Performance Of Support Vector Regression Research Articles

Speech features for telemonitoring of Parkinson's disease symptoms.

The aim of this paper is tracking Parkinson's disease (PD) progression based on its symptoms on vocal system using Unified Parkinsons Disease Rating Scale (UPDRS). We utilize a standard speech signal feature set, which contains 6373 static features as functionals of low-level descriptor (LLD) contours, and select the most informative ones using the maximal relevance and minimal redundancy based on correlations (mRMRC) criteria. Then, we evaluate performance of Gaussian mixture regression (GMR) and support vector regression (SVR) on estimating the third subscale of UPDRS, i.e., UPDRS: motor subscale (UPDRS-III). Among the most informative features, a list of features are selected after redundancy reduction. The selected features depict that LLDs providing information about spectrum flatness, spectral distribution of energy, and hoarseness of voice are the most important ones for estimating UPDRS-III. Moreover, the most informative statistical functions are related to range, maximum, minimum and standard deviation of LLDs, which is an evidence of the muscle weakness due to the PD. Furthermore, GMR outperforms SVR on compact feature sets while the performance of SVR improves by increasing number of features.

A Comparative Study on Machine Learning Algorithms for Smart Manufacturing: Tool Wear Prediction Using Random Forests

Manufacturers have faced an increasing need for the development of predictive models that predict mechanical failures and the remaining useful life (RUL) of manufacturing systems or components. Classical model-based or physics-based prognostics often require an in-depth physical understanding of the system of interest to develop closed-form mathematical models. However, prior knowledge of system behavior is not always available, especially for complex manufacturing systems and processes. To complement model-based prognostics, data-driven methods have been increasingly applied to machinery prognostics and maintenance management, transforming legacy manufacturing systems into smart manufacturing systems with artificial intelligence. While previous research has demonstrated the effectiveness of data-driven methods, most of these prognostic methods are based on classical machine learning techniques, such as artificial neural networks (ANNs) and support vector regression (SVR). With the rapid advancement in artificial intelligence, various machine learning algorithms have been developed and widely applied in many engineering fields. The objective of this research is to introduce a random forests (RFs)-based prognostic method for tool wear prediction as well as compare the performance of RFs with feed-forward back propagation (FFBP) ANNs and SVR. Specifically, the performance of FFBP ANNs, SVR, and RFs are compared using an experimental data collected from 315 milling tests. Experimental results have shown that RFs can generate more accurate predictions than FFBP ANNs with a single hidden layer and SVR.

Torque based defect detection and weld quality modelling in friction stir welding process

Efforts have been made towards the monitoring of friction stir welding (FSW) process using real-time torque signals in this research work. Signals were analyzed using discrete wavelet transform and statistical features namely dispersion, asymmetry and excess are computed. The computed features are further processed to develop effective methodology for internal defect identification in FSW process. A new indicator has been proposed combining the computed statistical features. The proposed indicator shows appreciable deviations for defective and defect free welds. Apart from defect detection using the computed signal features, they are also presented as inputs to a support vector machine learning based modelling tool for the prediction of ultimate tensile strength of the welded joints. The prediction accuracy of the model with computed signal features are found to be more than the model developed with process parameters. The comparison of the developed support vector regression (SVR) model with artificial neural network (ANN) model and general regression model yields that prediction performance of SVR is superior to ANN and general regression model. The proposed work can be modified for its successful use in real-time modelling of friction stir welding process.

Improving SVR and ANFIS performance using wavelet transform and PCA algorithm for modeling and predicting biochemical oxygen demand (BOD)

In recent years, the use of the artificial intelligence as an acceptable method in various issues, particularly in hydrology, have sharply risen. In this study, Support Vector Regression (SVR) and Adaptive Neural Fuzzy Inference System (ANFIS) models were used for predicting Biochemical Oxygen Demand (BOD) in Karun River in the west of Iran. In order to analyze hybrid models, wavelet transform was used as well. After decomposing parameters by wavelet transform, Principal Component Analysis (PCA) was used to recognize important components. Then, monthly time series of BOD index was used in Karun River in Mollasani station and also, covariates like Dissolved Oxygen (DO), monthly temperature, and river flow were used from 2002 to 2014. The results indicated that the SVR model with RMSE=0.0338mg/l and R2=0.843 has better performance than the ANFIS model with R2=0.828. Also, applying the wavelet transform on input data of the SVR model improved the results to R2=0.937 and RMSE=0.0210mg/l. Therefore, combining the SVR with the wavelet transform (WSVR) was a good idea to improve the prediction of the BOD value in Karun River. Finally, the combination was recognized as a suitable method and the BOD was predicted in six months.

Comparison of partial least squares and support vector regressions for predicting leaf area index on a tropical grassland using hyperspectral data

Leaf area index (LAI) is a key biophysical parameter commonly used to determine vegetation status, productivity, and health in tropical grasslands. Accurate LAI estimates are useful in supporting sustainable rangeland management by providing information related to grassland condition and associated goods and services. The performance of support vector regression (SVR) was compared to partial least square regression (PLSR) on selected optimal hyperspectral bands to detect LAI in heterogeneous grassland. Results show that PLSR performed better than SVR at the beginning and end of summer. At the peak of the growing season (mid-summer), during reflectance saturation, SVR models yielded higher accuracies (R2=0.902 and RMSE=0.371 m2 m−2) than PLSR models (R2=0.886 and RMSE=0.379 m2 m−2). For the combined dataset (all of summer), SVR models were slightly more accurate (R2=0.74 and RMSE=0.578 m2 m−2) than PLSR models (R2=0.732 and RMSE=0.58 m2 m−2). Variable importance on the projection scores show that most of the bands were located in the near-infrared and shortwave regions of the electromagnetic spectrum, thus providing a basis to investigate the potential of sensors on aerial and satellite platforms for large-scale grassland LAI prediction.

Time series forecasting for building energy consumption using weighted Support Vector Regression with differential evolution optimization technique

Electricity load forecasting is crucial for effective operation and management of buildings. Support Vector Regression (SVR) have been successfully used in solving nonlinear regression and time series problems related to building energy consumption forecasting. As the performance of SVR heavily depends on the selection of its parameters, differential evolution (DE) algorithm is employed in this study to solve this problem. The forecasting model is developed using weighted SVR models with nu-SVR and epsilon-SVR. The DE algorithm is again used to determine the weights corresponding to each model. A case of time series energy consumption data from an institutional building in Singapore is used to elucidate the performance of the proposed model. The proposed model can be used to forecast both, half-hourly and daily electricity consumption time series data for the same building. The results show that for half-hourly data, the model exhibits higher weight for nu-SVR, whereas for daily data, a higher weight for epsilon-SVR is observed. The mean absolute percentage error (MAPE) for daily energy consumption data is 5.843 and that for half-hourly energy consumption is 3.767 respectively. A detailed comparison with other evolutionary algorithms show that the proposed model yields higher accuracy for building energy consumption forecasting.

New analysis methods to push the boundaries of diagnostic techniques in the environmental sciences

In the last years, new and more sophisticated measurements have been at the basis of the major progress in various disciplines related to the environment, such as remote sensing and thermonuclear fusion. To maximize the effectiveness of the measurements, new data analysis techniques are required. First data processing tasks, such as filtering and fitting, are of primary importance, since they can have a strong influence on the rest of the analysis. Even if Support Vector Regression is a method devised and refined at the end of the 90s, a systematic comparison with more traditional non parametric regression methods has never been reported. In this paper, a series of systematic tests is described, which indicates how SVR is a very competitive method of non-parametric regression that can usefully complement and often outperform more consolidated approaches. The performance of Support Vector Regression as a method of filtering is investigated first, comparing it with the most popular alternative techniques. Then Support Vector Regression is applied to the problem of non-parametric regression to analyse Lidar surveys for the environments measurement of particulate matter due to wildfires. The proposed approach has given very positive results and provides new perspectives to the interpretation of the data.

Prediction of Henry Constants and Adsorption Mechanism of Volatile Organic Compounds on Multi-Walled Carbon Nanotubes by Using Support Vector Regression**Supported by the Innovative Talent Funds for Project 985 under Grant No WLYJSBJRCTD201102, the Fundamental Research Funds for the Central Universities under Grant No CQDXWL-2013-014, the Natural Science Foundation of Chongqing under Grant No CSTC2006BB5240, and the Program for New Century

Support vector regression (SVR) combined with particle swarm optimization for its parameter optimization is employed to establish a model for predicting the Henry constants of multi-walled carbon nanotubes (MWNTs) for adsorption of volatile organic compounds (VOCs). The prediction performance of SVR is compared with those of the model of theoretical linear salvation energy relationship (TLSER). By using leave-one-out cross validation of SVR test Henry constants for adsorption of 35 VOCs on MWNTs, the root mean square error is 0.080, the mean absolute percentage error is only 1.19%, and the correlation coefficient (R2) is as high as 0.997. Compared with the results of the TLSER model, it is shown that the estimated errors by SVR are all smaller than those achieved by TLSER. It reveals that the generalization ability of SVR is superior to that of the TLSER model. Meanwhile, multifactor analysis is adopted for investigation of the influences of each molecular structure descriptor on the Henry constants. According to the TLSER model, the adsorption mechanism of adsorption of carbon nanotubes of VOCs is mainly a result of van der Waals and interactions of hydrogen bonds. These can provide the theoretical support for the application of carbon nanotube adsorption of VOCs and can make up for the lack of experimental data.

Towards fast energy performance evaluation: A pilot study for office buildings

Given the growing concern about building energy efficiency and the difficulty in applying complex simulation tools during retrofit practices, the need for easily and quickly estimating the building energy performance becomes pressing. As a pilot test, this study proposes a systematic method to develop a model, which can immediately assess the annual electricity consumption for office buildings with fan coil system in Shanghai. First, a base-case building model is established by EnergyPlus to create a pool of candidate inputs using orthogonal experiment design. Then, analysis of variance is used to identify a total of 10 key building design parameters, which are selected as the input variables in the support vector regression (SVR) model based on a well-structured database. The performance of SVR is optimized using genetic algorithm (GA) based on radial basis function kernel. Finally, two real office buildings in Shanghai with reliable measured data serve to evaluate the developed hybrid model. The resulting differences between the predicted and measured values are generally within 10%. It is expected that the developed database and model can be used to assess the likely energy savings/penalty related with certain parameter changes to some extent during the retrofit process for office buildings.

Performance Comparison Between Support Vector Regression and Artificial Neural Network for Prediction of Oil Palm Production

The largest region that produces oil palm in Indonesia has an important role in improving the welfare of society and economy. Oil palm has increased significantly in Riau Province in every period, to determine the production development for the next few years with the functions and benefits of oil palm carried prediction production results that were seen from time series data last 8 years (2005-2013). In its prediction implementation, it was done by comparing the performance of Support Vector Regression (SVR) method and Artificial Neural Network (ANN). From the experiment, SVR produced the best model compared with ANN. It is indicated by the correlation coefficient of 95% and 6% for MSE in the kernel Radial Basis Function (RBF), whereas ANN produced only 74% for R2 and 9% for MSE on the 8th experiment with hiden neuron 20 and learning rate 0,1. SVR model generates predictions for next 3 years which increased between 3% - 6% from actual data and RBF model predictions.

COMPARISON OF CLASSICAL REGRESSION METHODS WITH DATA MINING REGRESSION METHODS IN PREDICTION OF HEALTH EXPENDITURE

Aim: One of the main difficulties about analyzing health expenditures is, the distribution of health expenditure is not normal and extremely positively skewed. This brings about overfitting problem and causes a decrease in regression model performance for predicting health expenditures. It is possible to use data mining based regression methods to improve classical regression model performances and overcome overfitting problem. Regression Tress, Random Forest Regression and Support Vector Regression are some of these methods. In this study it is aimed to compare prediction performances of different regression methods about predicting per capita health expenditures of member of total 214 World Bank countries. Materials and Methods: Before the analysis the distribution of health expenditure per capita normalized with using logarithmic and Box-Cox transformations. Multiple Linear Regression, Regression Tree, Random Forest Regression and Support Vector Machine Regression methods was used for prediction and R2, RMSE and MAE values are used for the assessment of prediction performances. Performance results are compared according to cross validation values determined by using different number of “k” parameters. Findings: Study findings show that prediction performance of Support Vector Regression is relatively higher compared with other regression methods when health expenditure per capita transformed by using Box-Cox transformation and when “k” parameter increases in cross validation. Results: Study results show that Support Vector Regression prediction performance is higher than other regression methods. It is advisable for future studies to examine Support Vector Regression performances using grid search methods which are one of hyperparameter optimization techniques.

Metode Hibrida FCM dan PSO-SVR untuk Prediksi Data Arus Lalu Lintas

Abstract — Traffic flow forecasting is one important part in Intelligent Transportation System. There are many methods had been developed for time series and traffic flow forecasting such as: Autoregressive Moving Average (ARIMA), Artificial Neural Network (ANN), and Support Vector Regression (SVR). SVR performance depend on kernel function and parameters of those kernel and data characteristic used in SVR as well. This research proposed hybrid method for traffic flow data clustering and forecasting. Fuzzy C-means is used in order to minimize the variance in whole dataset. Particle Swarm Optimization (PSO) is used in order to select the appropriate parameters for SVR. Experimental result shows the proposed method give MAPE below 4% in all test sites. Keywords—fuzzy c-means, particle swarm optimization, prediksi data lalu lintas, support vector regression, time-series.

SNOW DEPTH ESTIMATION USING TIME SERIES PASSIVE MICROWAVE IMAGERY VIA GENETICALLY SUPPORT VECTOR REGRESSION (CASE STUDY URMIA LAKE BASIN)

Abstract. Lake Urmia is one of the most important ecosystems of the country which is on the verge of elimination. Many factors contribute to this crisis among them is the precipitation, paly important roll. Precipitation has many forms one of them is in the form of snow. The snow on Sahand Mountain is one of the main and important sources of the Lake Urmia’s water. Snow Depth (SD) is vital parameters for estimating water balance for future year. In this regards, this study is focused on SD parameter using Special Sensor Microwave/Imager (SSM/I) instruments on board the Defence Meteorological Satellite Program (DMSP) F16. The usual statistical methods for retrieving SD include linear and non-linear ones. These methods used least square procedure to estimate SD model. Recently, kernel base methods widely used for modelling statistical problem. From these methods, the support vector regression (SVR) is achieved the high performance for modelling the statistical problem. Examination of the obtained data shows the existence of outlier in them. For omitting these outliers, wavelet denoising method is applied. After the omission of the outliers it is needed to select the optimum bands and parameters for SVR. To overcome these issues, feature selection methods have shown a direct effect on improving the regression performance. We used genetic algorithm (GA) for selecting suitable features of the SSMI bands in order to estimate SD model. The results for the training and testing data in Sahand mountain is [R²_TEST=0.9049 and RMSE= 6.9654] that show the high SVR performance.

Robust method based on optimized support vector regression for modeling of asphaltene precipitation

Abstract Precipitation of asphaltene largely affects the production rate of crude oil owing to the clogging of transportation pipeline as well as formation damage. Therefore, it is imperative to search for a robust method for the modeling of asphaltene precipitation. This paper aims to propose a new mathematical model for computing asphaltene precipitation as a function of titration data including dilution ratio, temperature, and molecular weight of solvent. This model is constructed based on integrating the support vector regression with the imperialist competitive algorithm. Optimization increases the performance of support vector regression by virtue of determining the optimal values of their parameters. The constructed model is applied to experimental data and its performance is compared with scaling equation which is traditionally employed for modeling of asphaltene precipitation. The results of this study show that the integration of imperialist competitive algorithm and support vector regression has a better performance than the scaling equation. This paper offers that the support vector regression optimized by virtue of the imperialist competitive algorithm is a reliable model for modeling asphaltene precipitation.

Support vector regression for near‐field multifocused antenna arrays considering mutual coupling

SummarySupport vector regression is proposed to obtain the coupling matrix that accounts for the real behavior of an antenna array in the near‐field multifocusing framework. The obtained model can then be easily included in this synthesis method in order to obtain the feeding weights which satisfy the radiation pattern requirements under real conditions. An example of the performance of support vector regression using information from method of moments is also provided to demonstrate the performance and accuracy of the technique. Copyright © 2015 John Wiley & Sons, Ltd.

Protein-Protein Interaction Prediction using PCA and SVR-PHCS

Protein-Protein Interactions (PPIs) play a key role in many biological systems. Thus, identifying PPIs is critical for understanding cellular processes. Many experimental techniques were applied to predict PPIs. The data extracted using these techniques are incomplete and noisy. In this regard, a number of computational methods include machine learning classification techniques have been developed to reduce the noise data and predict new PPIs. Since, using regression methods to solve classification problems has good results in other applications. Therefore, in this paper, a regression view is applied to the PPI prediction classification problem, so a new approach is proposed using Principal Component Analysis (PCA) and Support Vector Regression (SVR) which has been improved by a new Parallel Hierarchical Cube Search (PHCS) method. Firstly, PCA algorithm is implemented to select an optimal subset of features which leads to reduce processing time and to lessen the effect of noise. Then, the PPIs would be predicted, by using SVR. To get a better performance of SVR, a new PHCS method has been applied to select the appropriate values of SVR parameters. The obtained classification accuracy of the proposed method is 74.505% on KUPS (The University of Kansas Proteomics Service) dataset which outperforms the other methods.

A comparative assessment of support vector regression, artificial neural networks, and random forests for predicting and mapping soil organic carbon stocks across an Afromontane landscape

Soil organic carbon (SOC) is a key indicator of ecosystem health, with a great potential to affect climate change. This study aimed to develop, evaluate, and compare the performance of support vector regression (SVR), artificial neural network (ANN), and random forest (RF) models in predicting and mapping SOC stocks in the Eastern Mau Forest Reserve, Kenya. Auxiliary data, including soil sampling, climatic, topographic, and remotely-sensed data were used for model calibration. The calibrated models were applied to create prediction maps of SOC stocks that were validated using independent testing data. The results showed that the models overestimated SOC stocks. Random forest model with a mean error (ME) of −6.5MgCha−1 had the highest tendency for overestimation, while SVR model with an ME of −4.4MgCha−1 had the lowest tendency. Support vector regression model also had the lowest root mean squared error (RMSE) and the highest R2 values (14.9MgCha−1 and 0.6, respectively); hence, it was the best method to predict SOC stocks. Artificial neural network predictions followed closely with RMSE, ME, and R2 values of 15.5, −4.7, and 0.6, respectively. The three prediction maps broadly depicted similar spatial patterns of SOC stocks, with an increasing gradient of SOC stocks from east to west. The highest stocks were on the forest-dominated western and north-western parts, while the lowest stocks were on the cropland-dominated eastern part. The most important variable for explaining the observed spatial patterns of SOC stocks was total nitrogen concentration. Based on the close performance of SVR and ANN models, we proposed that both models should be calibrated, and then the best result applied for spatial prediction of target soil properties in other contexts.

Predicting thermal–hydraulic performances in compact heat exchangers by support vector regression

An alternative model using support vector regression (SVR) based on dynamically optimized search technique with k-fold cross-validation, was proposed to predict the thermal–hydraulic performance of compact heat exchangers (CHEs). 48 experimental data points from the author’s own study were used in the present work. The performance of SVR with different regularization parameter γ and kernel parameter σ2 had been investigated and the optimal values were obtained. According to predicted accuracy of indicating generalization capability, the model performance was compared and evaluated with the artificial neural network (ANN) model. As a result, it is found that, the SVR provides better prediction performances with the mean squared errors (MSE) of 2.645×10−4 for testing j factor and 1.231×10−3 for testing f factor, respectively. Also the computational time of SVR model was shorter than that of the ANN model. Moreover, the versatility of the configured SVR model was demonstrated by presenting the effects of some input variables on the output variables. The result indicated that SVR can offer an alternative and powerful approach to predict the thermal characteristics of new type fins in CHEs under various operating conditions.

A Comparative Study of Empirical Mode Decomposition-Based Short-Term Wind Speed Forecasting Methods

Wind speed forecasting is challenging due to its intermittent nature. The wind speed time series (TS) has nonlinear and nonstationary characteristics and not normally distributed, which make it difficult to be predicted by statistical or computational intelligent methods. Empirical mode decomposition (EMD) and its improved versions are powerful tools to decompose a complex TS into a collection of simpler ones. The improved versions discussed in this paper include ensemble EMD (EEMD), complementary EEMD (CEEMD), and complete EEMD with adaptive noise (CEEMDAN). The EMD and its improved versions are hybridized with two computational intelligence-based predictors: support vector regression (SVR) and artificial neural network (ANN). The EMD-based hybrid forecasting methods are evaluated with 12 wind speed TS. The performances of the hybrid methods are compared and discussed. It shows that EMD and its improved versions enhance the performance of SVR significantly but marginally on ANN, and among the EMD-based hybrid methods, the proposed CEEMDAN-SVR is the best method. Possible future works are also recommended for wind speed forecasting.

Application of support vector regression to CPUE analysis for southern bluefin tuna Thunnus maccoyii, and its comparison with conventional methods

This paper describes the catch per unit effort (CPUE) standardization using three models for data mining (support vector regression, neural network and tree regression model) and two conventional statistical methods (analysis of variance and generalized linear model) using the actual fishery data for southern bluefin tuna Thunnus maccoyii. Statistical performances of these five models were compared based on mean square error, mean absolute error and three correlation coefficients, which are measured by the difference between the observed and the corresponding predicted values. As a result, the performance of support vector regression is equivalent to (or better than) that of neural networks, and these two models are superior to the tree regression model, analysis of variance, and generalized linear model based on CPUE analyses of actual fishery data for southern bluefin tuna. We suggest a simple method for factorial analysis to extract the CPUE year trend based on the predicted values obtained from these data mining models. This method is expected to contribute markedly to reduce the difficulty of estimating the CPUE year trends by these models for data mining and should be applied to CPUE analyses because of its ease of use, general versatility and high performance .