Supervised Classification Techniques Research Articles

Bearing Health Monitoring Based on Hilbert–Huang Transform, Support Vector Machine, and Regression

The detection, diagnostic, and prognostic of bearing degradation play a key role in increasing the reliability and safety of electrical machines, especially in key industrial sectors. This paper presents a new approach that combines the Hilbert-Huang transform (HHT), the support vector machine (SVM), and the support vector regression (SVR) for the monitoring of ball bearings. The proposed approach uses the HHT to extract new heath indicators from stationary/nonstationary vibration signals able to tack the degradation of the critical components of bearings. The degradation states are detected by a supervised classification technique called SVM, and the fault diagnostic is given by analyzing the extracted health indicators. The estimation of the remaining useful life is obtained by a one-step time-series prediction based on SVR. A set of experimental data collected from degraded bearings is used to validate the proposed approach. The experimental results show that the use of the HHT, the SVM, and the SVR is a suitable strategy to improve the detection, diagnostic, and prognostic of bearing degradation.

Prediction of rare single-nucleotide causative mutations for muscular diseases in pooled next-generation sequencing experiments.

Next-generation sequencing (NGS) is a new approach for biomedical research, useful for the diagnosis of genetic diseases in extremely heterogeneous conditions. In this work, we describe how data generated by high-throughput NGS experiments can be analyzed to find single nucleotide polymorphisms (SNPs) in DNA samples of patients affected by neuromuscular disorders. In particular, we consider untagged pooled NGS data, where DNA samples of different individuals are combined in a single experiment, still providing information with an uncertainty limited to only two patients. At the moment, only few publications address the problem of SNPs detection in pooled experiments, and existing tools are often inaccurate. We propose a computational procedure consisting of two parts. In the first, data are filtered by means of decision rules. The second phase is based on a supervised classification technique. In the present work, we compare different de facto standard supervised and unsupervised procedures to identify and classify variants potentially related to muscular diseases, and we discuss results in terms of statistical and biological validation.

Targeted Land-Cover Classification

This paper addresses a specific typology of land-cover classification problems, hereinafter referred to as “targeted land-cover classification,” where the objective is the identification of only one or few specific “targeted” land-cover classes of interest, disregarding all the other potential classes present in the area under analysis. Such a challenging problem, which is common to a variety of operational information services and applications (e.g., agriculture, forestry, spatial planning, ecosystem monitoring, disaster management, habitat mapping, etc.), can be effectively solved by traditional supervised classification techniques provided that an exhaustive ground truth is available for all the land-cover classes present in the region of interest. Such a requirement is seldom satisfied and presents several practical drawbacks and limitations, both in terms of time and economic cost that may render this task difficult to achieve in most real-life cases. However, the possibility to perform an effective targeted classification using only ground-truth samples for the class(es) of interest (hence avoiding the burden and cost associated with the collection of a full and exhaustive ground-truth information) would represent a significant advantage. In this paper, we present a novel technique capable of identifying specific land-cover classes of interest by exploiting the ground truth only available for these targeted classes, while providing accuracies comparable to those of traditional fully supervised methods. The proposed technique jointly exploits both the unlabeled samples of the image under investigation and the training samples only available for the targeted classes. In particular, the expectation-maximization algorithm and Markov random fields are employed to estimate the probability density function of both the class(es) of interest and the unknown class representing the merger of all the unknown land-cover classes characterizing the study area for which no ground-truth information is available. An extensive experimental analysis and cross-comparisons with both fully supervised support vector machines and ensembles of multiple one-class support-vector-data-description classifiers on different data sets confirmed the effectiveness and the reliability of the proposed technique.

Performance analysis of landsat5 remote sensing data compression technique used for land use and land cover mapping

Multispectral band remote sensing imagery is used for environmental monitoring and land use and land cover mapping purposes. This image contains huge volume of data. Instead of using the entire data for land use land cove mapping, the spatially compressed images can also be used for mapping purposes. In this paper discrete wavelet transform is selected for compressing the Landsat5 image and the quality has been analysed using the parameters compression ratio, peak signal to noise ratio and digital number values. Using the digital number values the spectral signature graph is drawn. Finally only one wavelet is selected for land use and land cover mapping based on minimum cumulative error of the digital number values. Then the selected wavelet compressed image is classified using supervised classification technique and accuracy assessment is made by constructing the error matrix. Finally the selected wavelet compressed image is used for land use and land cover mapping. Keywords: Compression Ratio (CR), Peak Signal to Noise Ratio (PSNR), Digital Number (DN), Image Classification, Error Matrix, Spectral Signatures.

COMPARATIVE STUDY OF VARIOUS SUPERVISEDCLASSIFICATION METHODSFORANALYSING DEFORESTATION FACTORS

In this paper, various supervised classification techniques are compared and the results are demonstrated. Here the classification techniques like Decision tree, Bayesian method, Neural Networks and Rule Based method are discussed with regards to the data sets given. The population, built up and agriculture are the factors that play a vital role in the development of country which directly affects economic condition. In this paper, factors such as road , population, built up development ,agriculture and industry are considered as drivers of deforestation in the study area which is located in the Erode District of TamilNadu, India.

Application of remote sensing for gold exploration in the Fawakhir area, Central Eastern Desert of Egypt

Eastern desert of Egypt is well-endowed with noble mineral resources that provide challenges for the future economy and development in Egypt. Such mineral resources are not yet well-mapped and fully understood; however, various geological maps are available. The newly advanced sensors of earth observations such as advanced space thermal emission and reflection radiometer (ASTER) images were functioned to map the lithology, the structural patterns and the alteration zones related to gold mineralization at the Fawakhir area, Central Eastern Desert of Egypt. Integration of band rationing and principal component analysis (PCA) clearly determined the lithological units of the study area. The alteration zones of the study area were efficiently mapped from the ASTER radiance data using reference spectra based on two supervised classification techniques including the spectral angle mapper (SAM) and spectral information divergence (SID). Field visits and laboratory microscopic examination of thin and polished sections were used for validating the process. The final resulted map of the alteration mineral zones consists of sericite, clay minerals, quartz, carbonate, and pyrite. Auto-detection of the structural lineaments, which are zones of deformation and fracturing that implies secondary porosity, was mapped from the satellite image. These structures represent significant pathways for migration of hydrothermal solutions, which influence the creation of the alteration zones and become potential zones for mineral resources. The spatial analysis showed the high correlation between the density of lineaments and the occurrence of the alteration zones, which represent high probability of the presence of gold deposits that validated against the old gold mines.

A proxy for snow cover and winter ground surface cooling: Mapping Usnea sp. communities using high resolution remote sensing imagery (Maritime Antarctica)

Usnea sp. formations show a spatial distribution coinciding with wind-exposed locations on rock knobs or sedimentary bodies, while they are commonly absent from concave sites. Field collection of georeferenced ground truthing data in the Meseta Norte (Fildes Peninsula, Maritime Antarctica) and the application of supervised classification techniques over a summer high resolution QuickBird satellite scene showed excellent classification accuracy for the different landcover types. The results show that Usnea formation distribution maps are a viable proxy for areas with less snow during the cold season. Such an approach provides input for permafrost and active layer modelling since snow acts as a critical control on ground surface heat balance. Since snow mapping is extremely difficult in Maritime Antarctica our tested approach provides important added-value for empirical–statistical modelling of permafrost distribution.

LAND USE LAND COVER DYNAMICS OF NILGIRIS DISTRICT, INDIA INFERRED FROM SATELLITE IMAGERIES

Land use Land cover changes are critical components in managing natural resources especially in hilly region as they trigger the erosion of soil and thus making the zone highly vulnerable to landslides. T he Nilgiris district of Tamilnadu state in India is th e first biosphere in Western Ghats region with rare species of flora and fauna and often suffered by fr equent landslides. Therefore in this present study land use land cover dynamics of Nilgiri district has bee n studied from 1990 to 2010 using Satellite Remote Sensing Technique. The temporal changes of land use and land cover changes of Nilgiris district over t he period of 1990 to 2010 were monitored using LISS I and LISS III of IRS 1A and IRS-P6 satellites. Land use dynamics were identified using Maximum likelihood classification under supervised classification technique. From the remote sensing study, it is fou nd that during the study period of 1990 to 2010, ar ea of dense forest increased by 27.17%, forest plantat ion area decreased by 54.64%. Conversion of forest plantation, Range land and open forest by agricultu re and settlement leading to soil erosion and landslides. Tea plantation increased by 33.95% and agricultural area for plantation of vegetables increased rapidly to 217.56% in the mountain steep area. The accuracy of classification has been asses sed by forming confusion matrix and evaluating kappa coefficient. The overall accuracy has been obtained a s 83.7 and 89.48% for the years 1990 and 2010 respect ively. The kappa coefficients were reported as 0.80 and 0.88 respectively for the years 1990 and 2010.

Network-based Prediction of Cancer under Genetic Storm.

Classification of cancer patients using traditional methods is a challenging task in the medical practice. Owing to rapid advances in microarray technologies, currently expression levels of thousands of genes from individual cancer patients can be measured. The classification of cancer patients by supervised statistical learning algorithms using the gene expression datasets provides an alternative to the traditional methods. Here we present a new network-based supervised classification technique, namely the NBC method. We compare NBC to five traditional classification techniques (support vector machines (SVM), k-nearest neighbor (kNN), naïve Bayes (NB), C4.5, and random forest (RF)) using 50–300 genes selected by five feature selection methods. Our results on five large cancer datasets demonstrate that NBC method outperforms traditional classification techniques. Our analysis suggests that using symmetrical uncertainty (SU) feature selection method with NBC method provides the most accurate classification strategy. Finally, in-depth analysis of the correlation-based co-expression networks chosen by our network-based classifier in different cancer classes shows that there are drastic changes in the network models of different cancer types.

Cross-Device Automated Prostate Cancer Localization With Multiparametric MRI

Prostate cancer localization using supervised classification techniques has aroused considerable interest in medical imaging community in recent years. However, it is crucial to have an accurate training data set for supervised classification techniques. Since different devices with, e.g., different protocols and/or field strengths cause different intensity profiles, each device/protocol must have an accompanying training data set, which is very costly to obtain. It is highly desirable to adapt the existing classifier(s) trained for one device/protocol to help classify data coming from another device/protocol. In this paper, we propose a novel method that has the ability to design classifiers obtained from one imaging protocol and/or MRI device to be used on a data set from another protocol and/or imaging device. As an example problem, we consider prostate cancer localization with multiparametric MRI. We show that simple normalization techniques such as z-score are not sufficient for cross-device automated cancer localization. On the other hand, the method we have originally developed based on relative intensity allows us to successfully use a classifier obtained from one device to be applied on a test patient imaged with another device. Proposed method also allows us to employ T2-weighted MR images directly instead of an additional step to normalize T2-weighted images usually performed in an ad hoc manner when T2 maps are not available. To demonstrate the effectiveness of the proposed method, we use a multiparametric MRI data set acquired from 18 biopsy-confirmed cancer patients with two separate scanners: 1) 1.5-T (Excite HD) GE and 2) 1.5-T (Achieva) Philips Healthcare scanners. A comprehensive visual, quantitative, and statistical analysis of the results show that methods we have developed allow us to: 1) perform cross-device automated classification and 2) use T2-weighted images without an ad hoc subject-specific normalization.

Land Cover Change Assessment of Vaal Harts Irrigation Scheme using Multi-temporal Satellite Sata

Abstract Land cover change (LCC) is important to assess the land use/land cover changes with respect to the development activities like irrigation. The region selected for the study is Vaal Harts Irrigation Scheme (VHS) occupying an area of approximately 36, 325 hectares of irrigated land. The study was carried out using Land sat data of 1991, 2001, 2005 covering the area to assess the changes in land use/land cover for which supervised classification technique has been applied. The Normalized Difference Vegetation Index (NDVI) index was also done to assess vegetative change conditions during the period of investigation. By using the remote sensing images and with the support of GIS the spatial pattern of land use change of Vaal Harts Irrigation Scheme for 15 years was extracted and interpreted for the changes of scheme. Results showed that the spatial difference of land use change was obvious. The analysis reveals that 37.86% of additional land area has been brought under fallow land and thus less irrigation area (18.21%). There is an urgent need for management program to control the loss of irrigation land and therefore reclaim the damaged land in order to make the scheme more viable.

Application of remote sensing in exploration for uranium mineralization in Gabal El Sela area, South Eastern Desert, Egypt

This research aims at integrating remote sensing data and field studies to prospect for radioactive materials at Gabal El Sela area, South Eastern Desert of Egypt. This area is a part of the Arabo-Nubian shield, which is constituted by Late Precambrian basement rocks and developed during the Pan African tectono-thermal event. Geological interpretation of Landsat ETM+ image and field studies revealed that the study area is mainly covered by ophiolitic ultramafic rocks, intermediate metavolocanics, granodiorites, biotite granites and garnetiferous muscovite granites. These rocks are injected by pegmatitic and quartz veins and cut by acidic and basic dykes. The processed Landsat ETM+ data enable also to identify uraniferous alteration zones hosted in the granitic rocks (using band ratio 5/7, 4/5, 3/1 in RGB, density slicing and supervised classification techniques). The measured reflectance values (DNs) of six spectral bands of Landsat ETM+ for all exposed rock types in the study area show that the alteration zones have the highest reflectance values. The study recorded the presence of uranium mineralizations in alteration zones associating with the granitic rocks occurring along ENE–WSW trending shear zones. Three uranium-bearing alteration zones hosted in granitic rocks were detected. These zones are characterized by ferrugination, silicification, grizenization and kaolinitization. The most promising alteration zones (Site 1) occurs at the northeastern part of Gabal El Sela and is characterized by very high radioactive anomalies (up to 5000ppm eU). Microprobe analyses of samples collected from these alteration zones reveal the presence of uraninite crystals. It seems that the high fractures density of the granitic rocks acted as good channels for the ascending hydrothermal fluids and the percolating meteoric water, that leached the uranium minerals disseminated all over the rocks and redeposited them along the fractures of the shear zones. This study demonstrates the utility of orbital remote sensing data for finding unknown mineralized zones in the Eastern Desert of Egypt and similar arid regions.

Using Landsat data for assessing forest cover change and fragmentation in Laljhadi corridor of Kanchanpur district, Nepal

The study, carried out at Laljhadi corridor in Kanchanpur district of Nepal, aimed at assessing forest cover change and fragmentation using multi-temporal Landsat data. Post classifi cation change detection was applied on temporal forest cover class datasets obtained by supervised classifi cation technique with maximum likelihood algorithm. The overall change analysis indicated a decreasing trend in forest cover. Statistics on selected landscape metrics were generated to quantify the change in spatial structure resulting from fragmentation. The analysis of the landscape metrics depicted increase in fragmentation over the analysis time period along with progression of deforestation.DOI: http://dx.doi.org/10.3126/banko.v21i2.9142Banko Janakari Vol. 21, No. 2, 2011 Page: 40-44 Uploaded date: November 11, 2013

Multispectral remote sensing for site-specific nitrogen fertilizer management

The objective of this work was to evaluate the use of multispectral remote sensing for site-specific nitrogen fertilizer management. Satellite imagery from the advanced spaceborne thermal emission and reflection radiometer (Aster) was acquired in a 23 ha corn-planted area in Iran. For the collection of field samples, a total of 53 pixels were selected by systematic randomized sampling. The total nitrogen content in corn leaf tissues in these pixels was evaluated. To predict corn canopy nitrogen content, different vegetation indices, such as normalized difference vegetation index (NDVI), soil-adjusted vegetation index (Savi), optimized soil-adjusted vegetation index (Osavi), modified chlorophyll absorption ratio index 2 (MCARI2), and modified triangle vegetation index 2 (MTVI2), were investigated. The supervised classification technique using the spectral angle mapper classifier (SAM) was performed to generate a nitrogen fertilization map. The MTVI2 presented the highest correlation (R²=0.87) and is a good predictor of corn canopy nitrogen content in the V13 stage, at 60 days after cultivating. Aster imagery can be used to predict nitrogen status in corn canopy. Classification results indicate three levels of required nitrogen per pixel: low (0-2.5 kg), medium (2.5-3 kg), and high (3-3.3 kg).

Semi-Supervised Classification and its Application to Filtering IDS False Positives

Alert classifiers built with the supervised classification technique require large amounts of labeled training alerts. Preparing for such training data is very difficult and expensive. Thus accuracy and feasibility of current classifiers are greatly restricted. This paper employs semi-supervised learning to build alert classification model to reduce the number of needed labeled training alerts. Alert context properties are also introduced to improve the classification performance. Experiments have demonstrated the accuracy and feasibility of our approach.

Utilization of supervised classification in structural and lithological mapping of Wadi Al-Marwah Area, NW Arabian Shield, Saudi Arabia

Wadi Al-Marwah area is located in the northwestern part of the Arabian Shield, Saudi Arabia. It is mainly covered by Precambrian igneous and sedimentary rock units. This area was not subjected to previous detailed lithological or structural mapping. This study aims to apply supervised classification technique of remotely sensed digital satellite data of Landsat 7 for detailed lithological and structural mapping of the area. The fusion between multispectral Enhanced Thematic Mapper (ETM)+ data and high-resolution panchromatic ETM+ band-8 produced a color composite fused image for the study area, scale 1:50,000. The structural lineaments of the study area were extracted and interpreted from the digital imageries data. Little discrepancies or improvements were detected when combining the supervised classification results with the Landsat ratios or principal component analysis. These highlighted the benefits of multispectral classification, especially in terms of lithologic discrimination. The overall results of image processing techniques, applied in this work, were excellent and succeeded in the performance of a more detailed and accurate lithological and structural maps (scale 1:50,000) than the previous published maps for the investigated area.

Appraisal of land use/land cover of mangrove forest ecosystem using support vector machine

Human activities in many parts of the world have greatly changed the natural land cover. This study has been conducted on Pichavaram forest, south east coast of India, famous for its unique mangrove bio-diversity. The main objectives of this study were focused on monitoring land cover changes particularly for the mangrove forest in the Pichavaram area using multi-temporal Landsat images captured in the 1991, 2000, and 2009. The land use/land cover (LULC) estimation was done by a unique hybrid classification approach consisting of unsupervised and support vector machine (SVM)-based supervised classification. Once the vegetation and non-vegetation classes were separated, training site-based classification technology i.e., SVM-based supervised classification technique was used. The agricultural area, forest/plantation, degraded mangrove and mangrove forest layers were separated from the vegetation layer. Mud flat, sand/beach, swamp, sea water/sea, aquaculture pond, and fallow land were separated from non-vegetation layer. Water logged areas were delineated from the area initially considered under swamp and sea water-drowned areas. In this study, the object-based post-classification comparison method was employed for detecting changes. In order to evaluate the performance, an accuracy assessment was carried out using the randomly stratified sampling method, assuring distribution in a rational pattern so that a specific number of observations were assigned to each category on the classified image. The Kappa accuracy of SVM classified image was highest (94.53 %) for the 2000 image and about 94.14 and 89.45 % for the 2009 and 1991 images, respectively. The results indicated that the increased anthropogenic activities in Pichavaram have caused an irreversible loss of forest vegetation. These findings can be used both as a strategic planning tool to address the broad-scale mangrove ecosystem conservation projects and also as a tactical guide to help managers in designing effective restoration measures.

Automatic characterization of neointimal tissue by intravascular optical coherence tomography

Intravascular optical coherence tomography (IVOCT) is rapidly becoming the method of choice for assessing vessel healing after stent implantation due to its unique axial resolution <20 μm. The amount of neointimal coverage is an important parameter. In addition, the characterization of neointimal tissue maturity is also of importance for an accurate analysis, especially in the case of drug-eluting and bioresorbable stent devices. Previous studies indicated that well-organized mature neointimal tissue appears as a high-intensity, smooth, and homogeneous region in IVOCT images, while lower-intensity signal areas might correspond to immature tissue mainly composed of acellular material. A new method for automatic neointimal tissue characterization, based on statistical texture analysis and a supervised classification technique, is presented. Algorithm training and validation were obtained through the use of 53 IVOCT images supported by histology data from atherosclerotic New Zealand White rabbits. A pixel-wise classification accuracy of 87% and a two-dimensional region-based analysis accuracy of 92% (with sensitivity and specificity of 91% and 93%, respectively) were found, suggesting that a reliable automatic characterization of neointimal tissue was achieved. This may potentially expand the clinical value of IVOCT in assessing the completeness of stent healing and speed up the current analysis methodologies (which are, due to their time- and energy-consuming character, not suitable for application in large clinical trials and clinical practice), potentially allowing for a wider use of IVOCT technology.

Evaluation of various image classification techniques on Landsat to identify coral reefs

Coral reefs are one of the prominent marine ecosystems in the world. Coral reefs are facing threats from both natural and anthropogenic factors. Monitoring, protecting and studying these ecosystems are considered as a complex process because they are underwater features. Remote sensing can be quite useful in this by providing huge amount of database of satellite images. Landsat series have been providing satellite data for last 50 years and now it is one of the large databases providing satellite imageries. Accurate processing of these images can increase the accuracy of coral reef information extraction from these images. Classification is one of the important processes that can give adequate and precise information about coral reefs. This study is intended to compare various supervised classification techniques available for extraction of information from the image and to suggest best of them. Maximum Likelihood and Support Vector Machine have proven to be good classification module that can be used to classify Landsat Images to coral reef detection and monitoring.

An Analysis on Performance of Decision Tree Algorithms using Student’s Qualitative Data

Decision Tree is the most widely applied supervised classification technique. The learning and classification steps of decision tree induction are simple and fast and it can be applied to any domain. In this research student qualitative data has been taken from educational data mining and the performance analysis of the decision tree algorithm ID3, C4.5 and CART are compared. The comparison result shows that the Gini Index of CART influence information Gain Ratio of ID3 and C4.5. The classification accuracy of CART is higher when compared to ID3 and C4.5. However the difference in classification accuracy between the decision tree algorithms is not considerably higher. The experimental results of decision tree indicate that student's performance also influenced by qualitative factors.