Supervised Support Vector Machine Research Articles

Characterization of high background radiation of terrestrial naturally occurring radionuclides in a mining region of Senegal

A survey of natural radioactivity has been carried out to estimate the concentration of naturally occurring radionuclides and radiological risk associated in the south East mining region, which present high natural background radiation. An in-situ gamma-ray spectrometer was used to map natural environmental gamma-emitting radionuclides. A combined inferential statistical and chemometrics of naturally occurring radionuclides were used for data modelling and characterization. The radiological data surveys were explored using inferential statistical, principal component analysis and a supervised support vector machine learning. First, one-way Analysis of variance, on-parametric Kruskal Wallis test, was applied allowing pairwise comparison of radionuclides levels in sampling sites, second, data was submitted to PCA to extract noise-free data and reduced data was analyzed with support vector machine. PCA results show that 238U contribution (56%) is dominant in the first principal component with 40K (35%). The second component, with the cumulative variance explained of 88%, is dominated by 232Th (68%) and 40K (31%). Anova indicates that there is a significant difference in the mean mass concentration of samples type and soil activities are mostly lower. The best classification accuracy was 100% with the use of radial kernel density function. As potential uranium and gold mine site, these results will allow establishing both reference values for background radiation of the region and fingerprinting sources of naturally occurring radionuclides.

Rapid identification of edible oil species using supervised support vector machine based on low-field nuclear magnetic resonance relaxation features

Aimed to rapidly identify the edible oils according to their botanical origin, a novel method was proposed using supervised support vector machine based on low-field nuclear magnetic resonance and relaxation features. The low-field (LF) nuclear magnetic resonance (NMR) signals of 11 types of edible oils were acquired, and 5 features were extracted from the transverse relaxation decay curves and modeled using support vector machines (SVM) for the identification of edible oils. Two SVM classification strategies have been applied and discussed. Good performance can be achieved when the relative position of each edible oil has been determined by PCA before the designing of binary tree structure of SVM model, and the classification accuracy is 99.04%. The good robustness of this method has been verify at different data sets. It is almost a real time method, and the entire process takes only 144 s.

A Machine Learning Approach to Evaluating the Damage Level of Tooth-Shape Spur Dikes

Little research has been done on the application of machine learning approaches to evaluating the damage level of river training structures on the Yangtze River. In this paper, two machine learning approaches to evaluating the damage level of spur dikes with tooth-shaped structures are proposed: a supervised support vector machine (SVM) model and an unsupervised model combining a Kohonen neural network with an SVM model (KNN-SVM). It was found that the supervised SVM model predicted the damage level of the validation samples with high accuracy, and the unsupervised data-mining KNN-SVM model agreed well with the empirical evaluation result. It is shown that both machine learning approaches could become effective tools to evaluate the damage level of spur dikes and other river training structures.

Performance evaluation of textural features in improving land use/land cover classification accuracy of heterogeneous landscape using multi-sensor remote sensing data

Texture analysis of remote sensing images has been received a substantial amount of attention as it plays a vital role in improving the classification accuracy of heterogeneous landscape. However, it is inadequately studied that how the images from different sensors with varying spatial resolutions influence the choice of textural features. This study endeavors to examine the textural features from the Landsat 8-OLI, RISAT-1, Resourcesat 2-LISS III, Sentinel-1A and Resourcesat 2-LISS IV satellite images with spatial resolution of 30, 25, 23.5, 5×20 and 5.8 m respectively, for improving land use/land cover (LULC) classification accuracy. The textural features were extracted from the aforesaid sensor data with the assistance of gray-level co-occurrence matrix (GLCM) with different moving window sizes. The best combination of textural features was recognized using standard deviations and correlation coefficients following separability analysis of LULC categories based on training samples. A supervised support vector machine (SVM) classifier was employed to perform LULC classification and the results were evaluated using ground truth information. This work demonstrates the significance of textural features in improving the classification accuracy of heterogeneous landscape and it becomes more significant as the spatial resolution improved. It is also revealed that textures are vital especially in the case of SAR data.

Supervised Support Vector Machine in Predicting Foreign Exchange Trading

Supervised Support Vector Machine in Predicting Foreign Exchange Trading

Earthquake/Tsunami Damage Assessment for Urban Areas Using Post-Event PolSAR Data

Analyses of single-post-event polarimetric synthetic aperture radar (PolSAR) data permit fast and convenient post-disaster damage assessment work. By analyzing valid features, damaged and undamaged buildings can be quickly classified. However, the presence of oriented buildings in the disaster area makes the classification work more challenging. Many previous works extract the damage information of the disaster area by considering oriented buildings and undamaged parallel buildings as survived buildings. However, after-effect debris may create structures with random orientation angles. In our study on the Tohoku earthquake/tsunami disaster event, we found that some damaged buildings with large building orientation angles (with respect to the satellite flight path) are grouped as oriented buildings (undamaged buildings). In this paper, we propose a new earthquake/tsunami damage assessment method, particularly for urban areas, that takes this complex situation into consideration. The proposed method solves the problems of both urban-area extraction and damaged-building identification. For urban-area extraction, the proposed combined thresholding and majority voting method can accurately discriminate between urban and foreshortening mountain areas. Meanwhile, for damaged-building identification, the proposed new unsupervised damage assessment method classifies the buildings in a disaster area according to four conditions, and it outperforms the techniques used in existing works. The analysis results and the comparison with the supervised support vector machine (SVM) classification technique show that our proposed method can produce more accurate results for damage assessment using single-post-event PolSAR data.

Surface Defects Classification of Hot-Rolled Steel Strips Using Multi-directional Shearlet Features

In this paper, a method combining the use of discrete shearlet transform (DST) and the gray-level co-occurrence matrix (GLCM) is presented to classify surface defects of hot-rolled steel strips into the six classes of rolled-in scale, patches, crazing, pitted surface, inclusion and scratches. Feature extraction involves the extraction of multi-directional shearlet features from each input image followed by GLCM calculations from all extracted sub-bands, from which a set of statistical features is extracted. The resultant high-dimensional feature vectors are then reduced using principal component analysis. A supervised support vector machine classifier is finally trained to classify the surface defects. The proposed feature set is compared against the Gabor, wavelets and the original GLCM in order to evaluate and validate its robustness. Experiments were conducted on a database of hot-rolled steel strips consisting of 1800 grayscale images whose defects exhibit high inter-class similarity as well as high intra-class appearance variations. Results indicate that the proposed DST–GLCM method is superior to other methods and achieves classification rates of 96.00%.

Assessing urban development trends in representative Local Administrative Units before and after the Greek economic crisis

This study maps and quantifies urban development in four areas in Northern Greece for the last decade using remote sensing and urban sprawl indices. It aims to correlate the results with the effects of the Greek economic crisis on the construction activity. The remotely sensed data used are two multispectral cloud-free Landsat 5 Thematic Mapper (TM) images (from 2003 and 2009) and one Landsat 8 Operational Land Imager (OLI) image from 2014. A supervised Support Vector Machine (SVM) classifier is applied to produce land cover maps. The maps accuracy is assessed using high resolution airborne and Google Earth imagery for the 2003, 2009 and 2014 images. The high overall built-up area classification accuracy achieved ranges between 85.8% and 87.8%. The derived maps are further assessed to identify and monitor the impervious surface changes. A set of urban sprawl indices is applied to quantitatively analyse the spatial characteristics and spatial imprint of built-up area through time. The analysis of this dynamic phenomenon reveals that the economic crisis not only decreased the number of new constructions but also altered the way new construction occurs.

Semi-supervised learning in large scale text categorization

The rapid development of the Internet brings a variety of original information including text information, audio information, etc. However, it is difficult to find the most useful knowledge rapidly and accurately because of its huge number. Automatic text classification technology based on machine learning can classify a large number of natural language documents into the corresponding subject categories according to its correct semantics. It is helpful to grasp the text information directly. By learning from a set of hand-labeled documents, we obtain the traditional supervised classifier for text categorization (TC). However, labeling all data by human is labor intensive and time consuming. To solve this problem, some scholars proposed a semi-supervised learning method to train classifier, but it is unfeasible for various kinds and great number of Web data since it still needs a part of hand-labeled data. In 2012, Li et al. invented a fully automatic categorization approach for text (FACT) based on supervised learning, where no manual labeling efforts are required. But automatically labeling all data can bring noise into experiment and cause the fact that the result cannot meet the accuracy requirement. We put forward a new idea that part of data with high accuracy can be automatically tagged based on the semantic of category name, then a semi-supervised way is taken to train classifier with both labeled and unlabeled data, and ultimately a precise classification of massive text data can be achieved. The empirical experiments show that the method outperforms the supervised support vector machine (SVM) in terms of both F1 performance and classification accuracy in most cases. It proves the effectiveness of the semi-supervised algorithm in automatic TC.

Globally supported radial basis function based collocation method for evolution of level set in mass segmentation using mammograms

Computer-aided detection systems play an important role for the detection of breast abnormalities using mammograms. Global segmentation of mass in mammograms is a complex process due to low contrast mammogram images, irregular shape of mass, speculated margins, and the presence of intensity variations of pixels. This work presents a new approach for mass detection in mammograms, which is based on the variational level set function. Mesh-free based radial basis function (RBF) collocation approach is employed for the evolution of level set function for segmentation of breast as well as suspicious mass region. The mesh-based finite difference method (FDM) is used in literature for evolution of level set function. This work also showcases a comparative study of mesh-free and mesh-based approaches. An anisotropic diffusion filter is employed for enhancement of mammograms. The performance of mass segmentation is analyzed by computing statistical measures. Binarized statistical image features (BSIF) and variants of local binary pattern (LBP) are computed from the segmented suspicious mass regions. These features are given as input to the supervised support vector machine (SVM) classifier to classify suspicious mass region as mass (abnormal) or non-mass (normal) region. Validation of the proposed algorithm is done on sample mammograms taken from publicly available Mini-mammographic image analysis society (MIAS) and Digital Database for Screening Mammography (DDSM) datasets. Combined BSIF features perform better as compared to LBP variants with the performance reported as 97.12% sensitivity, 92.43% specificity, and 98% AUC with 5.12 FP/I on DDSM dataset; and 95.12% sensitivity, 92.41% specificity, and 95% AUC with 4.01FP/I on MIAS dataset.

Contributions of C-Band SAR Data and Polarimetric Decompositions to Subarctic Boreal Peatland Mapping

The objective of this paper is to assess the accuracy of C-band synthetic aperture radar (SAR) datasets in mapping peatland types over a region of Canada's subarctic boreal zone. This paper assessed contributions of quad-polarization linear backscatter intensities (σ°HH, σ°HV, σ°VV), image textures, and two polarimetric scattering decompositions: 1) Cloude-Pottier, and 2) Freeman-Durden. Four quad-polarimetric RADADSAT-2 images were studied at incidence angles of 19.4°, 23.1°, 45.8°, and 48.1°. The influence of combining dual-angular information acquired within a short temporal span was also assessed. These C-band SAR data were used to classify peatlands according to isolated flat bogs (bogs), channel fens (fens), raised peat plateaus (plateaus), and forested uplands (uplands) using a supervised support vector machine (SVM) classifier. Numerous classifications were examined to compare the unique contributions of these variables to classification accuracy. Results suggest linear backscatter variables in isolation produce comparable classification results with those of the Freeman-Durden and Cloude-Pottier decompositions. Combining polarimetric decomposition and texture data into classifications with linear backscatter data resulted in only minor (~1-3%) improvement. Combining classifications from small and large incidence angles (dual-angular) significantly improved classification results over those of a single image. Classification accuracy was the highest for isolated bogs and open water surfaces, whereas fens, uplands, and plateaus had lower accuracies. The highest accuracy classification (84% and kappa coefficient of 0.80) used a dual-angular approach, with additional decomposition and texture information. However, it is noted that texture information rarely improved classification results across all tests. This approach identified isolated flat bogs, channel fens, and raised peat plateaus with >76% producer's accuracies.

Fully automated diagnosis of papilledema through robust extraction of vascular patterns and ocular pathology from fundus photographs.

Rapid development in the field of ophthalmology has increased the demand of computer aided diagnosis of various eye diseases. Papilledema is an eye disease in which the optic disc of the eye is swelled due to an increase in intracranial pressure. This increased pressure can cause severe encephalic complications like abscess, tumors, meningitis or encephalitis, which may lead to a patient's death. Although there have been several papilledema case studies reported from a medical point of view, only a few researchers have presented automated algorithms for this problem. This paper presents a novel computer aided system which aims to automatically detect papilledema from fundus images. Firstly, the fundus images are preprocessed by going through optic disc detection and vessel segmentation. After preprocessing, a total of 26 different features are extracted to capture possible changes in the optic disc due to papilledema. These features are further divided into four categories based upon their color, textural, vascular and disc margin obscuration properties. The best features are then selected and combined to form a feature matrix that is used to distinguish between normal images and images with papilledema using the supervised support vector machine (SVM) classifier. The proposed method is tested on 160 fundus images obtained from two different data sets i.e. structured analysis of retina (STARE), which is a publicly available data set, and our local data set that has been acquired from the Armed Forces Institute of Ophthalmology (AFIO). The STARE data set contained 90 and our local data set contained 70 fundus images respectively. These annotations have been performed with the help of two ophthalmologists. We report detection accuracies of 95.6% for STARE, 87.4% for the local data set, and 85.9% for the combined STARE and local data sets. The proposed system is fast and robust in detecting papilledema from fundus images with promising results. This will aid physicians in clinical assessment of fundus images. It will not take away the role of physicians, but will rather help them in the time consuming process of screening fundus images.

Presurgical thalamocortical connectivity is associated with response to vagus nerve stimulation in children with intractable epilepsy.

Although chronic vagus nerve stimulation (VNS) is an established treatment for medically-intractable childhood epilepsy, there is considerable heterogeneity in seizure response and little data are available to pre-operatively identify patients who may benefit from treatment. Since the therapeutic effect of VNS may be mediated by afferent projections to the thalamus, we tested the hypothesis that intrinsic thalamocortical connectivity is associated with seizure response following chronic VNS in children with epilepsy. Twenty-one children (ages 5–21 years) with medically-intractable epilepsy underwent resting-state fMRI prior to implantation of VNS. Ten received sedation, while 11 did not. Whole brain connectivity to thalamic regions of interest was performed. Multivariate generalized linear models were used to correlate resting-state data with seizure outcomes, while adjusting for age and sedation status. A supervised support vector machine (SVM) algorithm was used to classify response to chronic VNS on the basis of intrinsic connectivity. Of the 21 subjects, 11 (52%) had 50% or greater improvement in seizure control after VNS. Enhanced connectivity of the thalami to the anterior cingulate cortex (ACC) and left insula was associated with greater VNS efficacy. Within our test cohort, SVM correctly classified response to chronic VNS with 86% accuracy. In an external cohort of 8 children, the predictive model correctly classified the seizure response with 88% accuracy. We find that enhanced intrinsic connectivity within thalamocortical circuitry is associated with seizure response following VNS. These results encourage the study of intrinsic connectivity to inform neural network-based, personalized treatment decisions for children with intractable epilepsy.

Automated diagnosis of macular edema and central serous retinopathy through robust reconstruction of 3D retinal surfaces

Background and objectivesMacular diseases tend to damage macula within human retina due to which the central vision of a person is affected. Macular edema (ME) and central serous retinopathy (CSR) are two of the most common macular diseases. Many researchers worked on automated detection of ME from optical coherence tomography (OCT) and fundus images, whereas few researchers have worked on diagnosing central serous retinopathy. But this paper proposes a fully automated method for the classification of ME and CSR through robust reconstruction of 3D OCT retinal surfaces. MethodsThe proposed system uses structure tensors to extract retinal layers from OCT images. The 3D retinal surface is then reconstructed by extracting the brightness scan (B-scan) thickness profile from each coherent tensor. The proposed system extracts 8 distinct features (3 based on retinal thickness profile of right side, 3 based on thickness profile of left side and 2 based on top surface and cyst spaces within retinal layers) from 30 labeled volumes (10 healthy, 10 CSR and 10 ME) which are used to train the supervised support vector machines (SVM) classifier. ResultsIn this research we have considered 90 OCT volumes (30 Healthy, 30 CSR and 30 ME) of 73 patients to test the proposed system where our proposed system correctly classified 89 out of 90 cases and has promising receiver operator characteristics (ROC) ratings with accuracy, sensitivity and specificity of 98.88%, 100%, and 96.66% respectively. ConclusionThe proposed system is quite fast and robust in detecting all the three types of retinal pathologies from volumetric OCT scans. The proposed system is fully automated and provides an early and on fly diagnosis of ME and CSR syndromes. 3D macular thickness surfaces can further be used as decision support parameter in clinical studies to check the volume of cyst.

Fuzzy c-means based support vector machines classifier for perfume recognition

Identification of more than three perfumes is very difficult for the human nose. It is also a problem to recognize patterns of perfume odor with an electronic nose that has multiple sensors. For this reason, a new hybrid classifier has been presented to identify type of perfume from a closely similar data set of 20 different odors of perfumes. The structure of this hybrid technique is the combination of unsupervised fuzzy clustering c-mean (FCM) and supervised support vector machine (SVM). On the other hand this proposed soft computing technique was compared with the other well-known learning algorithms. The results show that the proposed hybrid algorithm’s accuracy is 97.5% better than the others.

Mapping degraded grassland on the Eastern Tibetan Plateau with multi-temporal Landsat 8 data — where do the severely degraded areas occur?

The Tibetan Plateau in Western China is the world’s largest alpine landscape, sheltering a rich diversity of native flora and fauna. In the past few decades, the Tibetan Plateau was found to suffer from grassland degradation processes. Grassland degradation is assumed to not only endanger biodiversity but also to increase the risk for natural hazards in other parts of the country which are ecologically and hydrologically connected to the area. However, the mechanisms behind the degradation processes remain poorly understood due to scarce baseline data and insufficient scientific research.We argue that remote sensing data can help to better understand degradation processes and patterns by: (1) identifying the distribution of severely degraded areas and (2) comparing the patterns of key spatial attributes of the identified areas (altitude above sea level, aspect, slope, administrative districts) with existing theories on degradation drivers. Therefore, we applied four Landsat 8 images covering large portions of the three counties Jigzhi, Baima and Darlag in the Eastern Tibetan Plateau. The dates of the Landsat scenes were selected to cover differing phenological stages of the ecosystem. Reference data were collected with a remotely piloted aircraft and a standard consumer RGB camera. To exploit the phenological information in the Landsat data as well as deal with the problem of cloud cover in multiple images, we developed a straightforward PCA-based procedure to merge the Landsat scenes. The merged Landsat data served as input to a supervised support vector machine classification which was validated with an iterative bootstrap procedure and an additional independent validation set. The considered classes were “high-cover grassland”, “grassland (including several stages of grassland vitality)”, “(severely) degraded grassland”, “green shrubland”, “grey shrubland”, “urban areas” and “water bodies”. Kappa accuracies ranged between 0.84 and 0.93 in the iterative procedure, while the independent validation led to a kappa accuracy of 0.76. Mean producer’s and user’s accuracies for all classes were higher than 80%, and confusion mainly occurred between the two shrubland classes and between the three grassland classes.Analysis of the slope, aspect and altitude values of the vegetation classes revealed that the degraded areas mostly occurred at the higher altitudes of the study area (4300–4600m), with no strong connection to any specific slope or aspect. High-cover grassland was mostly located on sunny slopes at lower altitudes (less than 4300m), while shrubland preferred shady, relatively steep slopes across all altitudes. These observations proved to be stable across the examined counties, while the proportions of land-cover classes differed between the examined regions. Most counties showed 5–7% severely degraded land cover. Darlag, the county located at the edge of the permafrost zone, and featuring the highest average altitude and lowest annual temperature and precipitation, was found to suffer from larger areas of severe degradation (14%).Therefore, our findings support a strong connection between degradation patterns and climatic as well as altitudinal gradients, with an increased degradation risk for high altitude areas and areas in colder and drier climatic zones. This is relevant information for pastoral management to avoid further degradation of high altitude pastures.

Augmenting MRI-transrectal ultrasound-guided prostate biopsy with temporal ultrasound data: a clinical feasibility study.

In recent years, fusion of multi-parametric MRI (mp-MRI) with transrectal ultrasound (TRUS)-guided biopsy has enabled targeted prostate biopsy with improved cancer yield. Target identification is solely based on information from mp-MRI, which is subsequently transferred to the subject coordinates through an image registration approach. mp-MRI has shown to be highly sensitive to detect higher-grade prostate cancer, but suffers from a high rate of false positives for lower-grade cancer, leading to unnecessary biopsies. This paper utilizes a machine-learning framework to further improve the sensitivity of targeted biopsy through analyzing temporal ultrasound data backscattered from the prostate tissue. Temporal ultrasound data were acquired during targeted fusion prostate biopsy from suspicious cancer foci identified in mp-MRI. Several spectral features, representing the signature of backscattered signal from the tissue, were extracted from the temporal ultrasound data. A supervised support vector machine classification model was trained to relate the features to the result of histopathology analysis of biopsy cores obtained from cancer foci. The model was used to predict the label of biopsy cores for mp-MRI-identified targets in an independent group of subjects. Training of the classier was performed on data obtained from 35 biopsy cores. A fivefold cross-validation strategy was utilized to examine the consistency of the selected features from temporal ultrasound data, where we achieved the classification accuracy and area under receiver operating characteristic curve of 94 % and 0.98, respectively. Subsequently, an independent group of 25 biopsy cores was used for validation of the model, in which mp-MRI had identified suspicious cancer foci. Using the trained model, we predicted the tissue pathology using temporal ultrasound data: 16 out of 17 benign cores, as well as all three higher-grade cancer cores, were correctly identified. The results show that temporal analysis of ultrasound data is potentially an effective approach to complement mp-MRI-TRUS-guided prostate cancer biopsy, specially to reduce the number of unnecessary biopsies and to reliably identify higher-grade cancers.

Implementation of Support Vector Machines and Clustering of Intrusion Detection System for Computer Networks

Considering that intrusion detection systems and anomaly clearly recognize malicious activity. Nowadays, data mining based intrusion detection systems, security and more rapidly detect attacks.Therefore,in this article we use a combination of k-means clustering algorithm and is used supervised support vector machine algorithm to find the best line separator. This is leading to the separation of normal and attack traffic.

Object-based classification of QuickBird image and low point density LIDAR for tropical trees and shrubs mapping

This paper assessed the performance of object-based supervised support vector machine (SVM) and rule-based techniques in classifying tropical vegetated floodplain using 0.6m QuickBird image and LIDAR dataset of 1.4 points per square meter (PPSM). This is particularly significant in hydraulic modelling in which vegetation roughness is a big uncertainty and largely relies on land cover classification. The supervised classification resulted in 79.40% overall accuracy whilst the results improved by 8% with rule-based classification. 40 sample plots of trees and shrubs were measured to be compared to obtain the best classification results. The results showed a linear relationship between tree diameters and NDVI with a high Pearson correlation of 0.76 and coefficient of determination (r2) of 0.58. The canopy areas of shrubs were found to be representative spatially with an even higher Pearson correlation of 0.98 and r2 of 0.95. The study concluded that the fusion of QuickBird image and low point density LIDAR in rule-based classification together with field data were useful in quantifying tropical trees and shrubs.

Spatial–Spectral Information-Based Semisupervised Classification Algorithm for Hyperspectral Imagery

Semisupervised learning has shown its great potential in land cover mapping. It exploits the information of unlabeled training samples and converts those samples to labeled training samples to enhance classification. In this paper, the spatial information extracted by a two-dimensional (2-D) Gabor filter was stacked with spectral information first, and then the spatial neighborhood information of labeled training samples was combined with active learning (AL) algorithm to select the most useful and informative samples, which were used as the unlabeled set to aid the probability model-based supervised support vector machine (SVM). Experiments on two hyperspectral datasets showed that the spatial-spectral information-based semisupervised classification algorithm ($\bf{S}^{\bf{2}} \bf{ISC}$) can produce high classification accuracy with a small number of labeled samples, and outperformed the compared semisupervised algorithms.