Support Vector Regression Approach Research Articles

Approaches to denoise the diffuse optical signals for tissue blood flow measurement

Various diseases are relevant to the abnormal blood flow in tissue. Diffuse correlation spectroscopy (DCS) is an emerging technology to extract the blood flow index (BFI) from light electric field temporal autocorrelation data. To account for tissue heterogeneity and irregular geometry, we developed an innovative DCS algorithm (i.e., the Nth order linear algorithm, or simply the NL algorithm) previously, in which the DCS signals are fully utilized through iterative linear regressions. Under the framework of NL algorithm, the BFI to be extracted is significantly influenced by the linear regression approach adopted. In this study, three approaches were proposed and evaluated for performing the iterative linear regressions, in order to understand what are the appropriate regression methods for BFI estimation. The three methods are least-squared minimization (L2 norm), least-absolute minimization (L1 norm) and support vector regression (SVR), where L2 norm is a conventional approach to perform linear regression. L1 norm and SVR are the approaches newly introduced here to process the DCS data. Computer simulations and the autocorrelation data collected from liquid phantom and human tissues are utilized to evaluate the three approaches. The results show that the best performance is achieved by the SVR approach in extracting the BFI values, with an error rate of 2.23% at 3.0 cm source-detector separation. The L1 norm method gives a medium error of 2.81%. In contrast, the L2 norm method leads to the largest error (3.93%) in extracting the BFI values. The outcomes derived from this study will be very helpful for the tissue blood flow measurements, which is critical for translating the DCS technology to the clinic.

The evaluation of the effect of nappe breakers on the discharge capacity of trapezoidal labyrinth weirs by ELM and SVR approaches

The labyrinth weir is one type of overflow design used to direct and transfer water in open channels and to provide both routine flow and flood passage over dam spillways. Labyrinth weirs are primarily used at sites where the available spillway width is limited. Due to the increase in crest length, a labyrinth weir provides an increase in discharge capacity relative to conventional weir structures. It is important that the discharge coefficient be accurately represented to ensure appropriate and economical design. The discharge coefficient of trapezoidal labyrinth weirs (TLW) is estimated by using extreme learning machines (ELM) and support vector regression (SVR) techniques in this study. Additional discharge coefficient prediction models have been developed for applications that include the use of nappe breakers (NB). These are frequently included in the design as a mechanism to reduce the impact of vibrations and oscillations on these weirs. A total of 1128 test runs for discharge coefficient measurements of TLW with/without NB were performed in the present study. The statistical criteria used for the evaluation of the performance of models are Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Root Relative Squared Error (RRSE), Mean Absolute Percentage Error (MAPE) and Determination Coefficient (R2). Results of this investigation suggest that the models using Extreme Learning Machines (ELM) and Support Vector Regression (SVR) methods are successful in modeling the discharge coefficient of TLW with/without NB. The best correspondence between model and observation occurred using the ELM model; this resulted in an RMSE for the TLW with/without NB of 0.0188 and 0.0158, respectively.

Support Vector Regression Approach to Predict the Design Space for the Extraction Process of Pueraria lobata.

A support vector regression (SVR) method was introduced to improve the robustness and predictability of the design space in the implementation of quality by design (QbD), taking the extraction process of Pueraria lobata as a case study. In this paper, extraction time, number of extraction cycles, and liquid–solid ratio were identified as critical process parameters (CPPs), and the yield of puerarin, total isoflavonoids, and extracta sicca were the critical quality attributes (CQAs). Models between CQAs and CPPs were constructed using both a conventional quadratic polynomial model (QPM) and the SVR algorithm. The results of the two models indicated that the SVR model had better performance, with a higher R2 and lower root-mean-square error (RMSE) and mean absolute deviation (MAD) than those of the QPM. Furthermore, the design space was predicted using a grid search technique. The operational range was extraction time, 24–51 min; number of extraction cycles, 3; and liquid–solid ratio, 14–18 mL/g. This study is the first reported work optimizing the design space of the extraction process of P. lobata based on an SVR model. SVR modeling, with its better prediction accuracy and generalization ability, could be a reliable tool for predicting the design space and shows great potential for the quality control of QbD.

Pipeline leakage localization based on distributed FBG hoop strain measurements and support vector machine

Across the globe, pipelines help to carry all kinds of fluids across vast distances. Our prior work in fiber Bragg grating (FBG) hoop strain sensors are among the most recently reported technologies aimed at accomplishing the goal of continuous pipeline monitoring. Multiple hoop strain signals can be extracted from distributed FBG hoop strain sensors set along the pipeline to reflect leakage process. In this paper, we demonstrate the use of multiple, distributed FBG hoop strain sensors in cooperation with a support vector regression (SVR) to localize a leakage point along a model pipeline. Series of terminal hoop strain variations are extracted as the input variables to achieve multi regression analysis as to localize the leakage point. The parameters of different kernel functions are optimized through five-fold cross validation to obtain the highest leakage localization accuracy. The result shows that when taking radial basis kernel function (RBF) with optimized C and γ values, the localization mean square error (MSE) reaches as low as 0.043. The anti-noise capability of the SVR model is evaluated through superimposing Gaussian white noise of different levels. From the simulation study, the average localization error is still acceptable (≈500 m) even in 5% noise situation. The influence of hoop strain sensing points as input variables is also investigated. The system with more hoop strain sensing points shows more stable capability for different level noises. The results demonstrate feasibility and robustness of the SVR approach using multi-hoop strain measurements for pipeline leakage localization.

Interval Kernels for Combining Biometric Measurements from Multiple Prostate Samples per Patient in Prognostic Models with Transductive Semi-Supervised Support Vector Regression.

Prostate cancer is the most prevalent form of cancer and second most common form of cancer deaths among men in the United States. Physicians work with patients to make difficult treatment decisions. They are often aided by a multitude of prognostic models to assess risk and predict outcomes. These survival analysis models may analyze features characterizing biomolecular and histomorphic properties of the solid tumor sample. Commonly, multiple samples per patient are analyzed and features are combined in some way to generate a single result, such as by taking the median measurement. Recently, support vector regression approaches paired with a semi-supervised transduction framework have proven powerful in building such prognostic models. Separately, there has been work presenting the use of interval kernels to capture the range of measurements across multiple samples as an "interval" via the Hausdorff distance within a kernel based support vector approach. This paper presents the first results in exploring a combination of these two concepts. Namely, using an interval kernel based support vector approach within the aforementioned transduction framework to build prognostic models leveraging information from multiple tumor samples per patient. The results show that interval kernels yield more accurate prognostic models, and the semi-supervised transduction framework further improves performance. This suggests that this novel combination of unique recent advances can help build better prognostic models and improve the treatment of prostate cancer.

Support vector regression and neural networks analytical models for gas sensor based on molybdenum disulfide

In this study, MoS2 gas sensor based on field effect transistor has been proposed and the adsorption of NO2 molecules on the channel surface can lead to significant changes on its electronic and transport properties. The analytical models have been developed for the NO2 gas sensors by making an initial assumption that the gate voltage is directly proportional to the gas concentration. The performance of this sensor, is predicted and investigated by support vector regression (SVR) and artificial neural network (ANN) algorithms. The MoS2 gas sensor displays current changes upon exposure to very low concentrations of NO2. The comparison between analytical model, ANN and SVR with the empirical data shows the successful model construction. However, ANN outperforms the SVR approach and gives more accurate results.

Canopy Spectral Reflectance as a Predictor of Soil Water Potential in Rice

AbstractSoil water potential (SWP) is a key parameter for characterizing water stress. Typically, a tensiometer is used to measure SWP. However, the measurement range for commercially available tensiometers is limited to −90 kPa and a tensiometer can only provide estimate of SWP at a single location. In this study, a new approach was developed for estimating SWP from spectral reflectance data of a standing rice crop over the visible to shortwave‐infrared region (wavelength: 350–2,500 nm). Five water stress treatments corresponding to targeted SWP of −30, −50, −70, −120, and −140 kPa were examined by withholding irrigation during the vegetative growth stage of three rice varieties. Tensiometers and mechanistic water flow model were used for monitoring SWP. Spectral models for SWP were developed using partial‐least‐squares regression (PLSR), support vector regression (SVR), and coupled PLSR and feature selection (PLSRFS) approaches. Results showed that the SVR approach was the best model for estimating SWP from spectral reflectance data with the coefficient of determination values of 0.71 and 0.55 for the calibration and validation data sets, respectively. Observed root‐mean‐squared residuals for the predicted SWPs were in the range of −7 to −19 kPa. A new spectral water stress index was also developed using the reflectance values at 745 and 2,002 nm, which showed strong correlation with relative water contents and electrolyte leakage. This new approach is rapid and noninvasive and may be used for estimating SWP over large areas.

Mapping patterns of urban development in Ouagadougou, Burkina Faso, using machine learning regression modeling with bi-seasonal Landsat time series

Rapid urban population growth in Sub-Saharan Western Africa has important environmental, infrastructural and social impacts. Due to the low availability of reliable urbanization data, remote sensing techniques become increasingly popular for monitoring land use change processes in that region. This study aims to quantify land cover for the Ouagadougou metropolitan area between 2002 and 2013 using a Landsat-TM/ETM+/OLI time series. We use a support vector regression approach and synthetically mixed training data. Working with bi-seasonal image stacks, we account for spectral variability between dry and rainy season and incorporate a new class - seasonal vegetation - that describes surfaces that are soil and vegetation during parts of the year. We produce fraction images of urban surfaces, soil, permanent vegetation and seasonal vegetation for each time step. Statistical evaluation shows that a temporally generalized, bi-seasonal model over all time steps performs equally or better than yearly or mono-seasonal models and provides reliable cover fractions. Urban fractions can be used to visualize pixel-based spatial-temporal patterns of urban densification and expansion. A simple rule set based on a seasonal vegetation to soil ratio is appropriate to delineate areas of unplanned and planned settlements and, thus, contributes to monitoring urban development on a neighborhood scale.

Machine‐learning‐based system for multi‐sensor 3D localisation of stationary objects

Localisation of objects and people in indoor environments has been widely studied due to security issues and because of the benefits that a localisation system can provide. Indoor positioning systems (IPSs) based on more than one technology can improve localisation performance by leveraging the advantages of distinct technologies. This study proposes a multi-sensor IPS able to estimate the three-dimensional (3D) location of stationary objects using off-the-shelf equipment. By using radio-frequency identification (RFID) technology, machine-learning models based on support vector regression (SVR) and artificial neural networks (ANNs) are proposed. A k -means technique is also applied to improve accuracy. A computer vision (CV) subsystem detects visual markers in the scenario to enhance RFID localisation. To combine the RFID and CV subsystems, a fusion method based on the region of interest is proposed. We have implemented the authors' system and evaluated it using real experiments. On bi-dimensional scenarios, localisation error is between 9 and 29 cm in the range of 1 and 2.2 m. In a machine-learning approach comparison, ANN performed 31% better than SVR approach. Regarding 3D scenarios, localisation errors in dense environments are 80.7 and 73.7 cm for ANN and SVR models, respectively.

Prediction of Resilient Modulus from Post-compaction Moisture Content and Physical Properties Using Support Vector Regression

The present study assesses the use of support vector machine regression to predict the variation of resilient modulus with post-compaction moisture content of soils commonly encountered in Oklahoma, Pennsylvania and Wisconsin. Results show the prediction model using the support vector regression (SVR) approach is a function of degree of saturation, moisture content and plasticity index. The developed model is compared to current models in the literature. Results indicate the proposed SVR model gives more accurate values than current regression models. This model will better predict changes in the bearing capacity of pavements due to seasonal variations of moisture content.

Radio location of partial discharge sources: a support vector regression approach

Partial discharge (PD) can provide a useful forewarning of asset failure in electricity substations. A significant proportion of assets are susceptible to PD due to incipient weakness in their dielectrics. This study examines a low cost approach for uninterrupted monitoring of PD using a network of inexpensive radio sensors to sample the spatial patterns of PD received signal strength. Machine learning techniques are proposed for localisation of PD sources. Specifically, two models based on support vector machines are developed: support vector regression (SVR) and least-squares support vector regression (LSSVR). These models construct an explicit regression surface in a high-dimensional feature space for function estimation. Their performance is compared with that of artificial neural network (ANN) models. The results show that both SVR and LSSVR methods are superior to ANNs in accuracy. LSSVR approach is particularly recommended as practical alternative for PD source localisation due to its low complexity.

Using Support Vector Regression and Hyperspectral Imaging for the Prediction of Oenological Parameters on Different Vintages and Varieties of Wine Grape Berries

The performance of a support vector regression (SVR) model with a Gaussian radial basis kernel to predict anthocyanin concentration, pH index and sugar content in whole grape berries, using spectroscopic measurements obtained in reflectance mode, was evaluated. Each sample contained a small number of whole berries and the spectrum of each sample was collected during ripening using hyperspectral imaging in the range of 380–1028 nm. Touriga Franca (TF) variety samples were collected for the 2012–2015 vintages, and Touriga Nacional (TN) and Tinta Barroca (TB) variety samples were collected for the 2013 vintage. These TF vintages were independently used to train, validate and test the SVR methodology; different combinations of TF vintages were used to train and test each model to assess the performance differences under wider and more variable datasets; the varieties that were not employed in the model training and validation (TB and TN) were used to test the generalization ability of the SVR approach. Each case was tested using an external independent set (with data not included in the model training or validation steps). The best R2 results obtained with varieties and vintages not employed in the model’s training step were 0.89, 0.81 and 0.90, with RMSE values of 35.6 mg·L−1, 0.25 and 3.19 °Brix, for anthocyanin concentration, pH index and sugar content, respectively. The present results indicate a good overall performance for all cases, improving the state-of-the-art results for external test sets, and suggesting that a robust model, with a generalization capacity over different varieties and harvest years may be obtainable without further training, which makes this a very competitive approach when compared to the models from other authors, since it makes the problem significantly simpler and more cost-effective.

Fluid intelligence relates to the resting state amplitude of low-frequency fluctuation and functional connectivity: a multivariate pattern analysis.

The goal of this study was to investigate the relationship between fluid intelligence (gF) and the pattern of the functional characteristics in the resting state in adults using multivariate pattern analysis. Resting-state functional images from 100 participants in the Human Connectome Project data set were analyzed. The amplitude of low-frequency fluctuation (ALFF) was first calculated, and a support vector regression approach was used to identify the association with gF. To discover whether the connectivity of the gF-associated areas was also related to gF, we further checked the seed-based functional connectivity using the seeds from the ALFF. The ALFF showed that gF was correlated with the left anterior cingulate cortex, which is involved in high cognitive control processes. The functional connectivity showed that the connection between the right prefrontal cortex (Brodmann area 8) and the left anterior cingulate cortex could predict gF. The multivariate pattern analysis result indicated that the brain functional activity and functional integrity that we identified have the potential to become an objective biomarker for evaluating individual differences in gF.

Predicting Taxi-Out Time at Congested Airports with Optimization-Based Support Vector Regression Methods

Accurate prediction of taxi-out time is significant precondition for improving the operationality of the departure process at an airport, as well as reducing the long taxi-out time, congestion, and excessive emission of greenhouse gases. Unfortunately, several of the traditional methods of predicting taxi-out time perform unsatisfactorily at congested airports. This paper describes and tests three of those conventional methods which include Generalized Linear Model, Softmax Regression Model, and Artificial Neural Network method and two improved Support Vector Regression (SVR) approaches based on swarm intelligence algorithm optimization, which include Particle Swarm Optimization (PSO) and Firefly Algorithm. In order to improve the global searching ability of Firefly Algorithm, adaptive step factor and Lévy flight are implemented simultaneously when updating the location function. Six factors are analysed, of which delay is identified as one significant factor in congested airports. Through a series of specific dynamic analyses, a case study of Beijing International Airport (PEK) is tested with historical data. The performance measures show that the proposed two SVR approaches, especially the Improved Firefly Algorithm (IFA) optimization-based SVR method, not only perform as the best modelling measures and accuracy rate compared with the representative forecast models, but also can achieve a better predictive performance when dealing with abnormal taxi-out time states.

Support Vector Regression and Artificial Neural Network Approaches: Case of Economic Growth in East Africa Community

There has been increased interest of late on the application of nonlinear methods to economic and financial data due to their robustness in handling large and complex data. With increasingly complex ‘big data’, focus has shifted into use of robust techniques in analysis of data. Various nonlinear approaches have so far been established including support vector machine which is widely adapted in classification and regression problems. This research project applied support vector regression technique and neural network models in modeling and forecasting economic growth for the five countries in the East Africa Community including Kenya, Uganda, United Republic of Tanzania, Rwanda and Burundi. Data for the period 1990 to 2014 from World Bank databases was used for the research. Support vector model and neural network models were trained using the data for the 1990-2002 whereas the remaining data was used for prediction performance to determine the robustness of the two models on external datasets. The study revealed that specific country models had better performance compared to the combined model and that although the two models compared similarly under specific-country models, the neural network performed better in most countries. The study recommends the use of the two machine learning techniques in economic growth modeling. It also recommends that the performance be compared with the traditional econometric models but using countries with more data periods.

Short-Term Financial Time Series Forecasting Integrating Principal Component Analysis and Independent Component Analysis with Support Vector Regression

Financial time series forecasting could be beneficial for individual as well as institutional investors. But, the high noise and complexity residing in the financial data make this job extremely challenging. Over the years, many researchers have used support vector regression (SVR) quite successfully to conquer this challenge. In this paper, an SVR based forecasting model is proposed which first uses the principal component analysis (PCA) to extract the low-dimensional and efficient feature information, and then uses the independent component analysis (ICA) to preprocess the extracted features to nullify the influence of noise in the features. Experiments were carried out based on 16 years’ historical data of three prominent stocks from three different sectors listed in Dhaka Stock Exchange (DSE), Bangladesh. The predictions were made for 1 to 4 days in advance targeting the short term prediction. For comparison, the integration of PCA with SVR (PCA-SVR), ICA with SVR (ICA-SVR) and single SVR approaches were applied to evaluate the prediction accuracy of the proposed approach. Experimental results show that the proposed model (PCA-ICA-SVR) outperforms the PCA-SVR, ICA-SVR and single SVR methods.

Evaluation of PC-SAFT model and Support Vector Regression (SVR) approach in prediction of asphaltene precipitation using the titration data

Asphaltene deposition in porous media, wellbore and surface facilities has been a severe problem in petroleum industry which causes considerable remediation costs annually. Asphaltenes are heavy and polydisperse fractions of crude oil which are insoluble in n-alkanes such as n-heptane. In this work, three Iranian crude oils were prepared for titration experiments with n-pentane, n-heptane and n-dodecane at different solvent ratios and constant temperature. The experimental data were correlated by perturbed chain form of statistical associating fluid theory (PC-SAFT). The association of asphaltene molecules has been considered in this model with adjusting the uncertain parameters (such as association energy and association volume of asphaltene pseudo component) to match the experimental data. PC-SAFT parameters for other non-associating pseudo components have been calculated using the correlations proposed in literature. The present study also evaluated the performance of SVR method as a supervised learning approach in prediction of asphaltene precipitation. Deviation of proposed models has been validated using the statistical evaluation criteria and graphical analysis. The results show that the proposed models have AAD values less than 0.073 and a high potential in prediction of asphaltene precipitation.

Sensorless estimation of lake level by soft computing approach

PurposeThe purpose of the paper is to evaluate a lake lavel prediction model. Lake-level prediction is very important task for different crucial issues like planning of water resource, controlled drainage, etc. Therefore, in this study, a soft computing approach was applied to predict the lake levels based on the different prediction horizons.Design/methodology/approachThe main focus was to establish a sensorless estimation of the lake level based on different prediction horizons. Support vector regression approach was implemented for the lake-level prediction, as this approach is suitable for highly nonlinear prediction problems.FindingsLudoš Lake in Serbia was used for analysis.Originality/valueAccording the results, the soft computing models can be used confidently for the lake-level prediction based on the prediction accuracy.

State-of-health estimation for the lithium-ion battery based on support vector regression

Lithium-ion batteries have been widely used in many fields. The state-of-health is necessary and important for battery performance evaluation and lifetime prediction. A reliable state-of-health estimation is essential to help batteries work in a safe and suitable condition. In this paper, a novel state-of-health estimation approach is proposed for lithium-ion batteries based on statistical knowledge. An improved battery model, which combines the open-circuit-voltage modeling and the Thevenin equivalent circuit model, is proposed to improve the model accuracy and study the relation between internal parameters and states of the battery. The joint extended Kalman filter-recursive-least squares algorithm is employed to estimate battery state-of-charge and identify the model parameters and open-circuit-voltage simultaneously. Then a particle swarm optimization-least square support vector regression approach is employed to give a reliable state-of-health estimation result with high accuracy and good generalization ability, where the particle swarm optimization algorithm is used to improve the algorithm ability of global optimization. In order to verify the accuracy of the proposed method, static and dynamic current profile tests are carried out on lithium iron phosphate batteries in different aging levels. The experimental results indicate that the proposed method can present suitability for state-of-health estimation with high accuracy.

Improving Fractional Impervious Surface Mapping Performance through Combination of DMSP-OLS and MODIS NDVI Data

Impervious surface area (ISA) is an important parameter for many studies such as urban climate, urban environmental change, and air pollution; however, mapping ISA at the regional or global scale is still challenging due to the complexity of impervious surface features. The Defense Meteorological Satellite Program’s Operational Linescan System (DMSP-OLS) data have been used for ISA mapping, but high uncertainty existed due to mixed-pixel and data-saturation problems. This paper presents a new index called normalized impervious surface index (NISI), which is an integration of DMSP-OLS and Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) data, in order to reduce these problems. Meanwhile, this newly developed index is compared with previously used indices—Human Settlement Index (HSI) and Vegetation Adjusted Nighttime light Urban Index (VANUI)—in ISA mapping performance. We selected China as an example to map fractional ISA distribution through a support vector regression approach based on the relationship between the index and Landsat-derived ISA data. The results indicate that the proposed NISI provided better ISA estimation accuracy than HSI and VANUI, especially when the fractional ISA in a pixel is relatively large (i.e., >0.6) or very small (i.e., <0.2). This approach can be used to rapidly update ISA datasets at regional and global scales.