Conventional Support Vector Machine Research Articles

Train Support Vector Machine Using Fuzzy C-means Without a Prior Knowledge for Hyperspectral Image Content Classification

In this paper, a new cooperative classification method called auto-train support vector machine (SVM) is proposed. This new method converts indirectly SVM to an unsupervised classification method. The main disadvantage of conventional SVM is that it needs a priori knowledge about the data to train it. To avoid using this knowledge that is strictly required to train SVM, in this cooperative method, the data, that is, hyperspectral images (HSIs), are first clustered using Fuzzy C-means (FCM); then, the created labels are used to train SVM. At this stage, the image content is classified using the auto-trained SVM. Using FCM, clustering reveals how strongly a pixel is assigned to a class thanks to the fuzzification process. This information leads to gaining two advantages, the first one is that no prior knowledge about the data (known labels) is needed and the second one is that the training data selection is not done randomly (the training data are selected according to their degree of membership to a class). The proposed method gives very promising results. The method is tested on two HSIs, which are Indian Pines and Pavia University. The results obtained have a very high accuracy of the classification and exceed the existing manually trained methods in the literature.

Train Support Vector Machine Using Fuzzy C-means Without a Prior Knowledge for Hyperspectral Image Content Classification

In this paper, a new cooperative classification method called auto-train support vector machine (SVM) is proposed. This new method converts indirectly SVM to an unsupervised classification method. The main disadvantage of conventional SVM is that it needs a priori knowledge about the data to train it. To avoid using this knowledge that is strictly required to train SVM, in this cooperative method, the data, that is, hyperspectral images (HSIs), are first clustered using Fuzzy C-means (FCM); then, the created labels are used to train SVM. At this stage, the image content is classified using the auto-trained SVM. Using FCM, clustering reveals how strongly a pixel is assigned to a class thanks to the fuzzification process. This information leads to gaining two advantages, the first one is that no prior knowledge about the data (known labels) is needed and the second one is that the training data selection is not done randomly (the training data are selected according to their degree of membership to a class). The proposed method gives very promising results. The method is tested on two HSIs, which are Indian Pines and Pavia University. The results obtained have a very high accuracy of the classification and exceed the existing manually trained methods in the literature.

Monthly runoff prediction at Baitarani river basin by support vector machine based on Salp swarm algorithm

Accurate monthly runoff prediction is still challenging work regardless of the accessibility of different modelling techniques, like the knowledge-driven or data-driven models, and human activities and climate changes. To this context, applicability of hybrid SVM-SSA (Support Vector Machine with Salp Swarm Algorithm) model and conventional SVM and artificial neural network (ANN) models is investigated for runoff prediction in Baitarani river basin, Odisha, India. Potential of proposed techniques is measured utilising four quantitative indexes, root mean squared error (RMSE), coefficient of determination (R2), mean absolute error (MAE), and Willmott index (WI). Test results specify that hybrid model generates better prediction accurateness in comparison to applied conventional methods. The generalization and robustness of SVM-SSA techniques were very prominent, with R2 values of 0.9847 and 0.9844 for Anandpur and Champua stations during training phases. Similarly prominent value of WI are 0.9906, 0.9902 and minimum value of RSE and MAE are 20.019, 0.5928 and 0.0769, 0.5934 for Anandpur and Champua stations respectively. Therefore, SVM-SSA can be recommended for modeling complexity of interactions for rainfall-runoff process and predicting runoff.

A DPWM Modulation Strategy to Reduce Common-Mode Voltage for Indirect Matrix Converters Based on Active-Current Vector Amplitude Characteristics

Based on the characteristics that the amplitude of active-current vector in the rectifier stage is zero when the zero-voltage vector is used in the inverter stage, and any one from the six active-current vectors can be selected, a novel discontinuous pulsewidth modulation (DPWM) strategy for the indirect matrix converter is proposed in this article to reduce the common-mode voltage (CMV). In this DPWM strategy, the reference current vector in the rectifier stage is synthesized by two adjacent active-current vectors, and the reference voltage vector in the inverter stage is synthesized by two adjacent active-voltage vectors and one zero-voltage vector. When zero-voltage vectors are used in the inverter stage, an appropriate active-current vector with the lowest input phase voltage among the two adjacent active-current vectors is used to produce a lower CMV. Under the same switching loss, the proposed strategy has a better input performance and better output performance in high voltage transfer ratio than conventional SVM. Finally, the effectiveness of the modulation strategy is verified by simulation and experiments.

Fully Decomposed Singular Value and Fixed Dictionary Extreme Learning Machine for Bogie Fault Diagnosis

As an essential part in the rail train, the bogie plays an important role in the safety of the train operation. However, the fluctuant wheel-rail connection, as well as the structure and complex operating environment of the bogie always lead to low signal-to-noise ratio condition and complicated wheel-rail dynamic coupling relationship. The existing fault diagnosis methods can hardly perform well in this scenario. Concerning this issue, a novel feature extraction method named fully decomposed singular value (FdSV) is proposed in this paper. FdSV can decompose singular value characteristics of signals completely and increase the divergence of features to extract weak fault features effectively. Then, inspired by the theory of compressed perception and Hierarchy-ELM, a fixed dictionary extreme learning machine (FD-ELM) is also proposed for fault identification. This method calculates the weight matrix by formulas without randomization and removes the bias matrix. Therefore, it can easily discover the internal laws of data and improve the running speed and accuracy rapidly. Finally, the proposed algorithms have been verified by actual bogie data collected from bogies under low SNR and variable working conditions. Compared with SVD, the FdSV features are 1%-6% higher in testing accuracies. The accuracies of FD-ELM are 2-20% higher than the conventional ELM, H-ELM and SVM.

A New Hybrid Multiclass Approach Based on KNN and SVM

Support vector machine (SVM) is a machine learning method widely used in solving binary data classification problems due to its performance. Nevertheless, in practical problems of classification, there are often cases of the presence of more than two classes of objects in the original dataset. The paper considers a solution to the problem of SVM multiclass with the aim to increase the data classification quality based on a new way of hybridisation between SVM and [Formula: see text]-nearest neighbour (KNN) algorithms. The first phase of the approach is called the filtering phase. At this level, the feature space is split into two classes separated by a hyperplane. In the next step called review, we generate a second hyperplane, then we calculate the distance between each test pattern and the second hyperplane in the feature space using e.g. the KNN function. The result of the two phases is three classes instead of two produced by the conventional SVM. For evaluation purposes, dataset experiments are conducted on seven benchmark datasets that have high dimensionality and large size. Numerical experiments show that the 3SVM approach can improve not only the accuracy compared to other multiclass SVM approaches, but also the precision, recall, and [Formula: see text]-score.

Classification of Dysphonic Voices in Parkinson's Disease with Semi-Supervised Competitive Learning Algorithm.

This article proposes a novel semi-supervised competitive learning (SSCL) algorithm for vocal pattern classifications in Parkinson’s disease (PD). The acoustic parameters of voice records were grouped into the families of jitter, shimmer, harmonic-to-noise, frequency, and nonlinear measures, respectively. The linear correlations were computed within each acoustic parameter family. According to the correlation matrix results, the jitter, shimmer, and harmonic-to-noise parameters presented as highly correlated in terms of Pearson’s correlation coefficients. Then, the principal component analysis (PCA) technique was implemented to eliminate the redundant dimensions of the acoustic parameters for each family. The Mann–Whitney–Wilcoxon hypothesis test was used to evaluate the significant difference of the PCA-projected features between the healthy subjects and PD patients. Eight dominant PCA-projected features were selected based on the eigenvalue threshold criterion and the statistical significance level (p < 0.05) of the hypothesis test. The SSCL algorithm proposed in this paper included the procedures of the competitive prototype seed selection, K-means optimization, and the nearest neighbor classifications. The pattern classification experimental results showed that the proposed SSCL method can provide the excellent diagnostic performances in terms of accuracy (0.838), recall (0.825), specificity (0.85), precision (0.846), F-score (0.835), Matthews correlation coefficient (0.675), area under the receiver operating characteristic curve (0.939), and Kappa coefficient (0.675), which were consistently better than those results of conventional KNN or SVM classifiers.

Identifying molecular functional groups of organic compounds by deep learning of NMR data.

We preprocess the raw nuclear magnetic resonance (NMR) spectrum and extract key features by using two different methodologies, called equidistant sampling and peak sampling for subsequent substructure pattern recognition. We also provide a strategy to address the imbalance issue frequently encountered in statistical modeling of NMR data set and establish two conventional support vector machine (SVM) and K-nearest neighbor (KNN) models to assess the capability of two feature selections, respectively. Our results in this study show that the models using the selected features of peak sampling outperform those using equidistant sampling. Then we build the recurrent neural network (RNN) model trained by data collected from peak sampling. Furthermore, we illustrate the easier optimization of hyperparameters and the better generalization ability of the RNN deep learning model by detailed comparison with traditional machine learning SVM and KNN models.

Deep Learning-Based Device-Free Localization Scheme for Simultaneous Estimation of Indoor Location and Posture Using FMCW Radars.

Indoor device-free localization (DFL) systems are used in various Internet-of-Things applications based on human behavior recognition. However, the usage of camera-based intuitive DFL approaches is limited in dark environments and disaster situations. Moreover, camera-based DFL schemes exhibit certain privacy issues. Therefore, DFL schemes with radars are increasingly being investigated owing to their efficient functioning in dark environments and their ability to prevent privacy issues. This study proposes a deep learning-based DFL scheme for simultaneous estimation of indoor location and posture using 24-GHz frequency-modulated continuous-wave (FMCW) radars. The proposed scheme uses a parallel 1D convolutional neural network structure with a regression and a classification model for localization and posture estimation, respectively. The two-dimensional location information of the target is estimated for localization, and four different postures, namely standing, sitting, lying, and absence, are estimated simultaneously. We experimentally evaluated the proposed scheme and compared its performance with that of conventional schemes under identical conditions. The results indicate that the average localization error of the proposed scheme is 0.23 m, whereas that of the conventional scheme is approximately 0.65 m. The average posture estimation error of the proposed scheme is approximately 1.7%, whereas that of the conventional correlation, CSP, and SVM schemes are 54.8%, 42%, and 10%, respectively.

Fast Detection of Diarrhetic Shellfish Poisoning Toxins in Mussels Using NIR Spectroscopy and Improved Twin Support Vector Machines

Diarrhetic shellfish poisoning (DSP) toxins are potent marine biotoxins. It can cause a severe gastrointestinal illness by the consumption of mussels contaminated by DSP toxins. New methods for effectively and rapidly detecting DSP toxins-contaminated mussels are required. In this study, we used near-infrared (NIR) reflection spectroscopy combined with pattern recognition methods to detect DSP toxins. In the range of 950-1700 nm, the spectral data of healthy mussels and DSP toxins-contaminated mussels were acquired. To select optimal waveband subsets, a waveband selection algorithm with a Gaussian membership function based on fuzzy rough set theory was applied. Considering that detecting DSP toxins-contaminated mussels from healthy mussels was an imbalanced classification problem, an improved approach of twin support vector machines (TWSVM) was explored, which is based on a centered kernel alignment. The influences of parameters of the waveband selection algorithm and regularization hyperparameters of the improved TWSVM (ITWSVM) on the performance of models were analyzed. Compared to conventional SVM, TWSVM, and other state-of-the-art algorithms (such as multi-layer perceptron, extreme gradient boosting and adaptive boosting), our proposed model exhibited better performance in detecting DSP toxins and was little affected by the imbalance ratio. For the proposed model, the F-measure reached 0.9886, and detection accuracy reached 98.83%. We explored the physical basis for the detection model by analyzing the relationship between the occurrence of overtone and combination bands and selected wavebands. This study supports NIR spectroscopy as an innovative, rapid, and convenient analytical method to detect DSP toxins in mussels.

Early Fault Diagnosis Model Design of Reciprocating Compressor Valve Based on Multiclass Support Vector Machine and Decision Tree

According to the character of frequent fault occurrence, difficult diagnosis of large reciprocating compressor valves, an early fault diagnosis model of reciprocating compressor valve based on multiclass support vector machine and decision tree is designed. A series of simulation experiments of the suction valve and exhaust valve on a large-scale reciprocating compressor experimental bench are made and the valve fault principle is analyzed. Using the advantages of fast and efficient decision tree classification and the prominent characteristics of support vector machine in small sample binary classification, a multivariate classification and recognition model is constructed. The typical characteristic parameters of gearbox vibration signal are extracted as the fault feature vector training model under different fault conditions, and the samples are tested. The experimental results show that the recognition effect of this method is significantly better than that of the neural network method in the case of small samples, and the recognition efficiency is improved more than that of the conventional multivariate support vector machine method which can be effectively applied to reciprocating compressor valve fault diagnosis.

Application of Hybrid Support Vector Machine model for Streamflow Prediction in Barak valley, India

Forecasting streamflow (Qflow) is vital in flood and water management, determining potential of river water flow, agricultural practices, hydropower generation, and environmental flow study. This research aims to explore capability of hybrid support vector machines (SVM) with Whale Optimisation Algorithm (WOA) model for forecasting streamflow at Badarpur Ghat gauging station of Barak river basin and evaluate its enactment with the conventional SVM model. Root mean squared error (RMSE), mean absolute error (MAE), and Nash-Sutcliffe efficiency (NSE) statistical measures are considered as evaluating standards. Assessment of outcomes indicates that the optimization algorithm could enhance the accurateness of standalone SVM model in monthly streamflow forecasting. Compared to conventional artificial intelligence methods without a data pre-processing system, the comparatively good performance of applied hybrid model gives an effective alternate to achieve better precision in streamflow forecasting. Results confirm that enhanced SVM model can better process a multifaceted hydrogeological data set, have higher prediction accuracy, and possess better generalisation capability.

Classification of EEG signals using machine learning and deep learning techniques

Electroencephalogram (EEG) signals reveal electrical activity of brain in a person. Brain cells interact by impulses even during sleep. Any disruptions to these impulses induce problems in the individual. Hence clinicians analyze electrical activity of the brain by EEG readings to comprehend the disruptions in the impulses. EEG and its sub bands depict electrical pattern of human brain. EEG data comprises transient components, spikes, other sorts of artifacts due to eye blinking, movement of the individual, anxiousness etc. during EEG collection. Wavelet transformations are effective mathematical technique for sampling approximation to produce clear EEG. It also assists in filtering, sampling, interpolation, noise reduction, signal approximation and signal augmentation, feature extraction. In this study, a survey is done on EEG motor imagery signals and the various classifiers to assess them machine learning methods for EEG signal categorization are studied. Conventional SVM and logistic regression methods combined with basic 2-layer Neural Network (NN) are constructed using python in keras to assess the performances.

Brain MRI Analysis for Alzheimer's Disease Diagnosis Using CNN-Based Feature Extraction and Machine Learning.

Alzheimer’s disease is the most common form of dementia and the fifth-leading cause of death among people over the age of 65. In addition, based on official records, cases of death from Alzheimer’s disease have increased significantly. Hence, early diagnosis of Alzheimer’s disease can increase patients’ survival rates. Machine learning methods on magnetic resonance imaging have been used in the diagnosis of Alzheimer’s disease to accelerate the diagnosis process and assist physicians. However, in conventional machine learning techniques, using handcrafted feature extraction methods on MRI images is complicated, requiring the involvement of an expert user. Therefore, implementing deep learning as an automatic feature extraction method could minimize the need for feature extraction and automate the process. In this study, we propose a pre-trained CNN deep learning model ResNet50 as an automatic feature extraction method for diagnosing Alzheimer’s disease using MRI images. Then, the performance of a CNN with conventional Softmax, SVM, and RF evaluated using different metric measures such as accuracy. The result showed that our model outperformed other state-of-the-art models by achieving the higher accuracy, with an accuracy range of 85.7% to 99% for models with MRI ADNI dataset.

Efficient Reliability Analysis of Structures Using Symbiotic Organisms Search-Based Active Learning Support Vector Machine

Reliability-based design optimization considers the uncertainties that lie in the designing process of resilient buildings and structures. To model uncertainty, the major challenge is to lower the high computational expense incurred by the double-loop approach, where the design optimization (outer loop) repeatedly calls the reliability analysis of each structural design (inner loop). An alternative is to convert the reliability constraints to deterministic constraints by using optimality conditions. Yet, the approximated results are often inaccurate when constraint functions are highly non-linear, non-continuous, or non-differentiable. To achieve better accuracy while attaining sufficient flexibility, the present study proposes a new framework to classify the structural designs into feasible/infeasible designs. The proposed framework is called SOS-ASVM by integrating the symbiotic organism search (SOS) and the active-learning support vector machine (ASVM). ASVM is adopted as the surrogate model, while SOS is used to seek more representative samples to improve the classification accuracy of ASVM. The SOS-ASVM was validated by comparisons with popular classification tools: conventional support vector machine, artificial neural network, and Kriging model. Three practical engineering cases are used to demonstrate the performance of the SOS-ASVM: a cantilever beam, a bracket structure, and a 25-bar space truss. The comparison results confirm the superiority of the proposed framework to other tools.

At-line quality assurance of deep-fried instant noodles using pilot scale visible-NIR spectroscopy combined with deep-learning algorithms

Deep-fried instant noodles produced in a pilot scale facility were monitored for quality parameters using at-line visible-NIR spectroscopy (380–1650 nm) combined with deep-learning algorithms. To build robust calibrations, a wider range of quality parameters for instant noodles viz., moisture (1.6–11.04%), crude protein (8.34–14.39%), total fat (12.38–26.68%), and total ash (1.18–3.15%) were selected. The original raw spectral data was subjected to different pretreatments (standard normal variate (SNV), detrending (DT), first derivative (FD), multiplicative scattering correction (MSC), among others) before being optimized for wavelength selection feature algorithm by competitive adaptive reweighed sampling (CARS). The calibration models were established using deep-learning algorithms based on conventional partial least squares regression (PLSR) and support vector machine regression (SVMR). In general, the SVMR modeling gave an optimum prediction statistics (pretreatment method, coefficient of determination ( R 2 ), root mean square error of prediction (RMSEP), ratio of prediction to deviation (RPD)) for moisture (CARS–SNV–DT, 0.98, 0.32, 6.7), crude protein (CARS–FD, 0.98, 0.18, 8.15), and total fat (CARS–MSC, 0.99, 0.39, 10.15) whereas partial least squares regression (PLSR) gave for total ash (CARS–raw, 0.94, 0.12, 4.34). In particularly for noodle manufacturers, the unification of visible-NIR spectroscopy and deep-learning algorithm is a promising to realize sustainability in quality assurance and control. • Vis-NIR and deep-learning algorithm applied to quality monitoring of noodles. • Calibration models were optimized by CARS feature selection algorithm. • CARS-SVMR gave best predictions for moisture, crude protein, and total fat. • SVMR algorithm successfully used for at-line vis-NIR based quality monitoring.

Dataset for classification of computer graphic images and photographic images

&lt;span lang="EN-US"&gt;The recent advancements in computer graphics (CG) image rendering techniques have made it easy for the content creators to produce high quality computer graphics similar to photographic images (PG) confounding the most naïve users. Such images used with negative intent, cause serious problems to the society. In such cases, proving the authenticity of an image is a big challenge in digital image forensics due to high photo-realism of CG images. Existing datasets used to assess the performance of classification models are lacking with: (i) larger dataset size, (ii) diversified image contents, and (iii) images generated with the recent digital image rendering techniques. To fill this gap, we created two new datasets, namely, ‘JSSSTU CG and PG image dataset’ and ‘JSSSTU PRCG image dataset’. Further, the complexity of the new datasets and benchmark datasets are evaluated using handcrafted texture feature descriptors such as gray level co-occurrence matrix, local binary pattern and VGG variants (VGG16 and VGG19) which are pre-trained convolutional neural network (CNN) models. Experimental results showed that the CNN-based pre-trained techniques outperformed the conventional support vector machine (SVM)-based classifier in terms of classification accuracy. Proposed datasets have attained a low f-score when compared to existing datasets indicating they are very challenging.&lt;/span&gt;

Correct and remap solar radiation and photovoltaic power in China based on machine learning models

Accurate estimation of surface solar radiation (SSR) is crucial for photovoltaic (PV) systems design and solar PV power plants site selection. However, the SSR observations often suffer from inhomogeneity issues (e.g., aging equipment and instrument replacement) and low spatial–temporal coverage, which constrained the management and development of PV systems, particularly for countries with huge solar energy investments (e.g., China). To address them, we proposed seven different models (including four machine learning models, two empirical models and a multiple linear regression model) for accurate prediction of daily SSR using conventional meteorological data. Our results showed that the support vector machine (SVM) outperformed other models in estimating daily SSR. Based on SSR data corrected by the SVM model, the sharp downtrend in SSR (-9.64 W m−2 decade-1) due to the sensitivity drift of instrument aging before 1990 was moderated (-2.36 W m−2 decade-1) in China; and the abnormal jump caused by instrument replacement in SSR observations during 1990–1993 was removed. Furthermore, by combining the advantages of large network of conventional meteorological data and SVM model, we extended the limited station-based SSR data (∼100 meteorological stations) to include longer period and larger spatial coverage (2185 meteorological stations, 1971–2016) in China. By coupling the grid-based method and PV power model, we created reliable SSR and PV power maps with higher temporal and spatial resolution over China. Spatially, the main distribution of high SSR and PV power were in the northwestern China, Tibetan Plateau and some coastal areas in the southern China. Temporally, the PV power experienced a significant drop during 1971–2016 (-1.94 kWh m−2 decade-1) due to the significant decline in SSR (-0.91 W m−2 decade-1). Above all, the new maps of SSR and PV power prepared here should support PV investments within the Chinese territory.

Improving accuracy of SVM for monthly sediment load prediction using Harris hawks optimization

One of the most significant concerns for water resource projects is estimating sediment load (SL). In last three decades, the application of soft computing models for SL. prediction has broadly increased and provided effective outcomes. Present study applies a support vector machine (SVM) for estimating SL. at two stations of Baitarani river basin, Odisha, India. A meta-heuristic algorithm called Harris hawks optimization (HHO) is utilized for enhancing accuracy of SVM model in monthly SL prediction. Lagged rainfall, temperature, stage, runoff are considered different inputs for developing all models. The ability of SVM-HHO is benchmarked with SVM-PSO (particle swarm optimization), SVM-GWO (Grey wolves optimization), along with conventional SVM and ANN (artificial neural network) models. Collected data is divided into two subgroups (training and testing), and SL is estimated. Performance of applied methodologies is evaluated by quantitative statistical measures, such as coefficient of determination (R2), root mean square error (RMSE), and Nash-Sutcliffe efficiency (NSE). Obtained results revealed that highest value of NSE (0.9898), R2 (0.988), and a minimum of RMSE (0.6682) was generated by SVM-HHO in testing phase compared to other studied techniques. This leads to a conclusion regarding superiority of HHO evolutionary algorithm in enhancing accuracy of conventional SVM for monthly SL prediction. Ability of HHO algorithm to escape local optima marks SVM-HHO as a potential method in SL estimation. Findings of this study tend to ascertain the suitability of employed approach for precisely modelling SL in rivers.

Pattern Analysis and Regressive Linear Measure for Botnet Detection

Capturing the distributed platform with remotely controlled compromised machines using botnet is extensively analyzed by various researchers. However, certain limitations need to be addressed efficiently. The provisioning of detection mechanism with learning approaches provides a better solution more broadly by saluting multi-objective constraints. The bots’ patterns or features over the network have to be analyzed in both linear and non-linear manner. The linear and non-linear features are composed of high-level and low-level features. The collected features are maintained over the Bag of Features (BoF) where the most influencing features are collected and provided into the classifier model. Here, the linearity and non-linearity of the threat are evaluated with Support Vector Machine (SVM). Next, with the collected BoF, the redundant features are eliminated as it triggers overhead towards the predictor model. Finally, a novel Incoming data Redundancy Elimination-based learning model (RedE-L) is built to classify the network features to provide robustness towards BotNets detection. The simulation is carried out in MATLAB environment, and the evaluation of proposed RedE-L model is performed with various online accessible network traffic dataset (benchmark dataset). The proposed model intends to show better trade-off compared to the existing approaches like conventional SVM, C4.5, RepTree and so on. Here, various metrics like Accuracy, detection rate, Mathews Correlation Coefficient (MCC), and some other statistical analysis are performed to show the proposed RedE-L model's reliability. The F1-measure is 99.98%, precision is 99.93%, Accuracy is 99.84%, TPR is 99.92%, TNR is 99.94%, FNR is 0.06 and FPR is 0.06 respectively.