Combination Of Support Vector Machine Research Articles

New Importance Analysis Method for Structures with Correlated Inputs

For importance analysis of structural models with correlated inputs, a new method combining support vector machine regression and the single-loop Monte Carlo technique is proposed. In this new combined method, the correlated inputs are transformed to uncorrelated ones by orthogonally independent transformation first, then support vector machine regression is employed to establish the multiple-dimensional regression models involving all inputs, on which the single-loop Monte Carlo is used to obtain the importance measures of the correlated inputs. Support vector machine regression is also used to establish the one-dimensional regression model involving one concerned input. This direct support vector machine regression is proposed as a comparative method. Profiting from the high efficiency of support vector machine regression, the combined and direct methods can get the importance analysis results of the correlated inputs more efficiently than the Monte Carlo simulation. By comparing the results of combined single-loop Monte Carlo technique and those of the direct method, it is found that the former is more efficient and robust than the latter. Several examples illustrate the efficiency and accuracy of the proposed methods.

Classification of Polarimetric SAR Images Based on Modeling Contextual Information and Using Texture Features

This paper proposes a novel contextual method for classification of polarimetric synthetic aperture radar data. The method combines support vector machine (SVM) and Wishart classifiers to benefit from both parametric and nonparametric methods. This method computes the energy function of a Markov random field (MRF) in the neighborhoods of the pixel using Wishart distribution. It then relates the Markovian energydifference function to the SVM classifier. Therefore, the salt-and-pepper effect on the classified map is reduced using a contextual classifier. Moreover, to achieve the full advantage of spatial information, texture features are added into the contextual classification. Texture features are extracted from SPAN images and are added to the SVM classifier. In this paper, two Radarsat-2 polarimetric images acquired in the leaf-off and leaf-on seasons are used from a forest area. Efficient multitemporal information is exploited using composite kernels in SVM. Comparison of the proposed algorithm with the Wishart, Wishart-MRF, SVM, and SVM with composite kernel classifiers shows a 21.72%, 16.17%, 11.29%, and 8.19% improvement in overall accuracy, respectively. Moreover, incorporating texture features into classification results significant increase in the average accuracy in forest species compared with the use of only polarimetric features.

Arrhythmia recognition and classification using combined linear and nonlinear features of ECG signals

Arrhythmia is a cardiac condition caused by abnormal electrical activity of the heart, and an electrocardiogram (ECG) is the non-invasive method used to detect arrhythmias or heart abnormalities. Due to the presence of noise, the non-stationary nature of the ECG signal (i.e. the changing morphology of the ECG signal with respect to time) and the irregularity of the heartbeat, physicians face difficulties in the diagnosis of arrhythmias. The computer-aided analysis of ECG results assists physicians to detect cardiovascular diseases. The development of many existing arrhythmia systems has depended on the findings from linear experiments on ECG data which achieve high performance on noise-free data. However, nonlinear experiments characterize the ECG signal more effectively sense, extract hidden information in the ECG signal, and achieve good performance under noisy conditions. This paper investigates the representation ability of linear and nonlinear features and proposes a combination of such features in order to improve the classification of ECG data. In this study, five types of beat classes of arrhythmia as recommended by the Association for Advancement of Medical Instrumentation are analyzed: non-ectopic beats (N), supra-ventricular ectopic beats (S), ventricular ectopic beats (V), fusion beats (F) and unclassifiable and paced beats (U). The characterization ability of nonlinear features such as high order statistics and cumulants and nonlinear feature reduction methods such as independent component analysis are combined with linear features, namely, the principal component analysis of discrete wavelet transform coefficients. The features are tested for their ability to differentiate different classes of data using different classifiers, namely, the support vector machine and neural network methods with tenfold cross-validation. Our proposed method is able to classify the N, S, V, F and U arrhythmia classes with high accuracy (98.91%) using a combined support vector machine and radial basis function method.

Recombination Hotspot/Coldspot Identification Combining Three Different Pseudocomponents via an Ensemble Learning Approach.

Recombination presents a nonuniform distribution across the genome. Genomic regions that present relatively higher frequencies of recombination are called hotspots while those with relatively lower frequencies of recombination are recombination coldspots. Therefore, the identification of hotspots/coldspots could provide useful information for the study of the mechanism of recombination. In this study, a new computational predictor called SVM-EL was proposed to identify hotspots/coldspots across the yeast genome. It combined Support Vector Machines (SVMs) and Ensemble Learning (EL) based on three features including basic kmer (Kmer), dinucleotide-based auto-cross covariance (DACC), and pseudo dinucleotide composition (PseDNC). These features are able to incorporate the nucleic acid composition and their order information into the predictor. The proposed SVM-EL achieves an accuracy of 82.89% on a widely used benchmark dataset, which outperforms some related methods.

Land use/cover change from 2001 to 2010 and its socioeconomic determinants in Guangdong Province, a rapid urbanization area of China

Rapid economic and population growth exert profound effects on land use/cover change, especially in rapid economically developing areas, such as Guangdong Province-a pioneer of economic growth and urbanization in China. To elucidate the changes and mutual conversions of land-use types and their determinants in Guangdong Province, land-use maps of Guangdong Province from 2001 to 2010 were obtained from Moderate Resolution Imaging Spectroradiometer (MODIS) L1B images with 500 m resolution by using the supervised classification combined support vector machine (SVM) algorithm with a decision tree. The results indicate that during the studied time period, the area of urban land and orchards increased by 4186 km2 and 3615.5 km2, respectively, while the percentage of cropland decreased from 11.46% to 10.14%. Massive urbanization and cropland loss have occurred in the region. The conversion from wasteland to urban land, and the mutual transformation between wasteland and cropland were the most prevalent in Guangdong Province. This was especially obvious from 2007 to 2010, with an annual changing rate of 5.13% and -19.33%, respectively. Several socio-economic factors were also obtained to investigate their impacts on land-use changes in Guangdong Province. It is shown that marked increases in gross domestic product (GDP), total investment in fixed assets, total retail sales of consumer goods, and the increasing development of modern service industries have led to large-scale migration, and industrial structure adjustment and improvement; consequently dramatically rapid expansion of cities and cropland loss have occurred.

A new predictive model for the cyanotoxin content from experimental cyanobacteria concentrations in a reservoir based on the ABC optimized support vector machine approach: A case study in Northern Spain

Cyanotoxins, a kind of poisonous substances produced by cyanobacteria, are responsible for health risks in surface waters used for drinking or for recreation. Consequently, anticipation of its presence is a matter of importance to prevent risks. The aim of this study is to build a cyanotoxin diagnostic model by using support vector machines (SVMs) in combination with the artificial bee colony (ABC) technique from cyanobacterial concentrations determined experimentally in the Trasona reservoir (Northern Spain), to forecast the cyanotoxins' presence in the Trasona reservoir (Northern Spain). The ABC–SVM model is aimed at highly nonlinear biological problems with sharp peaks and the tests carried out have proven its high performance. The results of the present study are two-fold. In the first place, the significance of each biological and physical–chemical variables on the cyanotoxin content in the reservoir is presented through the model. Secondly, a predictive model of the cyanotoxin content is obtained. The agreement of the ABC–SVM-based model with experimental data confirmed its good performance. Finally, conclusions of this innovative research work are exposed.

GIS-based landslide susceptibility mapping with logistic regression, analytical hierarchy process, and combined fuzzy and support vector machine methods: a case study from Wolong Giant Panda Natural Reserve, China

Landslides, as one of the most destructive natural phenomena, distribute extensively in Wolong Giant Panda Natural Reserve and cause damage to both humans and endangered species. Therefore, landslide susceptibility zonation (LSZ) mapping is necessary for government agencies and decision makers to select suitable locations for giant pandas. The main purpose of this study is to produce landside susceptibility maps using logistic regression (LR), analytical hierarchy process (AHP), and a combined fuzzy and support vector machine (F-SVM) hybrid method based on geographic information systems (GIS). A total of 1773 landslide scarps larger than one cell (25 × 25 m2) were selected in the landslide inventory mapping, 70 % of which were selected at random to be used as test data, and the other 30 % were used as validation. Topographical, geological, and hydrographical data were collected, processed, and constructed into a spatial database. Nine conditioning factors were chosen as influencing factors related to landslide occurrence: slope degree, aspect, altitude, profile curvature, geology and lithology, distance from faults, distance from rivers, distance from roads, and normalized difference vegetation index (NDVI). Landslide susceptible areas were analyzed and mapped using the landslide occurrence factors by different methods. For conventional assessment, weights and rates of the affecting factors were assigned based on experience and knowledge of experts. In order to reduce the subjectivity, a combined fuzzy and SVM hybrid model was generated for LSZ in this paper. In this approach, the rates of each thematic layer were generated by the fuzzy similarity method, and weights were created by the SVM method. To confirm the practicality of the susceptibility map produced by this improved method, a comparison study with LR, AHP was assessed by means of their validation. The outcome indicated that the combined fuzzy and SVM method (accuracy is 85.73 %) is better than AHP (accuracy is 78.84 %), whereas it is relatively similar to LR (accuracy is 84.55 %). The susceptibility map based on combined the fuzzy and SVM approach also shows that 5.8 % of the study area is assigned as very highly susceptible areas, and 17.8 % of the study area is assigned as highly susceptible areas.

Automated biphasic morphological assessment of hepatitis B-related liver fibrosis using second harmonic generation microscopy.

Liver fibrosis assessment by biopsy and conventional staining scores is based on histopathological criteria. Variations in sample preparation and the use of semi-quantitative histopathological methods commonly result in discrepancies between medical centers. Thus, minor changes in liver fibrosis might be overlooked in multi-center clinical trials, leading to statistically non-significant data. Here, we developed a computer-assisted, fully automated, staining-free method for hepatitis B-related liver fibrosis assessment. In total, 175 liver biopsies were divided into training (n = 105) and verification (n = 70) cohorts. Collagen was observed using second harmonic generation (SHG) microscopy without prior staining, and hepatocyte morphology was recorded using two-photon excitation fluorescence (TPEF) microscopy. The training cohort was utilized to establish a quantification algorithm. Eleven of 19 computer-recognizable SHG/TPEF microscopic morphological features were significantly correlated with the ISHAK fibrosis stages (P < 0.001). A biphasic scoring method was applied, combining support vector machine and multivariate generalized linear models to assess the early and late stages of fibrosis, respectively, based on these parameters. The verification cohort was used to verify the scoring method, and the area under the receiver operating characteristic curve was >0.82 for liver cirrhosis detection. Since no subjective gradings are needed, interobserver discrepancies could be avoided using this fully automated method.

Prediction of human disease-associated phosphorylation sites with combined feature selection approach and support vector machine.

Phosphorylation is a crucial post-translational modification, which regulates almost all cellular processes in life. It has long been recognised that protein phosphorylation has close relationship with diseases, and therefore many researches are undertaken to predict phosphorylation sites for disease treatment and drug design. However, despite the success achieved by these approaches, no method focuses on disease-associated phosphorylation sites prediction. Herein, for the first time the authors propose a novel approach that is specially designed to identify associations between phosphorylation sites and human diseases. To take full advantage of local sequence information, a combined feature selection method-based support vector machine (CFS-SVM) that incorporates minimum-redundancy-maximum-relevance filtering process and forward feature selection process is developed. Performance evaluation shows that CFS-SVM is significantly better than the widely used classifiers including Bayesian decision theory, k nearest neighbour and random forest. With the extremely high specificity of 99%, CFS-SVM can still achieve a high sensitivity. Besides, tests on extra data confirm the effectiveness and general applicability of CFS-SVM approach on a variety of diseases. Finally, the analysis of selected features and corresponding kinases also help the understanding of the potential mechanism of disease-phosphorylation relationships and guide further experimental validations.

A Parallel Digital VLSI Architecture for Integrated Support Vector Machine Training and Classification

This paper presents a parallel digital VLSI architecture for combined support vector machine (SVM) training and classification. For the first time, cascade SVM, a powerful training algorithm, is leveraged to significantly improve the scalability of hardware-based SVM training and develop an efficient parallel VLSI architecture. The presented architecture achieves excellent scalability by spreading the training workload of a given data set over multiple SVM processing units with minimal communication overhead. Hardware-friendly implementation of the cascade algorithm is employed to achieve low hardware overhead and allow for training over data sets of variable size. In the proposed parallel cascade architecture, a multilayer system bus and multiple distributed memories are used to fully exploit parallelism. In addition, the proposed architecture is rather flexible and can be tailored to realize hybrid use of hardware parallel processing and temporal reuse of processing resources, leading to good tradeoffs between throughput, silicon overhead and power dissipation. Several parallel cascade SVM processors have been designed with a commercial 90-nm CMOS technology, which provide up to a $561\times $ training time speedup and a significant estimated $21\,859\times $ energy reduction compared with the software SVM algorithm running on a 45-nm commercial general-purpose CPU.

Using support vector machine and dynamic parameter encoding to enhance global optimization

This study presents an approach which combines support vector machine (SVM) and dynamic parameter encoding (DPE) to enhance the run-time performance of global optimization with time-consuming fitness function evaluations. SVMs are used as surrogate models to partly substitute for fitness evaluations. To reduce the computation time and guarantee correct convergence, this work proposes a novel strategy to adaptively adjust the number of fitness evaluations needed according to the approximate error of the surrogate model. Meanwhile, DPE is employed to compress the solution space, so that it not only accelerates the convergence but also decreases the approximate error. Numerical results of optimizing a few benchmark functions and an antenna in a practical application are presented, which verify the feasibility, efficiency and robustness of the proposed approach.

Radar target classification using support vector machine and subspace methods

Target classification is a significant research direction in radar field. The range profile is a good target electromagnetic scattering characteristic for real-time target classification. This study proposes a novel method which combines support vector machine (SVM) and subspace methods to achieve complex target classification. The performances of SVM and three representative subspace methods are analysed using range profiles generated by graphical electromagnetic computing method. Experimental results demonstrate that SVM classifier has better robustness in sample variation than conventional classifiers. The auxiliary effects of three subspace methods on classification have respective preponderances in different aspects.

Multi-muscle activation strategies during walking in female post-operative total joint replacement patients

Dynamic knee joint function requires coordinated multi-muscle activation patterns (MMP) that may be adversely affected by total knee arthroplasty (TKA). This study identified MMP changes in post-operative female TKA patients using a Support Vector Machine (SVM). It was hypothesised that TKA patients can successfully be classified and display significant alterations in temporal and spectral muscle activation characteristics. 19 female subjects (10 unilateral gender-specific TKA, 62.2±8.6yrs, BMI 28.2±5.4; and 9 healthy controls, 61.4±7.4yrs, BMI 25.6±2.4) were recruited. Surface electromyograms (EMG) were obtained for 7 lower limb muscles during walking. Stance phase (±30%) EMG data were processed using a wavelet transform and normalized to total power. Data across all muscles were combined to form MMPs and analyzed using a SVM. Recognition rates for all subjects were computed for MMPs and individual muscles. A binomial test was used to establish statistical significance (p<0.05). The results supported the hypothesis indicating significantly altered muscle activations for vastus medialis (recognition rate ∼68.4%) and biceps femoris (recognition rate ∼73.7%). Further analysis identified distinct between group differences across temporal, spectral and intensity domains. Application of a combined SVM and MMP approach may be beneficial for the future assessment of post-surgical dynamic muscle function.

A hybrid PSO optimized SVM-based method for predicting of the cyanotoxin content from experimental cyanobacteria concentrations in the Trasona reservoir: A case study in Northern Spain

There is an increasing need to describe cyanobacteria blooms since some cyanobacteria produce toxins termed cyanotoxins and, as a result, anticipate its presence is a matter of importance to prevent risks. Cyanobacteria blooms occur frequently and globally in water bodies, and they are a major concern in terms of their effects on other species such as plants, fish and other microorganisms, but especially by the possible acute and chronic effects on human health due to the potential danger from cyanobacterial toxins produced by some of them in recreational or drinking waters. Therefore, the aim of this study is to build a cyanotoxin diagnostic model by using support vector machines (SVMs) in combination with the particle swarm optimization (PSO) technique from cyanobacterial concentrations determined experimentally in the Trasona reservoir (recreational reservoir used as a high performance training center of canoeing in the Northern Spain). The Trasona reservoir is near Aviles estuary and after a short tour, the brackish waters of the Aviles estuary empty into the Cantabrian sea. This optimization technique involves kernel parameter setting in the SVM training procedure, which significantly influences the regression accuracy. Bearing this in mind, cyanotoxin contents have been predicted here by using the hybrid PSO–SVM-based model from the remaining measured water quality parameters (input variables) in the Trasona reservoir (Northern Spain) with success. In other words, the results of the present study are two-fold. In the first place, the significance of each biological and physical–chemical variable on the cyanotoxin content in the reservoir is presented through the model. Second, a predictive model able to forecast the possible presence of cyanotoxins is obtained. The agreement of the PSO–SVM-based model with experimental data confirmed its good performance. Finally, conclusions of this innovative research work are exposed.

Pixel- and feature-level fusion of hyperspectral and lidar data for urban land-use classification

The complexity of urban areas makes it difficult for single-source remotely sensed data to meet all urban application requirements. Airborne light detection and ranging (lidar) can provide precise horizontal and vertical point cloud data, while hyperspectral images can provide hundreds of narrow spectral bands which are sensitive to subtle differences in surface materials. The main objectives of this study are to explore: (1) the performance of fused lidar and hyperspectral data for urban land-use classification, especially the contribution of lidar intensity and height information for land-use classification in shadow areas; and (2) the efficiency of combined pixel- and object-based classifiers for urban land-use classification. Support vector machine (SVM), maximum likelihood classification (MLC), and object-based classifiers were used to classify lidar, hyperspectral data and their derived features, such as the normalized digital surface model (nDSM), normalized difference vegetation index (NDVI), and texture measures, into 15 urban land-use classes. Spatial attributes and rules were used to minimize misclassification of the objects showing similar spectral properties, and accuracy assessments were carried out for the classification results. Compared with hyperspectral data alone, hyperspectral–lidar data fusion improved overall accuracy by 6.8% (from 81.7 to 88.5%) when the SVM classifier was used. Meanwhile, compared with SVM alone, the combined SVM and object-based method improved OA by 7.1% (from 87.6 to 94.7%). The results suggest that hyperspectral–lidar data fusion is effective for urban land-use classification, and the proposed combined pixel- and object-based classifiers are very efficient and flexible for the fusion of hyperspectral and lidar data.

Hybrid PSO–SVM-based method for forecasting of the remaining useful life for aircraft engines and evaluation of its reliability

The present paper describes a hybrid PSO–SVM-based model for the prediction of the remaining useful life of aircraft engines. The proposed hybrid model combines support vector machines (SVMs), which have been successfully adopted for regression problems, with the particle swarm optimization (PSO) technique. This optimization technique involves kernel parameter setting in the SVM training procedure, which significantly influences the regression accuracy. However, its use in reliability applications has not been yet widely explored. Bearing this in mind, remaining useful life values have been predicted here by using the hybrid PSO–SVM-based model from the remaining measured parameters (input variables) for aircraft engines with success. A coefficient of determination equal to 0.9034 was obtained when this hybrid PSO–RBF–SVM-based model was applied to experimental data. The agreement of this model with experimental data confirmed its good performance. One of the main advantages of this predictive model is that it does not require information about the previous operation states of the engine. Finally, the main conclusions of this study are exposed.

Identifying DNA-binding proteins by combining support vector machine and PSSM distance transformation.

BackgroundDNA-binding proteins play a pivotal role in various intra- and extra-cellular activities ranging from DNA replication to gene expression control. Identification of DNA-binding proteins is one of the major challenges in the field of genome annotation. There have been several computational methods proposed in the literature to deal with the DNA-binding protein identification. However, most of them can't provide an invaluable knowledge base for our understanding of DNA-protein interactions.ResultsWe firstly presented a new protein sequence encoding method called PSSM Distance Transformation, and then constructed a DNA-binding protein identification method (SVM-PSSM-DT) by combining PSSM Distance Transformation with support vector machine (SVM). First, the PSSM profiles are generated by using the PSI-BLAST program to search the non-redundant (NR) database. Next, the PSSM profiles are transformed into uniform numeric representations appropriately by distance transformation scheme. Lastly, the resulting uniform numeric representations are inputted into a SVM classifier for prediction. Thus whether a sequence can bind to DNA or not can be determined. In benchmark test on 525 DNA-binding and 550 non DNA-binding proteins using jackknife validation, the present model achieved an ACC of 79.96%, MCC of 0.622 and AUC of 86.50%. This performance is considerably better than most of the existing state-of-the-art predictive methods. When tested on a recently constructed independent dataset PDB186, SVM-PSSM-DT also achieved the best performance with ACC of 80.00%, MCC of 0.647 and AUC of 87.40%, and outperformed some existing state-of-the-art methods.ConclusionsThe experiment results demonstrate that PSSM Distance Transformation is an available protein sequence encoding method and SVM-PSSM-DT is a useful tool for identifying the DNA-binding proteins. A user-friendly web-server of SVM-PSSM-DT was constructed, which is freely accessible to the public at the web-site on http://bioinformatics.hitsz.edu.cn/PSSM-DT/.

Expert system for predicting reaction conditions: the Michael reaction case.

A generic chemical transformation may often be achieved under various synthetic conditions. However, for any specific reagents, only one or a few among the reported synthetic protocols may be successful. For example, Michael β-addition reactions may proceed under different choices of solvent (e.g., hydrophobic, aprotic polar, protic) and catalyst (e.g., Brønsted acid, Lewis acid, Lewis base, etc.). Chemoinformatics methods could be efficiently used to establish a relationship between the reagent structures and the required reaction conditions, which would allow synthetic chemists to waste less time and resources in trying out various protocols in search for the appropriate one. In order to address this problem, a number of 2-classes classification models have been built on a set of 198 Michael reactions retrieved from literature. Trained models discriminate between processes that are compatible and respectively processes not feasible under a specific reaction condition option (feasible or not with a Lewis acid catalyst, feasible or not in hydrophobic solvent, etc.). Eight distinct models were built to decide the compatibility of a Michael addition process with each considered reaction condition option, while a ninth model was aimed to predict whether the assumed Michael addition is feasible at all. Different machine-learning methods (Support Vector Machine, Naive Bayes, and Random Forest) in combination with different types of descriptors (ISIDA fragments issued from Condensed Graphs of Reactions, MOLMAP, Electronic Effect Descriptors, and Chemistry Development Kit computed descriptors) have been used. Models have good predictive performance in 3-fold cross-validation done three times: balanced accuracy varies from 0.7 to 1. Developed models are available for the users at http://infochim.u-strasbg.fr/webserv/VSEngine.html . Eventually, these were challenged to predict feasibility conditions for ∼50 novel Michael reactions from the eNovalys database (originally from patent literature).

Diagnostic classification of intrinsic functional connectivity highlights somatosensory, default mode, and visual regions in autism.

Despite consensus on the neurological nature of autism spectrum disorders (ASD), brain biomarkers remain unknown and diagnosis continues to be based on behavioral criteria. Growing evidence suggests that brain abnormalities in ASD occur at the level of interconnected networks; however, previous attempts using functional connectivity data for diagnostic classification have reached only moderate accuracy. We selected 252 low-motion resting-state functional MRI (rs-fMRI) scans from the Autism Brain Imaging Data Exchange (ABIDE) including typically developing (TD) and ASD participants (n = 126 each), matched for age, non-verbal IQ, and head motion. A matrix of functional connectivities between 220 functionally defined regions of interest was used for diagnostic classification, implementing several machine learning tools. While support vector machines in combination with particle swarm optimization and recursive feature elimination performed modestly (with accuracies for validation datasets <70%), diagnostic classification reached a high accuracy of 91% with random forest (RF), a nonparametric ensemble learning method. Among the 100 most informative features (connectivities), for which this peak accuracy was achieved, participation of somatosensory, default mode, visual, and subcortical regions stood out. Whereas some of these findings were expected, given previous findings of default mode abnormalities and atypical visual functioning in ASD, the prominent role of somatosensory regions was remarkable. The finding of peak accuracy for 100 interregional functional connectivities further suggests that brain biomarkers of ASD may be regionally complex and distributed, rather than localized.

Prediction of heating parameters based on support vector machine

Considering the questions of complex non-linearity, large thermal inertia, retardance of a district heating system, it is very difficult to establish accurate mathematical models of heating parameters prediction for the heating system. Correlation analysis of influence factors is used to obtain the major factors influencing heating parameters through analysing operational data of a heating system; these factors serve as input parameters of the predicting model. This paper describes a prediction method that combines Support Vector Machine SVM with neural network. The method creates a network structure between heating parameters and its influence factors. Evaluation indexes of relative error and correlation coefficients are given to analyse the feasibility of the method within the scope of engineering applications through using the network model to regress and predict the heating parameters and compare them with testing data. It turned out that the prediction technique provides powerful guidance for operation of the district heating system.