Radial Kernel Research Articles

On the potential of a graph attention network in money laundering detection

PurposeIn a classroom, a support vector machines model with a linear kernel, a neural network and the k-nearest neighbors algorithm failed to detect simulated money laundering accounts generated from the Panama papers data set of the offshore leak database. This study aims to resolve this failure.Design/methodology/approachBuild a graph attention network having three modules as a new money laundering detection tool. A feature extraction module encodes these input data to create a weighted graph structure. In it, directed edges and their end vertices denote financial transactions. Each directed edge has weights for storing the frequency of money transactions and other significant features. Social network metrics are features of nodes for characterizing an account’s roles in a money laundering typology. A graph attention module implements a self-attention mechanism for highlighting target nodes. A classification module further filters out such targets using the biased rectified linear unit function.FindingsResulted from the highlighting of nodes using a self-attention mechanism, the proposed graph attention network outperforms a Naïve Bayes classifier, the random forest method and a support vector machines model with a radial kernel in detecting money laundering accounts. The Naïve Bayes classifier produces second accurate classifications.Originality/valueThis paper develops a new money laundering detection tool, which outperforms existing methods. This new tool produces more accurate detections of money laundering, perfects warns of money laundering accounts or links and provides sharp efficiency in processing financial transaction records without being afraid of their amount.

Numerical simulation of a degenerate parabolic problem occurring in the spatial diffusion of biological population

This paper studies a localized meshless algorithm for calculating the solution of a nonlinear biological population model (NBPM). This model describes the dynamics in the biological population and may provide valuable predictions under different scenarios. The solution of the NBPM is approximated by means of local radial basis function based on the partition of unity (LRBF-PU) technique. First, the partial differential equation (PDE) is converted into a system of ordinary differential equations (ODEs) with the help of radial kernels. Afterwards, the system of ODEs is solved through an ODE solver of high-order. The major advantage of this scheme is that it does not requires any linearization. The LRBF-PU approximation helps handling the issue of the matrix ill conditioning that arises in a global RBF approximation. Three examples highlight the efficiency and accuracy of the numerical method. It is verified that the proposed strategy is more efficient in terms of computational time and accuracy in comparison with others available in the literature.

ON POINTWISE APPROXIMATION OF FUNCTIONS OF SEVERAL VARIABLES BY SINGULAR INTEGRALS

We prove a theorem on weighted pointwise convergence of \ multidimensional integral operators with radial kernels to generating function of several variables, which are in general non-integrable in $n$-dimensional Euclidean space $E_{n}$ in the sense of Lebesgue$.$ Main result holds at almost every point of $E_{n}.$

Validation of ground truth fire debris classification by supervised machine learning

A set of 767 laboratory-generated fire debris samples of known ground truth as to whether an ignitable liquid residue was present (class IL) or absent (class SUB) were used to train five machine learning classifiers. Linear and quadratic discriminant analysis (LDA and QDA), k-nearest neighbors (kNN), and support vector machines with radial and linear kernels (SVMr and SVMl) were tested for their performance in correctly classifying the fire debris samples into class IL or class SUB. Each classifier was trained and tested/validated on 500 class-balanced data sets, each comprised of 400 fire debris samples (200 IL and 200 SUB) that were bootstrapped from the 767 laboratory-generated samples. Each bootstrapped data set was split into subsets for training (75%, 300 samples) and testing/validation (25%, 100 samples). The LDA, SVMr and SVMl were found to give satisfactory performance based on area under the receiver operating characteristic curve (0.86–0.92), equal error rates (17%−22%) and well-calibrated probabilities. The three satisfactory classifiers were further applied to a set of 129 fire debris samples produced in large-scale test burns. The classifications generated by the machine learning models were compared with the sample classes assigned by an informed analyst having knowledge of the chromatographic patterns of the ignitable liquids use to start the large-scale fires. The LDA and SVMl models gave results most closely aligned with the informed analyst.

Gene Polymorphisms of the Renin-Angiotensin System and Bleeding Complications of Warfarin: Genetic-Based Machine Learning Models.

This study aimed to investigate the effects of genetic variants and haplotypes in the renin–angiotensin system (RAS) on the risk of warfarin-induced bleeding complications at therapeutic international normalized ratios (INRs). Four single nucleotide polymorphisms (SNPs) of AGT, two SNPs of REN, three SNPs of ACE, four SNPs of AGTR1, and one SNP of AGTR2, in addition to VKORC1 and CYP2C9 variants, were investigated. We utilized logistic regression and several machine learning methods for bleeding prediction. The study included 142 patients, among whom 21 experienced bleeding complications. We identified a haplotype, H2 (TCG), carrying three single nucleotide polymorphisms (SNPs) of ACE (rs1800764, rs4341, and rs4353), which showed a significant relation with bleeding complications. After adjusting covariates, patients with H2/H2 experienced a 0.12-fold (95% CI 0.02–0.99) higher risk of bleeding complications than the others. In addition, G allele carriers of AGT rs5050 and A allele carriers of AGTR1 rs2640543 had 5.0- (95% CI 1.8–14.1) and 3.2-fold (95% CI 1.1–8.9) increased risk of bleeding complications compared with the TT genotype and GG genotype carriers, respectively. The AUROC values (mean, 95% CI) across 10 random iterations using five-fold cross-validated multivariate logistic regression, elastic net, random forest, support vector machine (SVM)–linear kernel, and SVM–radial kernel models were 0.732 (0.694–0.771), 0.741 (0.612–0.870), 0.723 (0.589–0.857), 0.673 (0.517–0.828), and 0.680 (0.528–0.832), respectively. The highest quartile group (≥75th percentile) of weighted risk score had approximately 12.0 times (95% CI 3.1–46.7) increased risk of bleeding, compared to the 25–75th percentile group, respectively. This study demonstrated that RAS-related polymorphisms, including the H2 haplotype of the ACE gene, could affect bleeding complications during warfarin treatment for patients with mechanical heart valves. Our results could be used to develop individually tailored intervention strategies to prevent warfarin-induced bleeding.

Classification of oils by ECOC based multi-class SVM using spectral analysis of acoustic signals

Acoustic signals passing through a material gather a great deal of analytical information. Such information act as a signature for classification of the materials. Therefore, the paper proposes a novel method for classification of oils by extracting spectral features from acoustic signals. A developed acoustic resonance spectrometry system has been used for the acquisition of acoustic signals from different oils. The system has a quartz V-tube and two similar piezoelectric transducers, attached at the two ends of the tube. One of the transducer acts as a transmitter and the others as a receiver. The transmitter generates the vibrations with the excitation of white noise. The vibrations transmitted through the tube along with oil from the transmitter end to the receiver end. At the receiver end, vibrations translated into the acoustic signal that has been recorded and analyzed. The experiments were performed on six different oils samples namely: Castor, Mustard, Olive, D.Almond, Petrol, and Tarpin. The Burg spectral analysis was carried out on acoustic signals and eight resonance peaks frequency were recorded. The principal component analysis was done on the collected resonance frequencies datasets. The first three principal components were given as input to ECOC SVM classifier using radial basic function kernel in five different output coding schemes one versus one (1VS1), one versus all (1VSA), ordinal (ORD), and binary complete (BNC). The accuracies of SVM_1VS1, SVM_1VSA, SVM_ORD, and SVM_BNC are found to be 100% in classifying Castor, Mustard, Olive, and Tarpin. The best classifier in classifying D.Almond and Petrol is SVM_ORD that has accuracies of 96.75% and 91.47% respectively. The total classification accuracies of classifiers SVM_1VS1, SVM_1VSA, SVM_ORD, and SVM_BNC are 95.71%, 93.08%, 98.03%, and 95.47% respectively. Hence, the variation in resonant peaks in the Burg autoregressive spectrum of acoustic signals showed significant potential in classifying the oils.

Numerical Solution of Fractional Order Anomalous Subdiffusion Problems Using Radial Kernels and Transform

By coupling of radial kernels and localized Laplace transform, a numerical scheme for the approximation of time fractional anomalous subdiffusion problems is presented. The fractional order operators are well suited to handle by Laplace transform and radial kernels are also built for high dimensions. The numerical computations of inverse Laplace transform are carried out by contour integration technique. The computation can be done in parallel and no time sensitivity is involved in approximating the time fractional operator as contrary to finite differences. The proposed numerical scheme is stable and accurate.

Semi‐analytical smoothed‐particle hydrodynamics correction factors for polynomial kernels and piecewise‐planar boundaries

AbstractThis work presents a method of exactly and efficiently evaluating integrals of compactly supported (and otherwise) polynomial radial kernels near piecewise‐planar boundaries. The technique is motivated by and has an immediate application in the smoothed‐particle hydrodynamics method employing the semi‐analytical boundary formulation. The current implementation is exact and fully general, avoiding any need for symbolic or numerical integration which previous implementations rely on. A detailed derivation and three compact implementations are provided and the latter can be easily modified to fit new and existing simulation codes. Applications and good performance are demonstrated on a number of test problems, where the semi‐analytical boundary method can simulate complex boundaries using roughly half the boundary particles required in a ghost particle boundary method.

Comparative Study of Several Machine Learning Algorithms for Classification of Unifloral Honeys.

Unifloral honeys are highly demanded by honey consumers, especially in Europe. To ensure that a honey belongs to a very appreciated botanical class, the classical methodology is palynological analysis to identify and count pollen grains. Highly trained personnel are needed to perform this task, which complicates the characterization of honey botanical origins. Organoleptic assessment of honey by expert personnel helps to confirm such classification. In this study, the ability of different machine learning (ML) algorithms to correctly classify seven types of Spanish honeys of single botanical origins (rosemary, citrus, lavender, sunflower, eucalyptus, heather and forest honeydew) was investigated comparatively. The botanical origin of the samples was ascertained by pollen analysis complemented with organoleptic assessment. Physicochemical parameters such as electrical conductivity, pH, water content, carbohydrates and color of unifloral honeys were used to build the dataset. The following ML algorithms were tested: penalized discriminant analysis (PDA), shrinkage discriminant analysis (SDA), high-dimensional discriminant analysis (HDDA), nearest shrunken centroids (PAM), partial least squares (PLS), C5.0 tree, extremely randomized trees (ET), weighted k-nearest neighbors (KKNN), artificial neural networks (ANN), random forest (RF), support vector machine (SVM) with linear and radial kernels and extreme gradient boosting trees (XGBoost). The ML models were optimized by repeated 10-fold cross-validation primarily on the basis of log loss or accuracy metrics, and their performance was compared on a test set in order to select the best predicting model. Built models using PDA produced the best results in terms of overall accuracy on the test set. ANN, ET, RF and XGBoost models also provided good results, while SVM proved to be the worst.

Support vector regression is an improvement for principle component analysis

The principle component analysis is used more frequently as a variables reduction technique. And recently, an evolving group of studies makes use of machine learning regression algorithms to improve the estimation of empirical models. One of the most frequently used machines learning regression models is support vector regression with various kernel functions. However, an ensemble of support vector regression and principal component analysis is also possible. So, this paper aims to investigate the competence of support vector regression techniques after performing principal component analysis to explore the possibility of reducing data and having more accurate estimations. Some new proposals are introduced and the behavior of two different models 𝜀𝜀-SVR and 𝑣𝑣-SVR are compared through an extensive simulation study under four different kernel functions; linear, radial, polynomial, and sigmoid kernel functions, with different sample sizes, ranges from small, moderate to large. The models are compared with their counterparts in terms of coefficient of determination (𝑅𝑅2 ) and root mean squared error (RMSE). The comparative results show that applying SVR after PCA models improve the results in terms of SV numbers between 30% and 60% on average and it can be applied with real data. In addition, the linear kernel function gave the best values rather than other kernel functions and the sigmoid kernel gave the worst values. Under 𝜀𝜀-SVR the results improved which did not happen with 𝑣𝑣-SVR. It is also drawn that, RMSE values decreased with increasing sample size.

Performance analysis of sentiments in Twitter dataset using SVM models

Sentiment Analysis is a current research topic by many researches using supervised and machine learning algorithms. The analysis can be done on movie reviews, twitter reviews, online product reviews, blogs, discussion forums, Myspace comments and social networks. The Twitter data set is analyzed using support vector machines (SVM) classifier with various parameters. The content of tweet is classified to find whether it contains fact data or opinion data. The deep analysis is required to find the opinion of the tweets posted by the individual. The sentiment is classified in to positive, negative and neutral. From this classification and analysis, an important decision can be made to improve the productivity. The performance of SVM radial kernel, SVM linear grid and SVM radial grid was compared and found that SVM linear grid performs better than other SVM models.

Large mass minimizers for isoperimetric problems with integrable nonlocal potentials

This paper is concerned with volume-constrained minimization problems derived from Gamow’s liquid drop model for the atomic nucleus, involving the competition of a perimeter term and repulsive nonlocal potentials. We consider a large class of potentials, given by general radial nonnegative kernels which are integrable on Rn, such as Bessel potentials, and study the behavior of the problem for large masses (i.e., volumes). Contrary to the small mass case, where the nonlocal term becomes negligible compared to the perimeter, here the nonlocal term explodes compared to it. However, using the integrability of those kernels, we rewrite the problem as the minimization of the difference between the classical perimeter and a nonlocal perimeter, which converges to a multiple of the classical perimeter as the mass goes to infinity. Renormalizing to a fixed volume, we show that, if the first moment of the kernels is smaller than an explicit threshold, the problem admits minimizers of arbitrarily large mass, which contrasts with the usual case of Riesz potentials. In addition, we prove that, any sequence of minimizers converges to the ball as the mass goes to infinity. Finally, we study the stability of the ball, and show that our threshold on the first moment of the kernels is sharp in the sense that large balls go from stable to unstable. A direct consequence of the instability of large balls above this threshold is that there exist nontrivial compactly supported kernels for which the problems admit minimizers which are not balls, that is, symmetry breaking occurs.

Novel Descriptors Derived from the Aggregation Propensity of Di- and Tripeptides Can Predict the Critical Aggregation Concentration of Longer Peptides

Self-assembling amphiphilic peptides have recently received special attention in medicine. Nonetheless, testing the myriad of combinations generated from at least 20 coded and several hundreds of noncoded amino acids to obtain candidate sequences for each application, if possible, is time-consuming and expensive. Therefore, rapid and accurate approaches are needed to select candidates from countless combinations. In the current study, we examined three conventional descriptor sets along with a novel descriptor set derived from the simulated aggregation propensity of di- and tripeptides to model the critical aggregation concentration (CAC) of amphiphilic peptides. In contrast to the conventional descriptors, the radial kernel model derived from the novel descriptor set accurately predicted the critical aggregation concentration of the test set with a residual standard error of 0.10. The importance of aromatic side chains, as well as neighboring amino acids in the self-assembly, was emphasized by analysis of the influential descriptors. The addition of very long peptides (70–100 residues) to the data set decreased the model accuracy and changed the influential descriptors. The developed model can be used to predict the CAC of self-assembling amphiphilic peptides and also to derive rules to apply in designing novel amphiphilic peptides with desired properties.

Experimental Comparison of Financial Distress Prediction Models Using Imbalanced data sets

From machine learning perspective, the problem of predicting financial distress is challenging because the distribution of the classes is extremely imbalanced. The goal of this study was comparing the performance of financial distress prediction models for the imbalanced data sets with different proportions. In this study, the data of the previous year before financial distress was used for 760 company year for the time period of 2007-2017. Besides using traditional classifications such as logistic regression, linear discriminant analysis, artificial neural network, and the classification models of least square support vector machine with four kernel functions, random forest and the Knn algorithm, the measures of the area under the curve and Friedman and Nemenyi tests were also utilized to determine the average rank and the difference significance of the Auc of the models. For selecting the models´ optimal parameters, the combined method of grid search optimization and cross validation was used. The results of this experimental study showed that for the balanced and imbalanced datasets with lower proportions, the best performance was for the random forest. For more imbalanced datasets, the best performance belonged to the least square support vector machine with sigmoid, radial, and linear kernel functions; performance of Knn algorithm had no significant difference from the other models and the performance of the artificial neural network was average or appropriate. Also, the performances of the linear logistic regression and linear discriminant analysis were weaker than other nonlinear models.

Strictly positive definite kernels on the 2-sphere: from radial symmetry to eigenvalue block structure

Abstract The paper introduces a new characterization of strict positive definiteness for kernels on the 2-sphere without assuming the kernel to be radially (isotropic) or axially symmetric. The results use the series expansion of the kernel in spherical harmonics. Then additional sufficient conditions are proven for kernels with a block structure of expansion coefficients. These generalize the result derived by Chen et al. (2003, A necessary and sufficient condition for strictly positive definite functions on spheres. Proc. Amer. Math. Soc., 131, 2733–2740) for radial kernels to nonradial kernels.

Anomaly Classification for Earthquake Prediction in Radon Time Series Data Using Stacking and Automatic Anomaly Indication Function

To minimize catastrophic impacts of earthquakes, many researchers have directed their efforts to developing novel computational techniques and models that forecast natural disasters based upon earthquake catalogues, anomalies in the ionosphere, and radon time series databases, etc. In this article, we propose a methodology that categorizes the soil radon gas (SRG) concentration into seismically active and non-active time series by taking the advantage of a stacking ensemble approach and utilizing an automatic anomaly indication function as a post-processing technique. The main idea behind the proposed methodology is of two layers. The first layer uses a stacking ensemble-based approach that incorporates three learners, i.e. a generalized linear model, linear regression and K-nearest neighbors, to train on seismically active and inactive periods to predict SRG concentration. These predictions are then combined with the labeled anomaly data to train a meta-learner, i.e., support vector machine with a radial kernel, that categorizes the series into active and non-active radon time series data. In the second layer, these classifications are then passed to an automatic anomaly indication function that further labels the time series in a group of readings where the level of received indications is greater than or equal to the indication factor. Radon time series (RTS) data has been recorded over a fault line present in Muzaffarabad, a city in the Pakistan territory of Kashmir, ranging from 1 March 2017 to 28 February 2018. Four seismic events occurred during the study period. The conclusion of the study reveals that the proposed methodology accurately locates the anomaly in the RTS data at different window sizes, i.e. with respect to individual days. The evaluation is based purely on the classification after processing of RTS data by both layers of the proposed methodology.

Komparasi model support vector machine dan backpropagation dalam peramalan jumlah wisatawan mancanegara di provinsi Bali

Tourism in Bali is one of the major industries which play an important role in developing the global economy in Indonesia. Good forecasting of tourist arrival, especially from foreign countries, is needed to predict the number of tourists based on past information to minimize the prediction error rate. This study compares the performance of SVM and Backpropagation to find the model with the best prediction algorithm using data from foreign tourists in Bali Province. The results of this study recommend the best forecasting using the SVM model with the radial kernel function. The best accuracy of the SVM model obtained the lowest error values of MSE 0.0009, MAE 0.0186, and MAPE 0.0276, compared to Backpropagation which obtained MSE 0.0170, MAE 0.1066, and MAPE 0.1539.

Radial kernelized regressive merkle–damgård cryptographic hash blockchain for secure data transmission with IoT sensor node

The concept of the Internet of Things (IoT) has drawn noteworthy attention to use the collected data anywhere in the world. Wireless Sensor Node (WSN) is integrated into the IoT for data collection. The IoT is very intense towards the security investigation. Patient data is medical information about an individual patient. Patient data contains their past and current health, treatment history, lifestyle options and genetic data. The patient data security is a most important issue in IoT. When a collected data get transmitted from the sensor node to the server it theft or altered by various intruders and it may cause a lack of security. In order to address these issues, in this research work, a novel technique called Radial Kernelized Regressive Merkle–Damgard Cryptographic Hash Blockchain (RKRMDCHB) technique is introduced to improve secure data transmission. The wireless sensor nodes are deployed to sense and monitor the data. At first, the sensing data are collected for secure transmission to the server. The RKRMDCHB technique uses the Radial Basis Kernelized Regression Function to perform the data classification. The regression function analyzes the input data and is categorized into various types of classes based on the radial basis kernel function. The classified data is given to the data block of the blockchain for secure transmission. The RKRMDCHB technique uses the Merkle–Damgard Cryptographic technique to generate the hash for each input data with the help of the one-way compression function. The data in the form of a hash is transmitted to the server through the internet. As a result, the RKRMDCHB technique ensures security and confidentiality to preserve the data and provide better communication. Experimental assessment is carried out on certain factors such as data confidentiality rate, data integrity, and processing time, with respect to a number of data sensed from the sensor device. The results demonstrate that the proposed RKRMDCHB technique provides an efficient solution for secure data transmission while preserving sensitive information against potential threats.

International carbon financial market prediction using particle swarm optimization and support vector machine

Carbon financial futures have both the characteristics of commodity futures and environmental protection and its price is affected by many factors. It is hard and complex for traditional analysis methods to get precise prediction results effectively. How to effectively predict the price trend of carbon financial futures has been focused on by both academia and traders. This study addresses the high prediction error of European allowance (EUA) futures price by constructing a novel approach by combining the support vector machine (SVM) and particle swarm optimization (PSO) algorithm. This article introduces a parameters optimization method, which provides the best parameters for SVM to improve the prediction performance of the EUA futures price. Furthermore, this research uses the realistic trading dataset containing 30,762 EUA futures closing prices to verify the effectiveness and efficiency of the PSO-SVM prediction model. The empirical results show that the prediction performance of the model, especially the radial kernel function, is significantly improved. And this approach can determine the parameters according to the characteristics of the dataset and input the parameters for training and prediction automatically. The PSO-SVM algorithm can effectively predict extreme price fluctuations and overcome the problem of high prediction error caused by parameter constraints.

Time-varying coefficient model estimation through radial basis functions

ABSTRACT In this paper, we estimate the dynamic parameters of a time-varying coefficient model through radial kernel functions in the context of a longitudinal study. Our proposal is based on a linear combination of weighted kernel functions involving a bandwidth, centered around a given set of time points. In addition, we study different alternatives of estimation and inference including a Frequentist approach using weighted least squares along with bootstrap methods, and a Bayesian approach through both Markov chain Monte Carlo and variational methods. We compare the estimation strategies mention above with each other, and our radial kernel functions proposal with an expansion based on regression spline, by means of an extensive simulation study considering multiples scenarios in terms of sample size, number of repeated measurements, and subject-specific correlation. Our experiments show that the capabilities of our proposal based on radial kernel functions are indeed comparable with or even better than those obtained from regression splines. We illustrate our methodology by analyzing data from two AIDS clinical studies.