Search

Extreme Support Vector Machine (ESVM), a variant of ELM, is a nonlinear SVM algorithm based on regularized least squares optimization. In this chapter, a regression algorithm, Extreme Support Vector Regression (ESVR), is proposed based on ESVM. Experiments show that, ESVR has a better generalization ability than the traditional ELM. Furthermore, ESVM can reach comparable accuracy as SVR and LS-SVR, but has much faster learning speed. [ABSTRACT FROM AUTHOR]

This paper addresses stock market forecasting indices. Generally, the stock market index exhibits clustering properties and irregular fluctuation. This paper presents the results of using real volatility clustering (RVC) to analyze the clustering in support vector regression (SVR), called ˵real volatility clustering of support vector regression″ (RVC-SVR). Combining RVC and SVR causes the parameters of estimation to become more difficult to solve, thus constituting a highly nonlinear optimization problem accompanied by many local optima. Thus, the genetic algorithm (GA) is used to estimate parameters. Data from the Taiwan stock weighted index (Taiwan), Hang Seng index (Hong Kong), and NASDAQ (USA) were used as the simulation presented in this paper. Based on the simulation results, the stock indices forecasting accuracy performance is significantly improved when the SVR model considers the RVC. [ABSTRACT FROM AUTHOR]

We combine the idea of sampling inequalities and Galerkin approximations of weak formulations of partial differential equations. The latter is a wellestablished tool for finite element analysis. We show that sampling inequalities can be interpreted as Pythagoras law in the energy norm of the weak form. This opens the way to consider regularization techniques known from machine learning in the context of finite elements. We show how sampling inequalities can be used to provide a deterministic worst case error estimate for reconstruction problems based on Galerkin type data. Such estimates suggest an a priori choice for regularization parameter(s). [ABSTRACT FROM AUTHOR]

Support vector regression (SVR) is a powerful machine learning technique in the framework of the statistical learning theory; while Kriging is a well-established prediction method traditionally used in the spatial statistics field. However, the two techniques share the same background of reproducing kernel Hilbert space (RKHS). SVR has recently shown promising performance in different spatial mapping tasks. In the present work, the problem of spatial data mapping is addressed using a multi-scale SVR (MS-SVR) approach. This can be considered as a multi-resolution analysis of the observed process. The multi-scale SVR approach is particularly attractive for its capability to deal, at the same time, with the nonlinear regression of the dependent variable on auxiliary variables and with the spatial interpolation. This capability makes the MS-SVR an optimal choice for automatic mapping system. In the present work MS-SVR was applied to soil heavy metal content mapping, in a study area site in the Italian Alps. The area complex landscape, modelled by both glacial and karsts phenomena, along with an heterogeneous nature of the parent material, makes the mapping of heavy metal content a difficult task to approach with linear regression or mixed geostatistical techniques. The result obtained outlines the Multi-scale SVR as a powerful technique for general inference and automatic mapping, with the only constraint of the requirement of a multi-parameter optimization. [ABSTRACT FROM AUTHOR]

Music video is a well-known medium in music entertainment which contains rich affective information and has been widely accepted as emotion expressions. Affective analysis plays an important role in the content-based indexing and retrieval of music video. This paper proposes a general scheme for music video affective estimation using correlation-based feature selection followed by regression. Arousal score and valence score with four grade scales are used to measure music video affective content in 2D arousal/valence space. The main contributions are in the following aspects: (1) correlation-based feature selection is performed after feature extraction to select representative arousal and valence features; (2) different regression methods including multiple linear regression and support vector regression with different kernels are compared to find the fittest estimation model. Significant reductions in terms of both mean absolute error and variation of absolute error compared with the state-of-the-art methods clearly demonstrate the effectiveness of our proposed method. [ABSTRACT FROM AUTHOR]

In this study, we compare six feature selection methods, i.e. five feature selection methods for k Nearest Neighborhood regression (kNNReg) and a rough set model based forward feature selection (FARNeM) for Support Vector Regression (SVR) for predicting the affinity of TAP binding peptides. The peptides were represented with binary, sequence associated amino acid properties, and binary plus properties of amino acids derived vectors, respectively. The weighted peptide features are input to the regression model and ranked according to the corresponding weights or the occurrence frequency, respectively. We find that SVR model performs better than kNNReg model for the prediction of the affinity of TAP transporter binding peptides. [ABSTRACT FROM AUTHOR]

Traditional spatial interpolation method of spatial interpolation such as geometric method and function method cannot estimate the theoretical error and cannot appreciate the interpolation precision, but statistical methods require data to meet a certain spatial distribution. This paper describes the basic principle of SVR and applies SVR theory to spatial interpolation, combines the SVR with the genetic algorithm for the selection of hyper parameters of SVR has a direct impact on the forecast performance, proposes a new spatial data interpolation algorithm based on SVR optimization by GA. The algorithm uses genetic algorithm to achieve the optimization of the hyper parameters using for SVR. In the process of evolution, this paper uses five folds cross-validation as the fitness function to achieve the training model΄s best generalization ability. Spatial interpolation comparison 97 standard data sets are selected as research object. Experimental results show that, spatial data interpolation algorithm based on SVR optimization by GA are more accurate, and it has less dependence on the data itself and higher prediction accuracy, which is a new interpolation method with good prospect. [ABSTRACT FROM AUTHOR]

Processing of data recorded by the MODIS sensors on board the Terra and Aqua satellites has provided AOT maps that in some cases show low correlations with ground-based data recorded by the AERONET. Application of SVR techniques to MODIS data is a promising, though yet poorly explored, method of enhancing the correlations between satellite data and ground measurements. The article explains how satellite data recorded over three years on central Europe are correlated in space and time with ground based data and then shows results of the application of the SVR technique which somewhat improves previously computed correlations. Hints about future work in testing different SVR variants and methodologies are inferred from the analysis of the results thus far obtained. [ABSTRACT FROM AUTHOR]

In theory, a gold futures possesses function of price discovery. However, futures including information must be disclosed by some effective way. This paper proposes a forecasting model which combines wavelet frame with Support vector regression (SVR). Wavelet frame is first used to decompose the series of gold futures price into sub-series with different scales, the SVR then uses the sub-series to build the forecasting model. Empirical research shows that the gold futures has the function of price discovery, and the two steps model is a good tool for making the price information clear and forecasting spot price. further research can try different basis function or other methods of disclosing information. [ABSTRACT FROM AUTHOR]

Color information is of great importance for the tongue inspection of computer-aided tongue diagnosis system. However, the RGB signals generated by different imaging device varied greatly due to dissimilar lighting conditions and usage of different kinds of digital cameras. This is a key problem for the tongue inspection and diagnosis. A common solution is to correct the tongue images to standard color space by the aid of colorchecker. In this paper, three general color correction techniques: polynomial regression, artificial neural network and support vector regression (SVR) are applied to the color correction of tongue image and compared for their performance of accuracy and time complexity. The experimental results of colorchecker correction show that when properly optimized, SVR performs the best among these three algorithms, with a training error of 0 and a test error of 0.68 to 3.03. The polynomial regression algorithm performs a little worse, but it is more robust to the fluctuations of the environmental illuminant and much faster than SVR to train the parameters. The ANN performs worst, and it is also time-consuming to train. Performance comparison to correct real tongue images shows that polynomial regression is better than SVR to achieve a close correction result to human perception. Finally, this paper is concluded that for tongue inspection in a computer-aided tongue diagnosis system, polynomial regression is suitable for online system correction to aid tongue diagnosis, while SVR technique offer a better alternative for the offline and automated tongue diagnosis. [ABSTRACT FROM AUTHOR]

Accurate grid resources prediction is crucial for a grid scheduler. In this study, support vector regression (SVR), which is an effective regression algorithm, is applied to grid resources prediction. In order to build an effective SVR model, SVR΄s parameters must be selected carefully. Therefore, we develop an ant colony optimization-based SVR (ACO-SVR) model that can automatically determine the optimal parameters of SVR with higher predictive accuracy and generalization ability simultaneously. The proposed model was tested with grid resources benchmark data set. Experimental results demonstrated that ACO-SVR worked better than SVR optimized by trial-and-error procedure (T-SVR) and back-propagation neural network (BPNN). [ABSTRACT FROM AUTHOR]

In order to manage the grid resources more effectively, the prediction information of grid resources is necessary in the grid system. This study developed a new model, ISGA-SVR, for parameters optimization in support vector regression (SVR), which is then applied to grid resources prediction. In order to build an effective SVR model, SVR΄s parameters must be selected carefully. Therefore, we develop genetic algorithms with improved simulated binary crossover (ISBX) that can automatically determine the optimal parameters of SVR with higher predictive accuracy. In ISBX, we proposed a new method to deal with the bounded search space. This method can improve the search ability of original simulated binary crossover (SBX) .The proposed model was tested with grid resources benchmark data set. Experimental results demonstrated that ISGA-SVR worked better than SVR optimized by genetic algorithm with SBX(SGA-SVR) and back-propagation neural network (BPNN). [ABSTRACT FROM AUTHOR]

Gene time series microarray experiments have been widely used to unravel the genetic machinery of biological process. However, most temporal gene expression data often contain noise, missing data points, and non-uniformly sampled time points, which will make the traditional analyzing methods to be unapplicable. One main approach to solve this problem is to reconstruct each gene expression profile as a continuous function of time. Then the continuous representation enables us to overcome problems related to sampling rate differences and missing values. In this paper, we introduce a novel reconstruction approach based on the support vector regression method. The proposed approach utilizes a framelet based kernel, which has the ability to approximate functions with multiscale structure and can reduce the influence of noise in data. To compensate the inadequate information from noisy and short gene expression data, we use its correlated genes as the test set to choose the optimal parameters. We show that this treatment can help to avoid over-fitting. Experimental results demonstrate that our method can improve the reconstruction accuracy. [ABSTRACT FROM AUTHOR]

In this paper, a novel PSO–SVR model that hybridized the constrict particle swarm optimization (PSO) and support vector regression (SVR) is proposed for grade estimation. This hybrid PSO–SVR model searches for SVR΄s optimal parameters using constrict particle swarm optimization algorithms, and then adopts the optimal parameters to construct the SVR models. The hybrid PSO–SVR grade estimation method has been tested on a number of real ore deposits. The result shows that method has advantages of rapid training, generality and accuracy grade estimation approach. It can provide with a very fast and robust alternative to the existing time-consuming methodologies for ore grade estimation. [ABSTRACT FROM AUTHOR]

In this study, a new monitoring system for carbon dioxide exchange is presented. The mission of the intelligent environment presented in this work, is to globally monitor the interaction between the ocean΄s surface and the atmosphere, facilitating the work of oceanographers. This paper proposes a hybrid intelligent system integrates case-based reasoning (CBR) and support vector regression (SVR) characterised for their efficiency for data processing and knowledge extraction. Results have demonstrated that the system accurately predicts the evolution of the carbon dioxide exchange. [ABSTRACT FROM AUTHOR]

This paper presents support vector regression with hybrid genetic algorithms and tabu search (GA/TS) algorithms (SVRGA/TS) models for the prediction of radio-wave path-loss in suburban environment. The support vector regression (SVR) model is a novel forecasting approach and has been successfully used to solve time series problems. However, the application of SVR model in a radio-wave path-loss forecasting has not been widely investigated. This study aims at developing a SVRGA/TS model to forecast radio-wave path-loss data. Furthermore, the genetic algorithm and tabu search techniques have be applied to select important parameters for SVR model. In this study, four forecasting models, Egli, Walfisch and Bertoni (W&B), generalized regression neural networks (GRNN) and SVRGA/TS models are employed for forecasting the same data sets. Empirical results indicate that the SVRGA/TS outperforms other models in terms of forecasting accuracy. Thus, the SVRGA/TS model is an effective method for radio-wave path-loss forecasting in suburban environment. [ABSTRACT FROM AUTHOR]

Accurate forecasting of rainfall has been one of the most important issues in hydrological research. In this paper, a novel neural network technique, support vector regression (SVR), to monthly rainfall forecasting. The aim of this study is to examine the feasibility of SVR in monthly rainfall forecasting by comparing it with back–propagation neural networks (BPNN) and the autoregressive integrated moving average (ARIMA) model. This study proposes a novel approach, known as particle swarm optimization (PSO) algorithms, which searches for SVR΄s optimal parameters, and then adopts the optimal parameters to construct the SVR models. The monthly rainfall in Guangxi of China during 1985–2001 were employed as the data set. The experimental results demonstrate that SVR outperforms the BPNN and ARIMA models based on the normalized mean square error (NMSE) and mean absolute percentage error (MAPE). [ABSTRACT FROM AUTHOR]

The objective of this paper is to investigate the use of Support Vector Regression (SVR) for Web development effort estimation when using a cross-company data set. Four kernels of SVR were used, linear, polynomial, Gaussian and sigmoid and two preprocessing strategies of the variables were applied, namely normalization and logarithmic. The hold-out validation process was carried out for all the eight configurations using a training set and a validation set from the Tukutuku data set. Our results suggest that the predictions obtained with linear kernel applying a logarithmic transformation of variables (LinLog) are significantly better than those obtained with the other configurations. In addition, SVR has been compared with the traditional estimation techniques, such as Manual StepWise Regression, Case-Based Reasoning, and Bayesian Networks. Our results suggest that SVR with LinLog configuration can provide significantly superior prediction accuracy than other techniques. [ABSTRACT FROM AUTHOR]

Researchers are taking interest in the computational prediction models to efficiently predict the structure of perovskites. we are using Support Vector Regression, Artificial Neural Network, Multiple Linear Regression and SPuDS program based approaches in predicting the lattice constants (LC) of double perovskites of A2BB΄O6-type. These prediction models correlate the LC to atomic parameters i.e., size of ionic radii, electro-negativity, and oxidation state. These models are developed using training data. Their performance is then estimated for validation data. To investigate the generalization capability, 48 new perovskites are also collected from recent literature. Analysis shows that SVR based proposed models are more accurate and generalized, reducing the prediction error effectively. [ABSTRACT FROM AUTHOR]

Support Vector Regression (SVR) has been very successful in pattern recognition, text categorization and function approximation. In real application systems, data domain often suffers from noise and outliers. When there is noise and/or outliers existing in sampling data, the SVR may try to fit those improper data and obtained systems may have the phenomenon of overfitting. In addition, the memory space for storing the kernel matrix of SVR will be increment with O (N2), where N is the number of training data. In this paper, a robust support vector regression is proposed for nonlinear function approximation problems with noise and outliers. [ABSTRACT FROM AUTHOR]

Resource planning in automotive industry is a very complex process which involves the management of material and human needs and supplies. This paper deals with the production of plastic injection moulds used to make car components in the automotive industry. An efficient planning requires, among other, an accurate estimation of the task execution times in the mould production process. If the relation between task times and mould parts geometry is known, the moulds can be designed with a geometry that allows the shortest production time. We applied two popular regression approaches, Support Vector Regression and Radial Basis Function, to this problem, achieving accurate results which make feasible an automatic estimation of the task execution time. [ABSTRACT FROM AUTHOR]

In this paper we target the automatic prediction of two personality traits, Extraversion and Locus of Control, in a meeting scenario using visual and acoustic features. We designed our task as a regression one where the goal is to predict the personality traits΄ scores obtained by the meeting participants. Support Vector Regression is applied to thin slices of behavior, in the form of 1-minute sequences. [ABSTRACT FROM AUTHOR]

The paper presents the Multiple Kernel Learning (MKL) approach as a modelling and data exploratory tool and applies it to the problem of wind speed mapping. Support Vector Regression (SVR) is used to predict spatial variations of the mean wind speed from terrain features (slopes, terrain curvature, directional derivatives) generated at different spatial scales. Multiple Kernel Learning is applied to learn kernels for individual features and thematic feature subsets, both in the context of feature selection and optimal parameters determination. An empirical study on real-life data confirms the usefulness of MKL as a tool that enhances the interpretability of data-driven models. [ABSTRACT FROM AUTHOR]

Development of performant state estimators for industrial processes like copper extraction is a hard and relevant task because of the difficulties to directly measure those variables on-line. In this paper a comparison between a dynamic NARX-type neural network model and a support vector machine (SVM) model with external recurrences for estimating the filling level of the mill for a semiautogenous ore grinding process is performed. The results show the advantages of SVM modeling, especially concerning Model Predictive Output estimations of the state variable (MSE < 1.0), which would favor its application to industrial scale processes. [ABSTRACT FROM AUTHOR]

We explore the pricing performance of Support Vector Regression for pricing S&P 500 index call options. Support Vector Regression is a novel nonparametric methodology that has been developed in the context of statistical learning theory, and until now it has not been widely used in financial econometric applications. This new method is compared with the Black and Scholes (1973) option pricing model, using standard implied parameters and parameters derived via the Deterministic Volatility Functions approach. The empirical analysis has shown promising results for the Support Vector Regression models. [ABSTRACT FROM AUTHOR]

This paper presents a method to estimate the position of object using contextual information. Although convention methods used only shape contextual information, color contextual information is also effective to describe scenes. Thus we use both shape and color contextual information. To estimate the object position from only contextual information, the Support Vector Regression is used. We choose the Pyramid Match Kernel which measures the similarity between histograms because our contextual information is described as histogram. When one kernel is applied to a feature vector which consists of color and shape, the similarity of each feature is not used effectively. Thus, kernels are applied to color and shape independently, and the weighted sum of the outputs of both kernels is used. We confirm that the proposed method outperforms conventional methods. [ABSTRACT FROM AUTHOR]

An Immune Particle Swarm Optimization (IPSO) for parameters optimization of Support Vector Regression (SVR) is proposed in this article. After introduced clonal copy and mutation process of Immune Algorithm (IA), the particle of PSO is considered as antibodies. Therefore, evaluated the fitness of particles by the Cross Validation standard, the best individual mutated particle for each cloned group will be selected to compose the next generation to get better parameters εCδ of SVR. It can construct high accuracy and generalization performance regression model rapidly by optimizing the combination of three SVR parameters at the same time. Under the datasets generated from sincx function with additive noise and forest fires dataset, experimental results show that the new method can determine the parameters of SVR quickly and the gotten models have superior learning accuracy and generalization performance. [ABSTRACT FROM AUTHOR]

At present, MTV has become an important favorite pastime to people. Affective analysis which can extract the affective states contained in MTVs could be a potential and promising solution for efficient and intelligent MTV access. One of the most challenging and insufficiently covered problems of affective analysis is that affective understanding is personal and various among users. Consequently, it is meaningful to develop personalized affective modeling technique. Because user΄s feedbacks and descriptions about affective sates provide valuable and relatively reliable clues about user΄s personal affective understanding, it is supposed to be reasonable to conduct personalized affective modeling by analyzing the affective descriptions recorded in user profile. Utilizing the user profile, we propose a novel approach combining support vector regression and psychological affective model to achieve personalized affective analysis. The experimental results including both user study and comparisons between current approaches illustrate the effectiveness and advantages of our proposed method. [ABSTRACT FROM AUTHOR]

This paper presents an adaptive image watermarking scheme. Watermark bits are embedded adaptively into the non-separable wavelet domain based on the Human Visual System (HVS) model trained by Support Vector Regression (SVR). Unlike conventional separable wavelet filter banks that limit the ability in capturing directional information, non-separable wavelet filter banks contain the basis elements oriented at a variety of directions and different filter banks are able to capture different detail information. After removing the high frequency components, the low frequency subband used for watermark embedding is more robust against noise and other distortions. In addition, owing to the good generalization ability of the support vector machine, watermark embedding strength can be adjusted according to the HVS value. The superiority of non-separable wavelet transform (DNWT) in capturing image features combined with the good generalization ability of support vector regression provide us with a promising way to design a more robust watermarking algorithm featuring a better trade-off between the robustness and imperceptivity, the main duality of watermarking algorithms. Experimental results show that the DNWT watermarking scheme is robust to noising, JPEG compression, and cropping. In particular, it is more resistant to JPEG compression and noise than the discrete separable wavelet transform based scheme. [ABSTRACT FROM AUTHOR]

Decomposition method has been widely used to efficiently solve the large size quadratic programming (QP) problems arising in support vector regression (SVR). In a decomposition method, a large QP problem is decomposed into a series of smaller QP subproblems, which can be solved much faster than the original one. In this paper, we analyze the global convergence of decomposition methods for SVR. We will show the decomposition methods for the convex programming problem formulated by Flake and Lawrence always stop within a finite number of iterations. [ABSTRACT FROM AUTHOR]

Risk level in the exchange rate market is dynamically evolving with complicated structures. To further refine the analysis process and achieve more accurate measurement, this paper proposes a novel kernel based nonlinear ensembled multi scale Value at Risk methodology for evaluating the risk level in the exchange rate market. In the proposed algorithm, wavelet analysis is introduced to analyze the multi scale heterogeneous risk structures across different time scales. The Principle Component Analysis is used to extract principle components from the redundant forecast matrixes. Then the support vector regression technique is integrated into the modeling process to nonlinearly ensemble forecast matrixes and produce more stable and accurate results. Taking Euro market as a typical test case, empirical studies employing the proposed algorithm shows the superior performance than benchmark ARMA-GARCH and realized volatility based approaches. [ABSTRACT FROM AUTHOR]

A new algorithm for modeling regression curve is put forward in the paper, it combines the B-spline network with improved support vector regression. Our experimental results on simulated data demonstrate that it is feasible and effective. [ABSTRACT FROM AUTHOR]

An appropriate parameters selection can significantly affect the accuracy of support vector regression (SVR) model. In this paper, a new evolutionary approach based on Differential Evolution (DE-SVR) is developed to train the SVR model. The approach evolves automatically the optimal model parameters by the differential mutation operations. Experimental results on several real-world datasets demonstrate that, comparing with the GA-based SVR and the Grid search methods, the DE-SVR can search the optimal parameters much more rapidly with less training time to build the SVR model, and has the comparable prediction accuracy as Grid search, even better than GA-based SVR. Therefore, the new evolutionary DE-SVM approach is an efficient method for automatic parameter determination of SVR problem. [ABSTRACT FROM AUTHOR]

In this paper, a new method to solve the classified problems by using Support Vector Regression is introduced. Proposed method is called as SVR-C for short. In the method, through reconstructing the training set, each class through reconstructing the training set, each class value corresponding to a new training set, then use the SVR algorithm to train it and get a constructed model. And then, to a new instance, use the constructed model to train it and approximate the target class to the maximization of output value. Compared with M5P-C, SMO, J48, the effectiveness of our approach is tested on 16 publicly available datasets downloaded from the UCI. Comprehensive experiments are performed, and the results show that the SVR-C outperforms M5P-C and J48, and takes on comparative performance to SMO but has low standard-deviation. Moreover, our approach performs well on multi-class problems. [ABSTRACT FROM AUTHOR]

In this paper, we proposed a novel non-rigid 2D-3D registration framework, which is based on Support Vector Regression (SVR) to compensate the disadvantages of generating large amounts of Digitally Rendered Radiographs (DRRs) in the stage of intra-operation for radiotherapy. It is successfully used to estimate similarity metric distribution from prior sparse target metric values against different featured transforming parameters of non-rigid registration. Through applying the appropriate selected features and kernel of SVR solution to our registration framework, experiments provide a precise registration result efficiently in order to assist radiologists locating the accurate positions of radiation surgery. Meanwhile, a medical diagnosis database is also built up to reduce the therapy cost and accelerate the procedure of radiotherapy in the case of future scheduling of multiple treatments. [ABSTRACT FROM AUTHOR]