K-means Classification Algorithm Research Articles

A deep learning model based glaucoma detection using retinal images

The retinal illness that causes vision loss frequently on the globe is glaucoma. Hence, the earlier detection of Glaucoma is important. In this article, modified AlexNet deep leaning model is proposed to category the source retinal images into either healthy or Glaucoma through the detection and segmentations of optic disc (OD) and optic cup (OC) regions in retinal pictures. The retinal images are preprocessed and OD region is detected and segmented using circulatory filter. Further, OC regions are detected and segmented using K-means classification algorithm. Then, the segmented OD and OC region are classified and trained by the suggested AlexNet deep leaning model. This model classifies the source retinal image into either healthy or Glaucoma. Finally, performance measures have been estimated in relation to ground truth pictures in regards to accuracy, specificity and sensitivity. These performance measures are contrasted with the other previous Glaucoma detection techniques on publicly accessible retinal image datasets HRF and RIGA. The suggested technique as described in this work achieves 91.6% GDR for mild case and also achieves 100% GDR for severe case on HRF dataset. The suggested method as described in this work achieves 97.7% GDR for mild case and also achieves 100% GDR for severe case on RIGA dataset. AIM: Segmenting the OD and OC areas and classifying the source retinal picture as either healthy or glaucoma-affected. METHODS: The retinal images are preprocessed and OD region is detected and segmented using circulatory filter. Further, OC region is detected and segmented using K-means classification algorithm. Then, the segmented OD and OC region classified are and trained by the suggested AlexNet deep leaning model. RESULTS: The suggested method as described in this work achieves 91.6% GDR for mild case and also achieves 100% GDR for severe case on HRF dataset. The suggested method as described in this work achieves 97.7% GDR for mild case and also achieves 100% GDR for severe case on RIGA dataset. CONCLUSION: This article proposes the modified AlexNet deep learning models for the detections of Glaucoma utilizing retinal images. The OD region is detected using circulatory filter and OC region is detected using k-means classification algorithm. The detected OD and OC regions are utilized to classify the retinal images into either healthy or Glaucoma using the suggested AlexNet model. The proposed method obtains 100% Sey, 93.7% Spy and 96.6% CA on HRF dataset retinal images. The proposed AlexNet method obtains 97.7% Sey, 98% Spy and 97.8% CA on RIGA dataset retinal images. The proposed method stated in this article achieves 91.6% GDR for mild case and also achieves 100% GDR for severe case on HRF dataset. The suggested method as described in this work achieves 97.7% GDR for mild case and also achieves 100% GDR for severe case on RIGA dataset.

Machine Learning Approach for an Adaptive E-Learning System Based on Kolb Learning Styles

In order to effectively implement adaptive learning within E-learning systems, it is crucial to accurately define thelearner's profile that reflects the characteristics necessary for optimal learning. Traditional methods of identifying profiles often relyon questionnaires to collect data from learners, which can be time-consuming and result in irrelevant data due to arbitrary responses.As a solution, we propose an intelligent and dynamic model for adaptive learning that takes into account the entire learning process,from diagnostic assessment to knowledge assimilation. Our approach utilizes the k-means classification algorithm to group learners based on similar characteristics, as defined by the KOLB model. To enhance the accuracy of our model, we also incorporate neural networks to automatically predict learning styles and using decision tree to propose a adaptative pedagogical content to learner. By doing so, we aim to improve the overall performance of our proposed model.

Detection of crack defects in carbon fiber composites using passive infrared thermography

Aiming at the problem of low detection accuracy in detection of damage location, This article proposes a method to reduce the local jitter in passive infrared thermal imaging by using wavelet hierarchic hard threshold denoising based on template convolution, and thresholds are defined using an unsupervised Kmeans classification algorithm, the traditional SAD template convolution matching algorithm and the Sobel operator edge detection are improved. The denoizing algorithm achieves improved the denoising strength, the smoothness, and effectively extracts the crack defect features. To increase the effectiveness of nondestructive examination, the rapid detection method of temperature data matching is utilized before the infrared detection experiment, and the damage position was marked on the normal sample to allow for intuitive assessment of the damage location. The results show that by matching temperature data, the damage degree of the specimen can be quickly determined, and that the damage features can be accurately extracted using improved image processing algorithm.

Evaluation of driver demand for in-vehicle information: An integrated method combining clustering and multivariate ordered probit model

Introduction: The proper execution of driving tasks requires information support. While new technologies have increased the convenience of information access, they have also increased the risk of driver distraction and information overload. Meeting drivers’ demands and providing them with adequate information are crucial to driving safety. Methods: Based on a sample of 1,060 questionnaires, research on driving information demands is conducted from the perspective of drivers. A principal component analysis and the entropy method are integrated to quantify the driving information demands and preferences of drivers. The K-means classification algorithm is selected to classify the different types of driving information demands, including dynamic traffic information demands (DTIDs), static traffic information demands (STIDs), automotive driving status information demands (ATIDs), and total driving information demands (TDIDs). Fisher's least significant difference (LSD) is used to compare the differences in the numbers of self-reported crashes among different driving information demand levels. A multivariate ordered probit model is established to explore the potential factors that influence the different types of driving information demand levels. Results: The DTID is the driver's most in-demand information type, and accordingly, gender, driving experience, average driving mileage, driving skills, and driving style significantly affect the driving information demand levels. Moreover, the number of self-reported crashes decreased as the DTID, ATID, and TDID levels decreased. Conclusion: Driving information demands are affected by a variety of factors. This study also provides evidence that drivers who have higher driving information demands are more likely to drive more carefully and safely than their counterparts who do not exhibit high driving information demands. Practical implications: The results are indicative of the driver-oriented design of in-vehicle information systems and the development of dynamic information services as a way to avoid negative impacts on driving.

Fast Treetops Counting Using Mathematical Image Symmetry, Segmentation, and Fast k-Means Classification Algorithms

Forests are important because they contribute to biodiversity, regulation of carbon dioxide, protection of hydrographic basins, wood production. This article presents a methodology for fast and effective counting of treetops using the mathematical symmetry of the grayscale image. For the treetop counting, the unsupervised k-means classification Algorithm was used with two groups or centroids: treetop and not-treetop. By using these groups and the mathematical symmetry of the image, a fast k-means classification Algorithm is generated. To solve the problem of treetop overlapping and perform a more accurate counting, the watershed Algorithm was used. This methodology has a mean treetop count accuracy of 98.3% with a confidence level of 99% in the interval (97.31, 99.7). Aerial images of the coniferous forest of Alcudia, Mallorca, Spain were used. Forests attenuate climatic changes originated by global warming. Drastic climatic changes cause catastrophes to humanity. This research would help the automatic, massive and recurring counting of treetops with the aim of obtaining forest inventories in order to take care of forests.

IoT-Fog-Cloud Centric Earthquake Monitoring and Prediction

Earthquakes are among the most inevitable natural catastrophes. The uncertainty about the severity of the earthquake has a profound effect on the burden of disaster and causes massive economic and societal losses. Although unpredictable, it can be expected to ameliorate damage and fatalities, such as monitoring and predicting earthquakes using the Internet of Things (IoT). With the resurgence of the IoT, an emerging innovative approach is to integrate IoT technology with Fog and Cloud Computing to augment the effectiveness and accuracy of earthquake monitoring and prediction. In this study, the integrated IoT-Fog-Cloud layered framework is proposed to predict earthquakes using seismic signal information. The proposed model is composed of three layers: (i) at sensor layer, seismic data are acquired, (ii) fog layer incorporates pre-processing, feature extraction using fast Walsh–Hadamard transform (FWHT), selection of relevant features by applying High Order Spectral Analysis (HOSA) to FWHT coefficients, and seismic event classification by K-means accompanied by real-time alert generation, (iii) at cloud layer, an artificial neural network (ANN) is employed to forecast the magnitude of an earthquake. For performance evaluation, K-means classification algorithm is collated with other well-known classification algorithms from the perspective of accuracy and execution duration. Implementation statistics indicate that with chosen HOS features, we have been able to attain high accuracy, precision, specificity, and sensitivity values of 93.30%, 96.65%, 90.54%, and 92.75%, respectively. In addition, the ANN provides an average correct magnitude prediction of 75%. The findings ensured that the proposed framework has the potency to classify seismic signals and predict earthquakes and could therefore further enhance the detection of seismic activities. Moreover, the generation of real-time alerts further amplifies the effectiveness of the proposed model and makes it more real-time compatible.

Spatial variability of soil quality within management zones: Homogeneity and purity of delineated zones

Fields are the original management zones used in agricultural ecosystems. Uniformity of soil within management zones (MZ) is crucial for sustainable soil management, long-term productivity, and avoiding environmental problems. When considering a new area for agricultural expansion or for improving the efficiency of existing agricultural practices, it is useful to identify homogeneous areas or MZs so that the land can be more sustainably used in the future. One way to identify MZs could be through soil quality assessment. Management zones were determined for an agroecosystem region in southern Iran with an area of 452 km2, and the homogeneity and purity of delineated zones were examined by soil quality assessment. Soil quality grades were calculated using 421 top-soil samples and two methods: i) the total data set (TDS) and ii) the minimum data set (MDS). The spatial distribution of soil quality grades was mapped using a random forest model. MZs were delineated using a fuzzy k-means classification algorithm based on the MDS. The random forest model mapped the spatial distribution of the soil quality well (R2 > 0.871). Among five soil quality grades, three soil quality grades, high (II), moderate (III), and low (IV), were found to cover 90.74 and 93.11% of the total studied area as predicted by the TDS and MDS, respectively. The subsequent classification of the soil quality data into MZs using fuzzy k-means identified two different MZs (p < 0.05). This means that there were heterogeneous soil quality grades in each of the MZs. Consequently, when fuzzy k-means is used to define MZs for the classification of agricultural ecosystems, areas with different soil quality may occur within each MZ. Soil management would be theoretically better if soil quality within management zones were homogenous. This study offers a framework to investigate the homogeneity of delineated MZs in terms of soil quality.

Placement and Quantitating of FACTS Devices in a Power System Including the Wind Unit to Enhance System Parameters

One of the main concerns of network operators is the enhancement of system parameters; accordingly, a set of different means to this end are posed. However, the use of renewable energies such as the wind could increase the importance of the debate over sustainability and conditions of power system parameters. In this study, the condition of said parameters is examined by placing FACTS (Flexible Alternating Current Transmission System) devices in a 24-bus power system including a wind farm. Research data entailing information on the wind and the amount of consumption load per year are classified by using the K-means classification algorithm; then, the objective function is obtained according to the parameters intended for optimization. This function is optimized by using the Honey-bee mating optimization (HBMO) algorithm followed by obtaining the suitable place and amount for FACTS devices. The results showed that the examined parameters are optimized when using FACTS devices.

Cancer genomics predicts disease relapse and therapeutic response to neoadjuvant chemotherapy of hormone sensitive breast cancers

Several studies provide insight into the landscape of breast cancer genomics with the genomic characterization of tumors offering exceptional opportunities in defining therapies tailored to the patient’s specific need. However, translating genomic data into personalized treatment regimens has been hampered partly due to uncertainties in deviating from guideline based clinical protocols. Here we report a genomic approach to predict favorable outcome to treatment responses thus enabling personalized medicine in the selection of specific treatment regimens. The genomic data were divided into a training set of N = 835 cases and a validation set consisting of 1315 hormone sensitive, 634 triple negative breast cancer (TNBC) and 1365 breast cancer patients with information on neoadjuvant chemotherapy responses. Patients were selected by the following criteria: estrogen receptor (ER) status, lymph node invasion, recurrence free survival. The k-means classification algorithm delineated clusters with low- and high- expression of genes related to recurrence of disease; a multivariate Cox’s proportional hazard model defined recurrence risk for disease. Classifier genes were validated by Immunohistochemistry (IHC) using tissue microarray sections containing both normal and cancerous tissues and by evaluating findings deposited in the human protein atlas repository. Based on the leave-on-out cross validation procedure of 4 independent data sets we identified 51-genes associated with disease relapse and selected 10, i.e. TOP2A, AURKA, CKS2, CCNB2, CDK1 SLC19A1, E2F8, E2F1, PRC1, KIF11 for in depth validation. Expression of the mechanistically linked disease regulated genes significantly correlated with recurrence free survival among ER-positive and triple negative breast cancer patients and was independent of age, tumor size, histological grade and node status. Importantly, the classifier genes predicted pathological complete responses to neoadjuvant chemotherapy (P < 0.001) with high expression of these genes being associated with an improved therapeutic response toward two different anthracycline-taxane regimens; thus, highlighting the prospective for precision medicine. Our study demonstrates the potential of classifier genes to predict risk for disease relapse and treatment response to chemotherapies. The classifier genes enable rational selection of patients who benefit best from a given chemotherapy thus providing the best possible care. The findings encourage independent clinical validation.

Habitat choices of African buffalo (Syncerus caffer) and plains zebra (Equus quagga) in a heterogeneous protected area

Abstract ContextAn understanding of large herbivore habitat choices in heterogeneous African protected areas is important for the better management of these key ecosystems. AimsTo determine habitat use of African buffalo (Syncerus caffer) and plains zebra (Equus quagga) in a heterogeneous protected area. MethodsZambezi National Park (ZNP), Zimbabwe, was divided into five vegetation types using an unsupervised classification on a Landsat satellite image that was classified into five land cover classes, using the K-means classification algorithm. African buffalo and plains zebra densities were then determined in each vegetation type using road transect surveys monthly between January 2013 and December 2015. Normalised difference vegetation index (NDVI), grass biomass, grass height and grass quality (nitrogen, calcium, phosphorus and acid detergent fibre content) were determined in each vegetation type during the wet (November to April) and dry (August to October) seasons to establish their quality as habitats for African buffalo and plains zebra. Key resultsBoth African buffalo and plains zebra mostly foraged in mixed and grassland areas, and avoided Zambezi teak vegetation type. Zambezi teak vegetation type had high NDVI due to the dense tree cover. Both African buffalo and plains zebra preferred vegetation types with intermediate grass biomass (approximately 300gm−2) and grass height (approximately 16cm). Grass nutritive value (in terms of nitrogen, phosphorus and acid detergent fibre) declined from wet to dry season in all vegetation types. ConclusionsAfrican buffalo and plains zebra in the ZNP confined their habitat use mostly to two vegetation types (mixed and grassland), which together covered 25% of the protected area. ImplicationsTeak (Baikiaea plurijuga) vegetation, which accounted for about 60% of the ZNP, was avoided by both African buffalo and plains zebra, suggesting that a significant part of the protected area was not used by the two herbivores.

Prediction of the Corn Grains Yield through Artificial Intelligence

Abstract Currently, the determination of the quality of the cereals is done manually by grain classifier experts prior to the marketing stage. In this paper we present a web software tool that allows determining the quality level of a corn sample automatically from an image of it. Image processing algorithms were implemented to correct distortions caused mainly by the capture process. The K-Means classification algorithm was used and a function was developed to calculate the hectolitre weight in relation to the sample area. The results obtained by the application for grades 1 and 2, are close to those measured by the experts. However, those for grade 3 have not been similar since the subsamples selected were not representative.

Wind Generation Impact in Transmission Expansion Planning

This research proposes a post-stage investment analysis in the transmission expansion network planning problem considering the intermittence of renewable energy sources, especially wind power sources. The main goal of this study is to analyze the voltage profiles in all the system buses and the level of electrical losses in the lines using AC flow for impact evaluation of renewable energy insertion. This research also performs an additional analysis considering wind energy insertion during the planning stage. This analysis is to demonstrate the effect in the stage of investment in transmission lines. The variability of these sources is represented through scenarios obtained by the K-means classification algorithm that allows the preservation of the correlation among generating stations. The methodology performance is tested using the IEEE 24-bus system, which is modified to include a significant share of wind energy.

Detecting malicious nodes via gradient descent and support vector machine in Internet of Things

IoT devices have become much popular in our daily lives, while attackers often invade network nodes to launch various attacks. In this work, we focus on the detection of insider attacks in IoT networks. Most existing algorithms calculate the reputation of all nodes based on the routing path. However, they rely heavily on the assumption that different nodes in the same routing path have equal reputation, which may be not invalid in practice and cause inaccurate detection results. To solve this issue, we formulate it as a multivariate multiple linear regression problem and use the K-means classification algorithm to detect malicious nodes. Further, we optimize the routing path and design an enhanced detection scheme. Our results indicate that our proposed methods could achieve a detection accuracy rate of 90% or above in a common case, and the enhanced scheme could reach an even lower false detection rate, i.e., below 5%.

Delineating vineyard zones by fuzzy K-means algorithm based on grape sampling variables

This study describes a method for delineating management zones using interpolated maps of grape characteristics recorded in 2013 and 2014 in a Godello vineyard located in the Bierzo Denomination of Origin (León, Northwest Spain). Ten variables were analyzed and recorded for the sampled vines (50 vines/ha). Interpolated maps reflecting each variable and year were created by spatial interpolation (kriging) from the sampled points. Principal component analysis was used to detect relationships between variables and to select the variables to be used to create the cluster classification. Using the fuzzy k-means classification algorithm implemented in the Management Zone Analyst (MZA v.1.0.0) software, several zones were delineated by combining the studied variables. The results delineated 2 different management areas composed of 3 zones each based on winery objectives: (1) to increase grape production (combining the yield for 2013 and 2014); and (2) to improve grape composition (combining the pH for 2013 and 2014).

Data Distribution Aware Classification Algorithm based on K-Means

Giving data driven decisions based on precise data analysis is widely required by different businesses. For this purpose many different data mining strategies exist. Nevertheless, existing strategies need attention by researchers so that they can be adapted to the modern data analysis needs. One of the popular algorithms is K-Means. This paper proposes a novel improvement to the classical K-Means classification algorithm. It is known that data characteristics like data distribution, high-dimensionality, the size, the sparseness of the data, etc. have a great impact on the success of the K-Means clustering, which directly affects the accuracy of classification. In this study, the K-Means algorithm was modified to remedy the algorithm’s classification accuracy degradation, which is observed when the data distribution is not suitable to be clustered by data centroids, where each centroid is represented by a single mean. Specifically, this paper proposes to intelligently include the effect of variance based on the detected data distribution nature of the data. To see the performance improvement of the proposed method, several experiments were carried out using different real datasets. The presented results, which are achieved after extensive experiments, prove that the proposed algorithm improves the classification accuracy of KMeans. The achieved performance was also compared against several recent classification studies which are based on different classification schemes.

Probabilistic relationships between wind and surface water circulation patterns in the SE Bay of Biscay

Non-linear K-means classification algorithm was used to obtain a comprehensive description of the winds and high-frequency radar-derived currents in the SE Bay of Biscay (study area), taking into account a wide range of scales, from several days to interannual variability. The results in the study area show that in summer, a stronger variability in winds and surface currents can be expected, while in winter, intense southwesterly winds and a cyclonic circulation prevail. In addition to the seasonal component of the currents, a significant spatial variability in terms of current patterns and a temporal variability at shorter and interannual scales were also identified, highlighting the complexity of the surface current dynamics. Moreover, the probabilistic relationships between wind and current patterns were explored, obtaining conditional probabilities. Most of the surface current patterns are clearly related to specific wind patterns that are recurrent in the study area. However, other common current patterns are not so clearly related to specific wind conditions. The presence of a seasonal slope current (Iberian Poleward Current, IPC) is one of the most relevant features of the local circulation. An IPC occurrence time series based on Sea Surface Temperature satellite imagery was used to obtain conditional probabilities with the high-frequency radar surface current patterns, showing a relation between the strongest IPC events and closed cyclonic currents, which are not linked to specific winds.

Surface texture analysis of a high-resolution DTM: Interpreting an alpine basin

This study focuses on a geomorphometric analysis of a small headwater basin of the Dolomites (Eastern Italian Alps). The analysis of the fine-scale surface morphology was performed on a high-resolution digital elevation model (cell size 2m) derived from an airborne LiDAR survey. The topography was synthesized by a set of spatial-statistical indexes representing a signature of the local surface morphology, and the relationships between the geomorphometric indexes and morphological features were evaluated. The studied basin was then segmented by means of a fuzzy k-means classification algorithm into different spatial subdomains characterized by specific local geomorphic signatures. This method allows us to objectively choose the optimal combination of the number of classes and degree of fuzziness characterizing the studied basin. Accordingly, the analysis can be adopted to the characteristics of the studied area, improving the interpretation of surface morphology and offering a valuable tool for comparative analysis between different geomorphological contexts. The results of the classification were compared with the actual surface morphology based on high-resolution orthophotos, aerial photos, and field surveys.

Early-season crop area estimates for winter crops in NE Australia using MODIS satellite imagery

To date, industry and crop forecasters have had a good idea of the potential crop yield for a specific season, but early-season information on crop area for a shire or region has been mostly unavailable. The question of “how early and with what accuracy?” area estimates can be determined using multi-temporal Moderate Resolution Imaging Spectroradiometer (MODIS) enhanced vegetation index (EVI) imagery was investigated in this paper. The study was conducted for two shires in Queensland, Australia for the 2003 and 2004 seasons, and focused on deriving total winter crop area estimates (including wheat, barley and chickpea). A simple metric ( Δ E ), which measures the green-up rate of the crop canopy, was derived. Using the unsupervised k-means classification algorithm, the accumulated difference of two consecutive images (one month apart) for three EVI threshold cut-offs ( Δ E i , where i = 250 , 500 and 750) at monthly intervals from April to October was calculated. July showed the highest pixel accuracy with percent correctly classified for all thresholds of 94% and 98% for 2003 and 2004, respectively. The differences in accuracy between the three cut-offs were minimal and the T500 threshold was selected as the preferred cut-off to avoid measuring too small or too large fluctuations in the differential EVI values. When compared to the aggregated shire data (surveyed) on crop area across shires and seasons, average percent differences for the Δ E T 500 for July and August ranged from −19% to 9%. To capture most of the variability in green-up within a region, the average Δ E T 500 of July and August was used for the early-season prediction of total winter crop area estimates. This resulted in high accuracy (R 2=0.96; RMSE = 3157 ha) for predicting the total winter crop from 2000 to 2004 across both shires. This result indicated that this simple multi-temporal remote sensing approach could be used with confidence in early-season crop area prediction at least one to two months ahead of anthesis.

Self-configurable fault monitoring in ad-hoc networks

While fault monitoring is a well-known issue in fixed wired networks, ad-hoc networks provide new challenges with respect to this issue: in particular, the monitoring task may be hindered at any time by the impossibility to observe a given node. This paper proposes a fault monitoring approach for ad-hoc networks which takes into account this constraint. Our approach is based on an information theory measure suitable to the intermittence of ad-hoc nodes and capable to detect network failures by inference. We define a distributed monitoring scheme with several collaborative detection methods, and we detail a self-configuration mechanism based on the K-means classification algorithm. We show how this solution can be integrated into a management architecture and evaluate its performance based an extensive set of simulations.

Robust classification of animal tracking data

This paper describes an application of the K-means classification algorithm to categorize animal tracking data into various classes of behavior. It was found that, even without explicit consideration of biological factors, the clustering algorithm repeatably resolved tracking data from cows into two groups corresponding to active and inactive periods. Furthermore, it is shown that this classification is robust to a large range of data sampling intervals. An adaptive data sampling algorithm is suggested for improving the efficiency of both energy and memory usage in animal tracking equipment.