Kernel Fuzzy C-means Research Articles

Fusion Kernel Fuzzy C-Means Clustering and Improved Distance Regularized Level Set Evolution Model of Thyroid Nodules Segmentation

Fusion Kernel Fuzzy <i>C</i>-Means Clustering and Improved Distance Regularized Level Set Evolution Model of Thyroid Nodules Segmentation

Hybrid Kernelized Fuzzy Clustering and Multiple Attributes Decision Analysis for Corporate Risk Management

This study introduces an emerging risk management architecture by extending balanced scorecards (BSC) with risk exposure considerations for corporate operating performance assessment and then constructs a hybrid mechanism that combines kernelized fuzzy C-means (KFCM), multiple attributes decision analysis (MADA), and extreme learning machine (ELM) for corporate operating performance forecasting. KFCM is implemented to do the clustering task for each corporate under each aspect of BSC. No specific corporate reaches optimal performance under each assessing measure—that is, dissimilar assessing criteria leads to dissimilar outcomes. This method can be transformed into a MADA task and a MADA algorithm that can yield a reliable outcome systematically. Sequentially, the outcome is fed into ELM to construct the performance forecasting mechanism. The introduced mechanism with outstanding forecasting performance comes with a critical challenge: it lacks interpretability, which impedes its real-life usage. To cope with this problem, the rough set theory (RST) is employed to extract the inherent decision logics from the black-box model and visualize it in human readable formats. The introduced model has been examined by real cases and is a promising alternative for corporate risk management.

Markov Prediction based on Semi-supervised Kernel Fuzzy Clustering

This paper proposes an improved Markov prediction, according to the temperature control problem of hot blast stove alternate supply air in the operation of blast furnace operation, namely, implement clustering for waiting to be processed data, used kernel fuzzy c-means clustering based on the pairwise constraints. The supply air temperature of hot blast stove is seen as without aftereffect things in this method, introduce semi-supervised learning mechanism in traditional fuzzy clustering to deal with the basic data, and using the kernel effectiveness index improved the FCM algorithm. Experiments show that the improved clustering algorithm is superior to other algorithms in accuracy and performance, at the same time, the improved prediction model comparison with the traditional values of temperature prediction, which has obvious advantages in defined temperature range and the fit of the temperature value, the guiding significance was significantly enhanced in industrial field.

Novel authentication procedures for preventing unauthorized access in social networks

Online Social Network (OSN) is the most widely used platform to communicate with one another. It is fast and cheap when compared to other modes of communication. However, Online Social Network profiles are being hacked often by employing various types of malicious attacks. Most of the times the attackers are successful in accessing the OSN account which illustrates that the present authentication mechanisms are not efficient enough. Once an online social network account is being hacked, it can be misused widely and the hacker makes the authorized user unable to login to the account by changing the login credentials. Security in OSN must be provided and it must prevent the attacks aimed by the attackers. However, to prevent unauthorized access in OSN a novel authentication procedure is needed and the novel authentication mechanism must abide the social network platform’s characteristics. This paper considers the attackers who aim to access the OSN account in an unauthorized manner and presents two novel authentication mechanisms to prevent the OSN platform. The proposed procedures make use of users’ chatting module statistics and relationship circle statistics to authenticate the users in Online Social Networks. This can be done by Multi kernel Fuzzy C-Means (MKFCM) Algorithm and Group Search Optimizer (GSO). These novel authentication procedures abide the OSN characteristics, as it makes use of Chatting and Relationship module.

Effective kernel FCM: Finding appropriate structure in cancer database

Finding available subclasses in high-dimensional medical databases using clustering techniques is considered as very important one in medical field. Due to similar intensities between the datapoints in high-dimensionality cancer medical database clustering techniques have failed to cluster the available subclasses with less error. Therefore this paper presents suitable fuzzy-based clustering techniques to find available subclasses in high-dimensional prostate and breast cancer databases. In addition this paper presents prototype initialization algorithm to avoid random initialization of initial prototypes. In order to evaluate the performance of proposed clustering techniques experimental study has been performed on benchmark databases. Finally the proposed methods have been successfully implemented to find the subclasses of cancers in prostate and breast cancer databases. The clustering results of proposed methods have been validated by evaluating clustering accuracy.

Identification of flight state under different simulator modes using improved diffusion maps

To identify the difference between dynamic and static simulator modes, a novel data analyzing method was presented in this paper using flight data sampled from manual flight task. The proposed method combined diffusion maps and kernel fuzzy c-means algorithm (KFCM) to identify types of flight data. Hybrid bacterial foraging (BF) and particle swarm optimization (PSO) algorithm (BF-PSO) was also introduced to optimize unknown parameters of the KFCM. This algorithm increased the possibility to find the optimal values avoided being trapped in local minima. The clustering accuracy of the proposed method applied in flight dataset demonstrated this method had the ability to recognize the types of flight state. The results of the paper indicated that the pilots movement sensing influenced pilot performance under the manual departure task.

Hybrid local diffusion maps and improved cuckoo search algorithm for multiclass dataset analysis

Data clustering is a meaningful tool that can, help people classify mixed data automatically. With rapid technological development, data in modern applications become large scale and high dimensional. Some original clustering methods are not suitable for complicated datasets. To improve the performance of the popular kernel fuzzy C-means (KFCM), this study proposed a local density adaptive diffusion maps (LDM) technique to obtain a reliable similarity description and dimensionality reduction. To find the valid cluster centroids of the dataset, this study also proposed an improved cuckoo search (ICS) to optimize the unknown parameters of the KFCM model. The ICS algorithm utilized quaternions to represent individuals who will be optimized. Variable step length of Lévy flights and discovery probability were also proposed, which were adjusted by the evolutional ratio of the cuckoo search process. To verify the availability of the ICS, 5 benchmark functions were tested. Finally, the proposed hybrid ICS and LDM based on KFCM (ICS-LDM-KFCM) was used to identify 4 standard artificial and 6 real world datasets. Compared with other clustering methods, the proposed method obtained more accurate results. This method is verified to be more suitable for complicated datasets with large number of attributes and clusters.

Kernelized Fuzzy C-means Clustering with Adaptive Thresholding for Segmenting Liver Tumors

Accurate liver segmentation from computed tomography (CT) scan images is an essential and crucial step in computer-aided diagnosis of liver tumor. The automatic and unsupervised methods are used to segment CT image that reduce the time and eliminate the need for manual interface. In this paper, the adaptive threshold, morphological processing, and kernel fuzzy C-means (KFCM) clustering algorithm have been used with spatial information for visualizing and measuring the tumor area of liver which is segmented from CT abdominal images. The proposed process was tested and evaluated to a series of test images on MICCAI 2008 liver tumor segmentation challenge datasets to extract the tumor region. The area of the tumor reason has been calculated also. The KFCM algorithm introduces a kernel function on fuzzy c-means clustering (FCM) to reduce the effect of noise and improves the ability of clustering. Experimental results showed positive results for the proposed algorithm. The PSNR and MSE have been calculated for FCM and KFCM method to compare the results to segment the images. An experimental result shows that the proposed method could be used effectively to segment the liver cancer region in terms of high PSNR and low MSE values and achieves high consistency to detect small changes in images.

An Analysis of Distributed Fuzzy K-Prototype Clustering Techniques over Peer to Peer Network Environment

The challenging problems in recent years is the Distributed Data clustering over Peer-to-Peer network (P2P). In this paper, there is a presentation of peer-to-peer based K-prototype framework named MEKPFCM, by incorporating kernel fuzzy c-means and possibilistic fuzzy clustering algorithms. This work also considers the distributed clustering problem where the data and computing resources are spread over the large Peer-to-Peer network. It offers two algorithms which produce an approximate result produced by the MEKPFCM clustering algorithm. In the dynamic P2P network, the distributed fuzzy k-prototype clustering algorithm is played a major role. This can produce clustering's by local synchronization only. The second algorithm uses sampled peers that are reliable and present analytical guarantees regarding the accuracy of clustering on a P2P network. The performance of the proposed method evaluation is done through accuracy and computation time and the experimentation shows that the proposed method gives good performance compared to the existing method.

Optimal Feature Selection for classification Using Rough Set Based CGA-NN Classifier

The overall process of the classification method is divided into two main steps, such as: 1) feature selection using RS-MKFCM algorithm; 2) classification using CGA-based neural network classifier. At first, the multiple kernel fuzzy c-means clustering with rough set theory (RS-MKFCM) algorithm is applied on the high dimensional micro array dataset to select the important features. After that, the classification is done through CGA-NN classifier. Selected features from micro array dataset are collected and fed to the neural network for training. In neural network, we utilise scaled conjugate gradient algorithm for training. It provides faster training with excellent test efficiency. To improve the classification performance, hybridisation of cuckoo search and genetic algorithm (CGA) is utilised with neural network for weight optimisation process. At last, the experimentation is performed by means of the five different micro array dataset. The experimentation result proves that the CGA-NN classifier outperformed the existing approach by attaining the maximum accuracy of 98.93% for ovarian cancer dataset when compared existing NN only achieved 51.87% and also support vector machine classifier achieves the 88.23%.

Smartwatch-Based Driver Vigilance Indicator With Kernel-Fuzzy-C-Means-Wavelet Method

A high-precision driver vigilance predictor could be a monetary countermeasure to reduce road accidents. Heart rate variability is a well-known measurement parameter to predict driver vigilance state, but the measurement is susceptible to motion artifact due to body movement where the electrocardiogram (ECG) sensor device had to be worn close to the heart. Thus, this paper presents a novel approach to measure the ECG from the driver palms while holding on the steering wheel. In addition, photoplethysmograms sensor attached on a driver finger can also measure the similar heart rate pattern, known as pulse rate variability. Another significant vigilance measurement parameter, respiratory rate variability, can be derived directly from the ECG with the squaring baseline method, without the usage of respiratory sensor. Furthermore, this paper is also focusing on the integration of age and gender as vigilance measurement parameter as each individual exhibits distinct signal pattern. Autonomous rules are derived from the data set that performs the kernel fuzzy c-means with if-then rules extraction, which subsequently classify the driver vigilance level into two predefined classes, that are drowsy and awake. The vigilance monitoring application is developed in smartwatch, able to perform the features extraction, and then predict the driver vigilance class based on the Kernel Fuzzy-C-Mean trained model. A vibration warning will be triggered to the driver if the driver is estimated as drowsy in five consecutive time frames. In fact, the experimental results stated that the prediction accuracy can be achieved at 97.28% on average across variant subjects.

Optimal feature selection for classification using rough set-based CGA-NN classifier

The overall process of the classification method is divided into two main steps, such as: 1) feature selection using RS-MKFCM algorithm; 2) classification using CGA-based neural network classifier. At first, the multiple kernel fuzzy c-means clustering with rough set theory (RS-MKFCM) algorithm is applied on the high dimensional micro array dataset to select the important features. After that, the classification is done through CGA-NN classifier. Selected features from micro array dataset are collected and fed to the neural network for training. In neural network, we utilise scaled conjugate gradient algorithm for training. It provides faster training with excellent test efficiency. To improve the classification performance, hybridisation of cuckoo search and genetic algorithm (CGA) is utilised with neural network for weight optimisation process. At last, the experimentation is performed by means of the five different micro array dataset. The experimentation result proves that the CGA-NN classifier outperformed the existing approach by attaining the maximum accuracy of 98.93% for ovarian cancer dataset when compared existing NN only achieved 51.87% and also support vector machine classifier achieves the 88.23%.

A Novel Multiple Fuzzy Clustering Method Based on Internal Clustering Validation Measures with Gradient Descent

In this paper, we propose a novel multiple fuzzy clustering method based on internal clustering validation measures with gradient descent. Firstly, some single fuzzy clustering algorithms such as Fuzzy C-Means, Kernel Fuzzy C-Means and Gustafson–Kessel are used to construct similarity matrixes for each partition. Secondly, those similarity matrixes are aggregated into a final one by means of the direct sum of weighted vectors where the values of weights are determined by internal clustering validation measures. Finally, final membership matrix is calculated by minimizing the sum of square errors through the gradient descent method. The proposed approach has been validated in terms of clustering quality on UCI Machine Learning Repository datasets. Experimental results show that the proposed approach’s performance is better than those of other ensemble and standalone methods.

An improved fuzzy C-means clustering algorithm for assisted therapy of chronic bronchitis.

Bronchitis is considered a non-specific inflammation in the peripheral tissues of the trachea and bronchus. Many therapeutic schemes for chronic bronchitis have been reported in existing research. This work attempted to conduct optimization analysis of the therapeutic scheme for chronic bronchitis using a data mining method. To overcome the shortfalls of the current fuzzy C-means clustering (FCM) algorithm, this research proposed an improved kernel fuzzy C-means (KFCM) clustering algorithm. The improved KFCM algorithm solved traditional cluster algorithm problems in two ways: firstly FCM clustering was mapped in high-dimensional kernel space; and the samples in the initial input space R(S) were mapped to high-dimensional feature space R(p). Finally, the improved and traditional algorithms by computer simulation experiments. Based on the analysis of the simulation experiments on IRIS dataset in this research, improved KFCM algorithm could improve calculation accuracy by 10% because the initial value greatly decreased the number of iterations and improved the accuracy of the calculation. The improved KFCM algorithm was used to optimize the relationship between data structures in the process of iteration clustering so as to accelerate iteration convergence. The simulation results show that the improved KFCM algorithm performs better in terms of both calculating performance and clustering correctness.

A novel spectral clustering method with superpixels for image segmentation

Similarity measure is critical to the performance of spectral clustering. The most commonly used similarity measure for spectral clustering is Gaussian kernel similarity measure. However, the selection of accurate scaling parameter in Gaussian kernel function is difficult. To reduce the sensitivity of scaling parameter, in this paper, a novel spectral clustering method with superpixels for image segmentation (SCS) is proposed. In particular, a novel kernel fuzzy similarity measure is presented, which uses membership distribution in partition matrix obtained by kernel fuzzy C-means clustering(KFCM). In addition, the superpixel is introduced into image segmentation to alleviate the computational burden of affinity matrix. The experimental results show that our approach is able to perform steadily under different parameters, and obtain good clustering results on various natural images. Moreover, the evaluation comparisons also indicate that our method can achieve comparable accuracy and significantly outperform most state-of-the-art algorithms.

Leveraging longitudinal driving behaviour data with data mining techniques for driving style analysis

Accurately understanding driving behaviour is of crucial importance for advanced driving assistant systems such as adaptive cruise control system and intelligent forward collision warning system. To understand different driving styles, this study employs the clustering method and topic model to extract latent driving states, which can elaborate and analyse the commonness and individuality of driving behaviour characteristics with the longitudinal driving behaviour data collected by the instrumented vehicle. To handle the large set of data and discover the valuable knowledge, the data mining techniques including ensemble clustering method based on the kernel fuzzy C-means algorithm and the modified latent Dirichlet allocation model are employed in this study. The 'aggressive', 'cautious' and 'moderate' driving states are discovered and the underlying quantified structure is built for the driving style analysis. Language: en

Recognition of radar emitter signals based on SVD and AF main ridge slice

Recognition of radar emitter signals is one of core elements in radar reconnaissance systems. A novel method based on singular value decomposition (SVD) and the main ridge slice of ambiguity function (AF) is presented for attaining a higher correct recognition rate of radar emitter signals in case of low signal-to-noise ratio. This method calculates the AF of the sorted signal and ascertains the main ridge slice envelope. To improve the recognition performance, SVD is employed to eliminate the influence of noise on the main ridge slice envelope. The rotation angle and symmetric Holder coefficients of the main ridge slice envelope are extracted as the elements of the feature vector. And kernel fuzzy c-means clustering is adopted to analyze the feature vector and classify different types of radar signals. Simulation results indicate that the feature vector extracted by the proposed method has satisfactory aggregation within class, separability between classes, and stability. Compared to existing methods, the proposed feature recognition method can achieve a higher correct recognition rate.

Kidney segmentation in CT sequences using SKFCM and improved GrowCut algorithm.

BackgroundOrgan segmentation is an important step in computer-aided diagnosis and pathology detection. Accurate kidney segmentation in abdominal computed tomography (CT) sequences is an essential and crucial task for surgical planning and navigation in kidney tumor ablation. However, kidney segmentation in CT is a substantially challenging work because the intensity values of kidney parenchyma are similar to those of adjacent structures.ResultsIn this paper, a coarse-to-fine method was applied to segment kidney from CT images, which consists two stages including rough segmentation and refined segmentation. The rough segmentation is based on a kernel fuzzy C-means algorithm with spatial information (SKFCM) algorithm and the refined segmentation is implemented with improved GrowCut (IGC) algorithm. The SKFCM algorithm introduces a kernel function and spatial constraint into fuzzy c-means clustering (FCM) algorithm. The IGC algorithm makes good use of the continuity of CT sequences in space which can automatically generate the seed labels and improve the efficiency of segmentation. The experimental results performed on the whole dataset of abdominal CT images have shown that the proposed method is accurate and efficient. The method provides a sensitivity of 95.46% with specificity of 99.82% and performs better than other related methods.ConclusionsOur method achieves high accuracy in kidney segmentation and considerably reduces the time and labor required for contour delineation. In addition, the method can be expanded to 3D segmentation directly without modification.

Performance Comparison of Wavelet and Contourlet Frame Based Features for Improving Classification Accuracy in Remote Sensing Images

Conventional classification algorithms makes the use of only multispectral information in remote sensing image classification. Wavelet provides spatial and spectral characteristics of a pixel along with its neighbours and hence this can be utilized for an improved classification. The major disadvantage of wavelet transform is the non availability of spatial frequency features in its directional components. The contourlet transform based laplacian pyramid followed by directional filter banks is an efficient way of extracting features in the directional components. In this paper different contourlet frame based feature extraction techniques for remote sensing images are proposed. Principal component analysis (PCA) method is used to reduce the number of features. Gaussian Kernel fuzzy C-means classifiers uses these features to improve the classification accuracy. Accuracy assessment based on field visit data and cluster validity measures are used to measure the accuracy of the classified data. The experimental result shows that the overall accuracy is improved to 1.73 % (for LISS-II), 1.81 % (for LISS-III) and 1.95 % (for LISS-IV) and the kappa coefficient is improved to 0.933 (for LISS-II), 0.0103 (for LISS-III) and 0.0214 (for LISS-IV) and also the cluster validity measures gives better results when compared to existing method

Automatic recognition of weave pattern and repeat for yarn-dyed fabric based on KFCM and IDMF

This paper proposes an automatic recognition method to analyze the weave pattern and repeat of yarn-dyed fabrics. Firstly, the warp and weft floats of preprocessing yarn-dyed fabric images with the solid color are segmented through gray projection method. The kernel fuzzy c-means clustering (KFCM) algorithm is utilized to classify the weave points based on the texture features of gray means, gray variances and gray level co-occurrence matrix (GLCM). The exact state of the two floats is judged by comparing average gray means of each cluster. With warp floats (1s) and weft floats (0s), fabric image is represented as binary value weave diagram and coded digital matrix. Then, improved distance matching function (IDMF) is employed to obtain the weave repeat of weave diagram, which is used to correct error floats and improve the accuracy of identification result. Moreover, IDMF is directly applied to yarn-dyed fabrics with different color yarns and obtained the accurate weave repeat with faster speed. The experimental results have shown that the proposed algorithm can recognize weave pattern and repeat accurately and faster, and output the corresponding binary value weave diagram of the identified fabric.