel, and Polynomial Kernel) using SVM and looking for the relationship between the sentiment analysis result and electability from multiple survey institutions using Pearson Product Moment Correlation Technique by using data from Twitter. Data from Twitter has been categorized into five emotions (love, happiness, sadness, fear, and anger). When three kernels are compared for accuracy, the Polynomial Kernel comes out on top with a score of 78.51%. The positive correlation for Basuki Tjahaja Purnama - Djarot Saiful Hidayat is -0.81, whereas the positive correlation for Anies Rasyid Baswedan - Sandiaga Salahuddin Uno is 0.98. [ABSTRACT FROM AUTHOR]

An efficient priority queue data structure has been designed to improve the performance of applications that utilize a min/max heap, by reducing the execution time. An adaptive multivariate integration application was identified as an excellent candidate for experimentation. Originally, a max heap was at the heart of the algorithm, keyed with the absolute error estimates of the subregions generated by adaptive partitioning. The algorithm consists of selecting/deleting a region from the heap, dividing the region into several subregions, evaluating the subregions, and inserting the subregions back into the heap. The steps continue until the error is reduced below a user-prescribed threshold, or a prescribed number of function samples has been reached. The max heap was replaced by the priority queue data structure that we designed. The data structure is bucket-based and uses a hashing function to determine the appropriate bucket in which to insert a data item. The hashing function allows the data structure to determine the appropriate bucket with a constant (풪(1)) time complexity. This outperforms a max heap, which determines the appropriate place to insert with an 풪(log n) time complexity. The adaptive multivariate integration application calls a Compute Unified Device Architecture (CUDA) kernel function to evaluate the subregions. We reengineered the parallel algorithm to call the kernel function from another kernel function, allowing multiple regions to be processed on the GPU simultaneously. This further improves the performance of the application by enhancing the speedup 2-3 fold. [ABSTRACT FROM AUTHOR]

In this study, the Support Vector Regression (SVR) model was used to forecast air quality time series data which is the particulate matter 10 micrometers or less in diameter (PM10) in Malaysia. Several SVR kernel functions, which are the Linear, Polynomial and Radial Basis Function (RBF) kernels, were considered in this study to determine the most suitable kernel function for forecasting the PM10 time series. The period of the data is from 5th July 2017 to 31st January 2019 consists of five air quality monitoring stations which are Kangar station in Perlis, Tasek Ipoh station in Perak, Shah Alam station in Selangor, Pasir Gudang station in Johor and Kuala Terengganu station in Terengganu. Model performance was compared based on the testing dataset's mean squared error (MSE) values. The results show that the SVR model with Radial Basis Function kernel is more suitable for forecasting the PM10 time series compared to the Linear and Polynomial kernels. [ABSTRACT FROM AUTHOR]

In this study, the Support Vector Regression (SVR) model was used to forecast air quality time series data which is the particulate matter 10 micrometers or less in diameter (PM10) in Malaysia. Several SVR kernel functions, which are the Linear, Polynomial and Radial Basis Function (RBF) kernels, were considered in this study to determine the most suitable kernel function for forecasting the PM10 time series. The period of the data is from 5th July 2017 to 31st January 2019 consists of five air quality monitoring stations which are Kangar station in Perlis, Tasek Ipoh station in Perak, Shah Alam station in Selangor, Pasir Gudang station in Johor and Kuala Terengganu station in Terengganu. Model performance was compared based on the testing dataset's mean squared error (MSE) values. The results show that the SVR model with Radial Basis Function kernel is more suitable for forecasting the PM10 time series compared to the Linear and Polynomial kernels. [ABSTRACT FROM AUTHOR]

In the past two decades, chemical looping reforming (CLR) is gaining increasing research interest because of the advantages of not using pure oxygen in the reaction, which allows the saving of energy-intensive air separation requirements in conventional reforming processes. The modeling of hydrogen production by CLR based on input process parameters and employing steam as an oxidant has been investigated. CLR produces considerable amounts of hydrogen, and it has been discovered that input factors such as reaction temperature, catalyst loading, and steam-to-methane ratio all have a major impact on this amount of hydrogen produced. The hydrogen produced was positively influenced by both the reaction temperature and the steam-to-methane ratio. Increases in both the reaction temperature and the steam-to-methane ratio were shown to cause a significant rise in the amount of hydrogen produced. On the contrary, increasing the amount of catalyst used resulted in peak performance in terms of hydrogen production at 0.4%. There is a considerable impact of the three kernel functions (linear, quadratic, and cubic) on the performance of the support vector machine (SVM) when modeling the nonlinear connection between the process parameters and the desired output. The robustness of the SVM is further demonstrated by the R2 values of 0.80, 0.89, and 090 obtained for the prediction of hydrogen production. The root mean square error (RMSE) and mean absolute error (MAE) values suggest that the SVM produces minimum predictive errors. [ABSTRACT FROM AUTHOR]

The first case of COVID-19 has been verified in Indonesia detected on March 2, 2020. Since then, the number of cases of COVID-19 in Indonesia has continued to increase every day. West Java is one of the provinces with confirmed positive cases due to local transmission, which is high in Indonesia, reaching 513,000 cases as of September 3, 2021. The wide area coverage and the rapid spread of the COVID-19 virus from one region to another indicate the presence of spatial effect. The goal of this research is to figure out what factors impact the frequency of positive instances of COVID-19 in districts/cities in West Java province. Because the data is in the form of count data, the analysis that can be used is the Poisson regression model. However, we found cases of overdispersion and there is a spatial effect, so the right model to overcome this problem is the Geographically Weighted Negative Binomial Regression (GWNBR) model for each district and city in West Java Province. The weighting matrix used is a fixed gaussian kernel function. The results of the mapping inform that the factors that influence the number of positive cases of Covid-19 globally in all districts and cities in West Java Province include population density, the percentage of households that have access to proper sanitation, and the number of villages that have health facilities. In addition, six groups of district and city classifications were obtained with each variable that influenced it. [ABSTRACT FROM AUTHOR]

The internet as a medium of communication today is a very important part of life for humans, including students. The use of the internet by students is not only related to academic activities but also to meet their daily needs. Various factors influence student decisions in online shopping. These factors are broadly classified in the form of customer profiles as intrinsic factors and online shop profiles as extrinsic. Online business actors need a process of identifying potential buyers of their products to further market these products. The forming factors of online shopping behavior are classified using the Support Vector Machine (SVM) which has a strong generalization ability on high-dimensional data. SVM can solve the linear and nonlinear regression problems of the kernel in classification activities. On data that cannot be separated linearly, SVM uses Kernel Trick. on various kernel functions including a nonlinear kernel, polynomial, sigmoid, and Radial Base Function (RBF). The weakness in SVM is that it has problems in selecting the appropriate online shopping parameters to improve optimization. To overcome its weaknesses, SVM can be strengthened by a learning algorithm to perform classification and regression. The genetic algorithm is used to find the optimum online shopping parameter values and is suitable for the SVM method. The results of the comparison of the accuracy values obtained show that the search for SVM parameter values using genetic algorithms results in achieving more optimal accuracy. [ABSTRACT FROM AUTHOR]

A contact problem for an elastic spherical shell of medium thickness and a rigid plane is considered. The shell model is formulated as a 2D constrained continuum on the background of a general higher-order shell theory. To find the contact pressure distribution, a static axisymmetric contact problem for a spherical shell and an absolutely solid surface is solved. The main governing equation is an integral equation using the transient function for a shell as a kernel. To close the system, the main governing equation is supplemented by the equilibrium condition and an additional relation determining the position of the contact area boundary. This system is solved iteratively using the quadrature method. The transient function is found by solving the corresponding auxiliary problem using series with respect to Legendre and Gegenbauer polynomials. By orthogonal expansion the formulated problem is reduced to the system of algebraic equations for the coefficients of the orthogonal series. The convergence of the constructed solution for the transient function is studied. The test problem on the given pressure impact on the spherical shell has been solved. Finally, the implementation of the theory of multicomponent dry friction in some engineering problems of the dynamics of composite shells interacting with rigid rough planes is proposed. The main attention is paid to the construction of analytical models of combined dry friction, taking into account the real distribution of normal and tangential contact stresses. [ABSTRACT FROM AUTHOR]

Our goal is to include a brief overview of the WSN system and its potential to smooth the path to link the real and virtual environment to a global network centered on the Internet. Thus, in these modern technologies, sensor nodes play a significant role in transmitting information from a connection node to another defined node within its full scope. According to the ideal condition of WSN, the issue of radiofrequency network localization is of considerable significance in many technological applications such as military hardware for the identification of local points of attack or protection, environmental science besides, in general, sensor nodes. The most advanced equipment for acquiring a direct Global Positioning System location. We look at the primary structure of WSNs for kernel functions and methods to identify the most suitable and great access value based on the localization action after a brief history of the introduction part. Then, to put in place, the suitable approach for accessing terminals in the best obtainable state in WSNs, we restate the algorithms and techniques to build. As a result, we conclude on the points raised to compare and make a conclusion. Therefore, the final aim of this analysis is to clarify the position's quality and effectiveness based on diverse viewpoints. The results show that optimum placement could modify this immense technology to find new ways to find nodes based on the unexpected structure of WSNs in the best nodes. [ABSTRACT FROM AUTHOR]

This paper presents a mathematical model based on the influence of unmonitored asymptomatic individuals, vaccinations and individuals returning home to the spread of COVID 19. The concept used is that individual populations moves in 3 regions with each region having 1 interface or 1 connecting route. Individual movement is expressed by a weight function which in modeling use the Kernel density function in the normal group. The mathematical model obtained is in the form of a System of Integro-Partial Differential Equations consisting of 3 regional sub-models and an entire regional system model. Leipzig constant analysis was carried out in order to obtain model validation that was suitable for the phenomenon that occurred. [ABSTRACT FROM AUTHOR]

The increase of job competition is a challenge for job seekers as well as for organizations to find and retain the right and competitive people. Inadequate recruitment processes will result in lower productivity and turnover. Therefore the organization needs an accurate decision support system to get employees who match the qualifications of the organization. This study aims to design support vector machine (SVM) -based decision support for employee recruitment. The data used are 400 data which are divided into two classes, namely accepted and not accepted. The use of different kernel functions in the design of the SVM classification model shows that the polynomial kernel function is based on the experimental parameter value C is 23, d is 5 and the ratio experiment of 80% and 20%, the accuracy is 97.50% and the AUC is 0.99. The results of this study indicate that the designed classification model has a good performance in employee recruitment. [ABSTRACT FROM AUTHOR]

The massive growth of abnormal brain cells leads to Brain tumor. The accurate detection of brain tumor is thus imperative. In this paper, to obtain satisfactory segmentation of brain MRI image, optimized Fuzzy C means clustering is used. The proposed method consists of various steps. First, to upgrade the quality of brain MRI image, contrast enhancement is used. For skull stripping, morphological operations are presented. Second, its result is fed to segmentation stage, where optimized Fuzzy C means clustering with genetic algorithm is utilized. Next, feature extraction with combination of GLRLM and GLCM method is proposed. The SVM classifier, based on extracted features, classifies tumor and normal brain images. The training of SVM is carried out using 150 MRI images dataset. More accurate results are obtained using 5 kernels methods that are Linear, Radial Basis Function, Quadratic, Polynomial and MLP kernel. The highest accuracy of 94.8% is obtained using quadratic kernel function. Successfully, proposed method amends rate of accuracy and identifies tumor and non-tumor images. [ABSTRACT FROM AUTHOR]

We investigated unsteady elastic diffusion torsional vibrations of a simply supported rectangular isotropic Kirchhoff-Love plate. The problem solution is sought in integral form, and Green's functions are kernels of these integrals. We used the Laplace transform in time and the expansion into double trigonometric Fourier series in spatial coordinates to find the Green's functions. The Green's functions in the image domain are represented as rational functions depending on the Laplace transform parameter. The transition to the original domain is done analytically through residues and tables of operational calculus. Analytical expressions of the Green's functions are obtained. A study of unsteady mechanical and diffusion fields interaction is done for an isotropic plate using a three-component continuum. [ABSTRACT FROM AUTHOR]

In this paper, we propose a new method for solving linear partial fuzzy integro-differential equations (PFIDE)with a convolution kernel using double fuzzy Natural transform (DFNT). We provide fundamental results of the double fuzzy Natural transform for fuzzy partial derivatives of the m-th order and fuzzy convolution. By using them we construct the solution of PFIDE. We show that DFNT method is a simple and reliable approach for solving such equation analytically. Finally, we give illustrative example to show the ability of the proposed method. [ABSTRACT FROM AUTHOR]

The paper develops the forecasting model for time series via Kernel-based fuzzy clustering algorithm (KFC). This algorithm not only finds the appropriate number of groups, clustering for time series but also determines the fuzzy relationship of elements inside of time series. One of the most advantages of the proposed method is the improvement of fuzzy clustering problem based on the Kernel function to increase the fuzzy relationship between centroid clusters and elements in series. Experimental results show that the proposed algorithm has more advantages than other ones. In addition, this algorithm also shows potential and effective in comparing with other algorithms. [ABSTRACT FROM AUTHOR]

In this note, the homotopy analysis method (HAM) is applied as a tool for solving the system of non-linear Fredholm-Volterra integral equations. The generalized chain rule is implemented for differentiation of the non-linear kernel functions with many variables, and the non-linear problem is reduced into a sequence of known non-linear integral equa- tions. The proposed method is compared with the homotopy perturbation method (HPM) given in Yusufogli [20] and Biazar and Ghazyini [21]. Numerical results reveal that the proposed method is very effective with high accuracy and dominated the HPM. [ABSTRACT FROM AUTHOR]

A growing complexity of modern engineering problems emphasises the need to methods working with discrete mathematical structures, such as methods of discrete potential theory and discrete function theory, are gaining more and more popularity nowadays. However, solution procedure of boundary value problems by these methods requires at first a construction of the discretisation of a given domain. Thus, in this paper, we present a general discretisation algorithm for discrete potential and function theories. The algorithm allows a consistent discretisation of a domain for interior and exterior boundary value problems based only on a definition of a given continuous domain; and the complete discretisation procedure is based on the application of well-known set-theoretic operations originating from the characterisation of mesh points by help of star-Laplacian. Thus, the algorithm results in a rectangular lattice established over an arbitrary two-dimensional domain and its exterior. [ABSTRACT FROM AUTHOR]

Increased humidity of building envelopes frequently leads to the appearance and growth of mold, which is one of the most important factors concerning Sick Building Syndrome evaluation. Detection of Volatile Organic Compounds (VOCs) emitted by the fungi can be performed using gas sensor arrays. Array output in a form of multidimensional electric signals has to be analyzed by means of appropriate statistical methods. The idea presented within this paper is to use Support Vector Machine (SVM) in the classification of the mold-infested buildings because of different admixture of fungal VOCs within their indoor atmosphere. The mappings used by SVM schemes are designed to ensure that dot products of pairs of input data vectors are computed in terms of the variables in the original space, by defining them in terms of a kernel function k(x,y) selected to suit the problem. Using different types of kernel function actual levels of contamination could be assessed based on readouts from a Metal Oxide Semiconductor (MOS) sensors array. SVM method is appropriated for situations where the functional form of the relationship between signals from sensor array and the classes of contamination is unknown or relationship is complex. The multidimensional sensor readouts pertaining to the air sampled near the building envelopes in varying degree of mold-contamination, compared with clean and synthetic air, are interpreted and presented. [ABSTRACT FROM AUTHOR]

Stationarity is an essential requirement in modeling GSTAR space-time. GSTAR modeling adopted from the three iterative Box-Jenkins time series modeling. The stages are model identification, parameter estimation, and validation. In the validation, stage consists of two tests, namely parameter stationarity test and residual test. In the parameter stationary test, the Eigenvalue method has been used, in this paper the inverse matrix Autocovariance M ˜ 1 the technique is used. The spatial weight matrix of the GSTAR model, uses a kernel approach so that it is random. The results obtained that A GSTAR(1;1) process with kernel weight matrix which has IAcM is stationary if and only if the M ˜ 1 matrix is positive definite and the modulus from all the eigenvalues of the matrix parameter is less than 1. [ABSTRACT FROM AUTHOR]

As the progress of network transmission media need to be secured as data communication is strongly felt and as the volume of media shared (Video, image, audio) increase, although acceleration speed of encryption cannot retain aside. Lately, fractal-based cryptosystem has become a focus of active research in computer network system because of its disordered behaviour. Nowadays, Numba supports CUDA GPU programming by directly compiling a constrained subset of Python code into CUDA kernels and device functions following the CUDA execution model. The proposed method makes the comparison of process speed between of Mandelbrot function between GPU-Numba over CPU as Mandelbrot develop a non-transitional key cryptosystem. The process starts with the CPU speed take to encrypt image using Mandelbrot function with the help of few python libraries. After executing on CPU. The method was implemented as well by GPUNumba to accelerate the processing speed. As per experimental result and performance analysis, to encrypt an image took 30 minutes on CPU while n GPU Numba took only 20 seconds to encrypted the same image as CPU. As a result, the CUDA Numba has high optimized due to its fast which marks the process of securing image with efficient execution time which helps to achieve a secure communication network situation for data transmission. [ABSTRACT FROM AUTHOR]

In this paper, we consider the linear ill-posed inverse problem with noisy data in the statistical learning setting. The Tikhonov regularization scheme in Hilbert scales is considered in the reproducing kernel Hilbert space framework to reconstruct the estimator from the random noisy data. We discuss the rates of convergence for the regularized solution under the prior assumptions and link condition. For regression functions with smoothness given in terms of source conditions the error bound can explicitly be established. [ABSTRACT FROM AUTHOR]

In this paper we refer to the computational issues in solving Maxwell’s curl equations without using any connectivity among the points in which the problem domain is discretized. The adopted procedure is able to approximate the electric and magnetic vector fields making use of the derivatives of a kernel function at points arranged in the computational domain. In order to improve the numerical accuracy a suitable strategy is considered. The computational core of the overall process requires elementary linear algebra operations. In the paper the method is presented and the discussion is addressed to the computational issues. Moreover, some numerical simulations are presented to validate the numerical process. [ABSTRACT FROM AUTHOR]

A numerical solution of the Fredholm integral equations can be obtained by many methods. Most of them lead to a solution of a system of linear equations with fully populated matrices. In the case of collocation or product integration methods, each element of the matrix is an integral, which needs to be calculated. It causes high computing time in multidimensional problems. Computing time can be reduced by the Nyström method. It is based on substitution of the integral by a numerical integration rule. It has the advantage that only diagonal elements of the matrix are integrals. When the kernel function is singular, a singularity subtraction is needed. However it can not be used for every kernel function and every integration rule. The main point of this paper is the convergence conditions of the Nyström method as applied to a special multidimensional integral equation. The paper includes an illustrative example. [ABSTRACT FROM AUTHOR]

Edge waste is an inevitable consequence of the cast film extrusion process, since the edges of the film are trimmed due to dimensional distortion and unpleasant layer distribution. On the other hand, most of the materials which are used in film production are not biodegradable. Therefore, for protecting the environment, the waste output should be reduced. For this reason, nowadays film production lines are equipped with inline recycling units which are also acceptable from the economic aspect, although, it caused degradation in polymer materials. Upon degradation, the film quality decreases which is associated with the molecular weight and molecular weight distribution (MWD) alterations. Despite this, film quality should be kept at a high rate as much as possible to be acceptable in the market. Hence, chasing MWD which is changed during the recycling process, is one of the hints for controlling the film quality. In this study, the degradation of the film grade polypropylene through recycling by twin screw extruder at 170°C and 220°C was investigated. It was assumed the material in the extruder was degraded by extremely high temperature and shear rate. The degradation of polypropylene subjected to multiple extrusions was evaluated by following the changes in the MWD curves in each run, which were indirectly obtained by rheometry based on the generalizing mixing rule. In this equation, Kernel function describes the relaxation behavior of a monodisperse component of molecular weight, and time is related to molecular weight, and the exponent ß characterizes the mixing behavior of the chains. As its calibration, gel permeation chromatography (GPC) was also used. In this regard, two values of ß (ß=2.5 and ß=3) and different kernel functions such as Tumilino, single exponential and Doi-Edward were used. According to the MWDs obtained from rheometry and GPC data, chain scission during the recycling processes was evaluated. [ABSTRACT FROM AUTHOR]

The intended work uses the TKEMPR method which decomposes a linear integral operator on univariate functions by using high dimensional modelling with the basic idea to use repeated bivariate Enhanced Multivariance Products Representation (EMPR) technique which is called Tridiagonal Kernel Enhanced Multivariance Products Representation (TKEMPR). It uses EMPR bivariate function decomposition consecutively such that in each step the remainder term is expanded to again a bivariate EMPR but with different support functions. [ABSTRACT FROM AUTHOR]

The purpose of this paper is to introduce a new method we call “Tridiagonal Folded Kernel Enhanced Multivariance Products Representation (TFKEMPR)”, which involves both Tridiagonal Kernel Enhanced Multivariance Products Representation (TKEMPR) and Tridiagonal Folmat Enhanced Multivariance Products Representation (TFEMPR) as its very specific circumstances. [ABSTRACT FROM AUTHOR]

The mechanical and physical properties of palm kernel shell (PKS) reinforced with high-density polyethylene (HDPE) have been investigated by using there different filler loading of 5wt.%, 10wt.% and 15wt.%. The PKS filler was prepared for two different particle sizes namely fine size (150μm-212μm) and coarse size (300μm- 422μm). The HDPE/PKS composites were compounding by using twin screw extruder and turn them into pallet form. Then, pallet form of HDPE/PKS composites sample was prepared with an injection moulded machine produced dogbone shape of samples. Tensile test and flexural test has been used as the mechanical testing for this study. Based on the results of tensile strength, the fine size PKS provided better result than the coarse size PKS for the 10wt.% PKS loading. Optimum PKS loading was the sample with 10wt.% PKS loading comes with fine size PKS as the finer the size of PKS give better tensile strength. [ABSTRACT FROM AUTHOR]

In this study, the alkali treatment of palm kernel shell using different concentration (5% and 10%) of sodium hydroxide (NaOH) is performed. The interaction between palm kernel shell and high-density polyethylene (HDPE) was examined by Scanning Electron Microscopy (SEM). For comparison purpose, the treated and untreated palm kernel shell filler was reinforced in High Density Polyethylene (HDPE) matrix and have been tested under tensile and flexural modes using Universal Testing Machine (UTM). The results indicate that the strength and modulus of elasticity properties were improved in treating filler composites for both tensile and flexural modes. [ABSTRACT FROM AUTHOR]

Cancer is one disease that needs a proper treatment. There are more than 150 types of cancer, one of them is brain cancer. Taking advantage from microarray data, machine learning methods can be applied to help brain cancer prediction according to its type. This problem can be referred to as a multiclass classification problem. Using one versus one approach, the multiclass problem with k classes can be transformed into k(k+1)/2 binary class classification. To improve the accuracy, the features candidate will be evaluated using feature selection. In this research, Kernel Function is implemented as the feature selection method and Multiple Support Vector Machine (MSVM) method is implemented as the classification method. The results obtained showed the comparison accuracy of MSVM use and without feature selection. [ABSTRACT FROM AUTHOR]

The present work deals with the numerical solution for a system of second-ordered singular integral equations of the first kind. The analytical solution of hypersingular operator is obtained and used to develop an efficient approximate solution. Relating the Hadamard integrals to Cauchy principal-value integrals, we expand the singular kernel and the density function of hypersingular integral equation based first and second kinds Chebyshev series of order N. Furthermore, the analytical findings are applied successfully for the mathematical model of antenna work problem where the kernel function has a discontinuity at the feeding points. A numerical example are given to demonstrate the performance of the present schemes. [ABSTRACT FROM AUTHOR]

Linear regression is a method that can be used to model the relationship between a dependent variable to one or more independent variables. There are some assumptions that must be fulfilled in the linear regression model, such as the error term is normally distributed with mean zero, constant error variance (homoscedasticity), and the error between observations are independent. When analyzing spatial data using a linear regression model, sometimes the homoscedastic assumption cannot be fulfilled because data condition on one location can be different with data condition in other location. Geographically Weighted Regression (GWR) model can be used to overcome the spatial heterogeneity problem. Parameters of GWR model can be estimated using Weighted Least Squares (WLS) method as basic of estimating parameters. As the weight is kernel weighting function. Kernel weighting function used in this paper is Gaussian kernel weighting function. There is an example of the GWR model application by using inpatient claims data of PT. XYZ members to see the relationship between the total inpatient cost to the hospitalization duration and hospital's room type for Typhoid Fever. Based on the map of parameter estimation on GWR model, it can be seen that there is a variation of the total inpatient cost in every subjects location. If only the linear regression model is used to analyze this data, there will be a misleading interpretation so that it is suitable to model the data with GWR model. [ABSTRACT FROM AUTHOR]

This paper describes one method for a numerical solution of a Fredholm integral equation of the second kind. The kernel function has a special type of singularity called the diagonal one. The main result of this paper is the error estimation of the numerical solution based on the Nystr¨om method with singularity subtraction. The error estimate is compared to numerical examples. The error behavior and computing time of the Nytr¨om method is compared with alternative method. [ABSTRACT FROM AUTHOR]

The numerical reactor system development is a hot topic in the field of high-performance computing. There are many phases in the entire development life cycle of the numerical reactor with regards to a variety of domains. These stages are discussed exhaustively in this paper respectively. Several fundamental modules or abilities, which should be incorporated into the numerical reactor, are defined for the purpose of high-precision calculations. Meanwhile, a hierarchical architecture of the system is depicted for a reference. [ABSTRACT FROM AUTHOR]

Molecular dynamics simulation is an important tool for studying materials microstructure evolution under radiation effects. It is difficult to simulate atomic diffusion in materials owing to the time-scale limitations of the molecular dynamics method. Accelerated molecular dynamics has been developed as a solution and the parallel replica method is the simplest and most accurate of the accelerated dynamics techniques. The simulation time scale can reach several orders more than the direct molecular dynamics, while retaining the complete atomic detail. In parallel replica method, the entire system is replicated on each of M available parallel or distributed processors, it is in line with the characteristics of high-performance parallel computing and can be optimized from the program level. [ABSTRACT FROM AUTHOR]

Thermal-Hydraulics computing softwares play an important role in nuclear reactor design and safety analysis. Among them, Nek5000 is an excellent open-source Computational Fluid Dynamics based Thermal-Hydraulics computing software. To in-depth investigate the performance of Nek5000, Exascale Simulation of Advanced Reactors research group has developed Nekbone mini-application. Nek5000 resembles the kernel basic structure and functionalities of the complete Nek5000 software. We investigate the weak and strong scaling performance on two Chinsed supercomputers which are Tianhe II and Era. Tianhe II was crowned as the fastest supercomputer in 2013 by TOP500 list. The average MFlops and Time-scaling performance metric represent the excellent weak scaling performance of Nekbone on both Tianhe II an Era supercomputers. [ABSTRACT FROM AUTHOR]

This paper presents a new framework for the development of generalized composite kernels machines for discrete Hopfield neural network and to upgrading the performance of logic programming in Hopfield network by applying kernels machines in the system. We put up a new family of generalized composite kernels which exhibit great flexibility when combining the spectral and the spatial information contained in the analyze data that make use of kernels. The present an analysis based on kernels machines (Linear kernel Hopfield neural network, Polynomial kernel Hopfield neural network, and Gaussian kernel Hopfield neural network) that measure how the capacity of Hopfield network to use for solving the combinatorial problem that always occurs in Hopfield network and guaranty that the solution is optimal. This work is merely focusing on the ways to upgrade the performance of logic programming in Hopfield network. We carried out computer simulations to demonstrate the ability of kernels Hopfield neural network in enhancing the performance of the system. By applying kernels Hopfield neural network in the system, it does not only produce better quality solutions but it also can handle the network better even though the complexity increased. Besides that, the system also makes the solutions converge faster. Thus, the presence of this kernels Hopfield neural network in the system will produce solutions with better quality [ABSTRACT FROM AUTHOR]

In this letter, we propose a novel Laplacian least squares support vector machine in sum space (Lap-LS-SVM-SS) for semi-supervised HIC with multiscale kernels, which can simultaneously deal with the high- and low- frequency components of the target classification function with small and large scale kernels, respectively. The proposed method can solve the multi-class problem of hyperspectral dataset directly. The experimental results on AVIRIS Indian Pines dataset demonstrate that Lap-LS-SVM-SS yields superior classification performance over the conventional semi-supervised classifiers (e.g., LapSVM) in challenging small training labeled samples and a large amount of unlabeled samples. [ABSTRACT FROM AUTHOR]

This paper proposed an improved image registration method combining Hu-based invariant moment contour information and feature points detection, aiming to solve the problems in traditional image stitching algorithm, such as time-consuming feature points extraction process, redundant invalid information overload and inefficiency. First, use the neighborhood of pixels to extract the contour information, employing the Hu invariant moment as similarity measure to extract SIFT feature points in those similar regions. Then replace the Euclidean distance with Hellinger kernel function to improve the initial matching efficiency and get less mismatching points, further, estimate affine transformation matrix between the images. Finally, local color mapping method is adopted to solve uneven exposure, using the improved multiresolution fusion algorithm to fuse the mosaic images and realize seamless stitching. Experimental results confirm high accuracy and efficiency of method proposed in this paper. [ABSTRACT FROM AUTHOR]

Let Ω be a homogeneous function with zero degree on ℝn. Fractional integral operator with rough kernel TΩ,α maps function f to convolution between f and kernel function KΩ,α = Ω(x)|x|α-n where 0 < α < n. In this paper, we show Spanne type result and Adam type result for TΩ,α on generalized Morrey spaces. [ABSTRACT FROM AUTHOR]

We consider a generalized Langevin equation with an exponentially decaying memory kernel as a model for the firing process of a resonate-and-fire neuron. The effect of temporally correlated random neuronal input is modeled as Ornstein-Uhlenbeck noise. In the noise-induced spiking regime of the neuron, we derive exact analytical formulas for the dependence of some statistical characteristics of the output spike train, such as the probability distribution of the interspike intervals (ISIs) and the survival probability, on the parameters of the input stimulus. Particularly, on the basis of these exact expressions, we have established sufficient conditions for the occurrence of memory-time-induced transitions between unimodal and multimodal structures of the ISI density and a critical damping coe_cient which marks a dynamical transition in the behavior of the system. [ABSTRACT FROM AUTHOR]

Based on the dynamics of a generalized Langevin equation subjected to a power-law-type memory kernel with a finite memory time a resonate-and-fire neuron model is considered. The effect of temporally correlated random neuronal input is modeled as a Mittag-Leffler noise. Using a first-passage-time formulation, the exact expression for the output interspike interval density is derived and its dependence on input parameters, especially on the memory time and on the memory exponent, is analyzed. Particularly, in the case of external white noise it is shown that at intermediate values of the memory exponent the survival probability (the probability that spikes are not generated) is significantly enhanced in comparison with the cases of strong and weak memory, which causes a resonance-like suppression of the interspike interval distribution versus the memory exponent. Similarities and differences between the behaviors of the models at internal and external noises are also discussed. [ABSTRACT FROM AUTHOR]

Proposed a new method of constructing nonparametric dynamic models of the oculomotor system system (OMS) in the form of human multidimensional transition functions on the basis of experimental data "input-output". As the test signals used bright points on the long duration of the computer screen. OMS response is measured using information technology Eye-tracking and recorded on video. As a result data processing of the experiment we receive function based "pupil coordinate - time". Using the method of least squares (Ordinary Least Squares, OLS) defined transition functions of the first, second and third order - integral transformations of Volterra kernels, representing a model of OMS. Completed experimental studies using computer simulations confirm the adequacy of the constructed approximation model as a real system. [ABSTRACT FROM AUTHOR]

A semi-Markov (SM) process is defined by a renewal kernel and an initial distribution of states or another equivalent parameters. Those quantities contain full information about the process and they allow us to find many characteristics and parameters of the process. Constructing the semi-Markov reliability model means building the kernel of the process based on some assumptions. Many characteristics and parameters of the SM process have a natural interpretation in the semi-Markov reliability model. [ABSTRACT FROM AUTHOR]

Although the correlation filter-based trackers achieve the competitive results both on accuracy and robustness, there are still some problems in handling scale variations, object occlusion, fast motions and so on. In this paper, a multiscale kernel correlation filter algorithm based on random fern detector was proposed. The tracking task was decomposed into the target scale estimation and the translation estimation. At the same time, the Color Names features and HOG features were fused in response level to further improve the overall tracking performance of the algorithm. In addition, an online random fern classifier was trained to re-obtain the target after the target was lost. By comparing with some algorithms such as KCF, DSST, TLD, MIL, CT and CSK, experimental results show that the proposed approach could estimate the object state accurately and handle the object occlusion effectively. [ABSTRACT FROM AUTHOR]

Support vector machine is becoming more and more popular in classification. In this paper, we focus on the nonhomogeneous polynomial kernels and present a new way on the choice of the exponent d to the polynomial. We conclude that the optimal d in terms of accuracy rate is always in the range constructed by our method and our new modified support vector machine algorithm is fast and efficient. [ABSTRACT FROM AUTHOR]