Robust Clustering Technique Research Articles

Robust clustering based on trimming

AbstractClustering is one of the most widely used unsupervised learning techniques. However, it is well‐known that outliers can have a significantly adverse impact on commonly applied clustering methods. On the other hand, clustered outliers can be particularly detrimental to (even robust) statistical procedures. Therefore, it makes sense to combine concepts from Robust Statistics and Cluster Analysis to deal with both clusters and outliers simultaneously through robust clustering approaches. Among the existing robust clustering techniques, we focus on those that rely on (impartial) trimming. Trimming offers the user an easy interpretation, as standard well‐known clustering methods are applied after a fraction of the potentially most outlying observations is removed. This trimming approach, when combined with appropriate constraints on the clusters' dispersion parameters, has shown a good performance and can be implemented efficiently thorough available algorithms.This article is categorized under: Statistical Learning and Exploratory Methods of the Data Sciences > Clustering and Classification Statistical and Graphical Methods of Data Analysis > Robust Methods

Clustering and Financial Performance Analysis of Indonesian Coal Mining Industry Stock Prices

In the comprehensive study of the Indonesian coal mining industry, a rigorous exploration and clustering approach was applied to historical stock price data and financial metrics from coal companies listed for over a decade. Data sourced from Yahoo Finance underwent an automated download process, ensuring consistency and efficiency. The research utilized robust clustering techniques, including K-means, Hierarchical, and Correlation Clustering, to discern the stock price movements during the dynamic market conditions of 2022 and 2023. Financial performance analysis, focusing on key metrics such as ROA, NPM, and EPS, highlighted the unique financial dynamics of companies like PTIS, IATA, and AIMS. The study's results provide a multifaceted understanding of the coal industry's financial trends, emphasizing varied company responses to market conditions and revealing significant financial performance divergences among key players. This research not only offers invaluable insights into the coal industry's financial landscape but also presents an innovative methodology poised to transform financial analytics in the sector.

Semi-Supervised Clustering-Based DANA Algorithm for Data Gathering and Disease Detection in Healthcare Wireless Sensor Networks (WSN).

Wireless sensor networks (WSNs) have emerged as a promising technology in healthcare, enabling continuous patient monitoring and early disease detection. This study introduces an innovative approach to WSN data collection tailored for disease detection through signal processing in healthcare scenarios. The proposed strategy leverages the DANA (data aggregation using neighborhood analysis) algorithm and a semi-supervised clustering-based model to enhance the precision and effectiveness of data collection in healthcare WSNs. The DANA algorithm optimizes energy consumption and prolongs sensor node lifetimes by dynamically adjusting communication routes based on the network's real-time conditions. Additionally, the semi-supervised clustering model utilizes both labeled and unlabeled data to create a more robust and adaptable clustering technique. Through extensive simulations and practical deployments, our experimental assessments demonstrate the remarkable efficacy of the proposed method and model. We conducted a comparative analysis of data collection efficiency, energy utilization, and disease detection accuracy against conventional techniques, revealing significant improvements in data quality, energy efficiency, and rapid disease diagnosis. This combined approach of the DANA algorithm and the semi-supervised clustering-based model offers healthcare WSNs a compelling solution to enhance responsiveness and reliability in disease diagnosis through signal processing. This research contributes to the advancement of healthcare monitoring systems by offering a promising avenue for early diagnosis and improved patient care, ultimately transforming the landscape of healthcare through enhanced signal processing capabilities.

Development of a Computerized Diagnostic System for Brain MRI Tumor Scanning Using a Robust Information Clustering Technique

This paper presents the development of a computerized diagnostic system for brain MRI tumor scanning using a robust information clustering technique. The method used for this study is data collection, data processing, feature extraction, artificial neural network, activation function, training algorithm, and classification. The method was modelled using a structural approach that developed the Artificial Neural Network (ANN) algorithm, using tansig activation function and back-propagation training algorithm. The brain tumor detection algorithm developed was implemented with MATLAB Simulink application, and tested with Mean Square Error (MSE) and Regression (R) analysis. The result showed that the MSE is 0.002488 and the Regression result is 0.9933. The algorithm was also comparatively compared with an existing system and the result showed that the new system achieved better regression performance than the others. Then it was deployed as a clinical decision system for the diagnosis of brain tumors and tested, the result showed that it was able to detect patients with brain MRI data. Keyword : Back-Propagation, Magnetic Resonance Imaging (MRI), Neural Network, Simulink, Tansig.

Improved space breakdown method - A robust clustering technique for spike sorting.

https://github.com/ArdeleanRichard/Space-Breakdown-Method.

A Kemeny Distance-Based Robust Fuzzy Clustering for Preference Data

We propose two robust fuzzy clustering techniques in the context of preference rankings to group judges into homogeneous clusters even in the case of contamination due to outliers or, more generally, noisy data. The two fuzzy C-Medoids clustering methods, based on the same suitable exponential transformation of the Kemeny distance, belong to two different approaches and differ in the way they introduce the fuzziness in the membership matrix, the one based on the “m” exponent and the other on the Shannon entropy. As far as the Kemeny distance is concerned, it is equivalent to the Kendall distance in the case of complete rankings but differs from the latter in the way of handling tied rankings. Simulations prove that our methods are able to recover the natural structure of the groups neutralizing the effect of possible noises and outliers. Two applications to real datasets are also provided.

A fuzzy clustering technique for enhancing the convergence performance by using improved Fuzzy c-means and Particle Swarm Optimization algorithms

Fuzzy clustering is a well-established technique among the well-known clustering techniques in several real-world applications due to easy implementation and produces satisfactory clustering result. However, it has some deficiency such as sensitive to outliers, result dependency on choosing initial centroid, etc. To eradicate the shortcoming of FCM algorithm, this article introduces a robust clustering technique, particle swarm optimization improved fuzzy c-means is developed by the hybridization of particle swarm optimization and improved fuzzy c-means techniques, to deal with noisy data and initialization problem. In this article, a fuzzy clustering technique is developed to increase the convergence performance of clustering techniques. Fuzzy c-means is improved by developing a new metric to tolerate the noisy environment. Particle swarm optimization has an inbuilt guidance strategy which leads the solution in particle swarm optimization to obtain useful information from the better solution and thereby helping them improve their own solution. To handle the initialization problem of fuzzy c-means, particle swarm optimization technique is used. PSO effectively enhance the performance of improved FCM to increase the effectiveness of clustering. The effectiveness of the proposed clustering technique over existing techniques in literature has been illustrated by adopting eight real worlds and three artificial data sets. The results show that the proposed algorithm generates encouraging results as compared to the established clustering technique in literature.

Capturing the multidimensionality of motivation in physical education: A self-organizing maps approach to profiling students

This study aimed to capture the multidimensionality of adolescents’ motivation in the physical education (PE) setting, within self-determination theory, by employing self-organizing maps (SOM) analysis. Particularly, it examined the topological relationships among students’ basic psychological needs satisfaction, their perception of more or less self-determined motivation and the affective and behavioural consequences in PE lessons across several sociodemographic variables. A nationally representative sample of 3029 Spanish students (51% girls), aged 12 to 18 years, was surveyed. SOM mapped well-defined students’ profiles that embraced interrelatedly a considerable number of students’ motivational characteristics. Four target profiles, out of 10, were explored. The first two profiles, highly motivated to be active girls and boys, mainly experienced senses of self-determination, but also controlled reasons for participating in PE lessons, high perceived competence, relatedness and autonomy fulfilment, perceived exerted effort and satisfaction. However, the reluctance to be physically active presented two gendered motivational profiles. Barely motivated to be active girls showed the lowest levels of self-determined motivation, including introjected regulation, low perceptions of competence, autonomy, relatedness, and dissatisfaction in PE. Vaguely motivated to be active boys revealed that despite their perceptions of competence the neglect of the other two psychological needs was more likely to determine a controlled motivation and, consequently, maladaptive outcomes. SOM proved to be a more robust and accurate clustering technique than the k-means algorithm and helped to portray and visualize the complexity behind the decision to become an active person considering the motivational processes in PE. Implications are provided for practitioners.

A Robust Tensor-Based Submodule Clustering for Imaging Data Using l12 Regularization and Simultaneous Noise Recovery via Sparse and Low Rank Decomposition Approach.

The massive generation of data, which includes images and videos, has made data management, analysis, information extraction difficult in recent years. To gather relevant information, this large amount of data needs to be grouped. Real-life data may be noise corrupted during data collection or transmission, and the majority of them are unlabeled, allowing for the use of robust unsupervised clustering techniques. Traditional clustering techniques, which vectorize the images, are unable to keep the geometrical structure of the images. Hence, a robust tensor-based submodule clustering method based on regularization with improved clustering capability is formulated. The induced tensor nuclear norm (TNN), integrated into the proposed method, offers better low rankness while retaining the self-expressiveness property of submodules. Unlike existing methods, the proposed method employs a simultaneous noise removal technique by twisting the lateral image slices of the input data tensor into frontal slices and eliminates the noise content in each image, using the principles of the sparse and low rank decomposition technique. Experiments are carried out over three datasets with varying amounts of sparse, Gaussian and salt and pepper noise. The experimental results demonstrate the superior performance of the proposed method over the existing state-of-the-art methods.

Clustering Accelerometer Activity Patterns from the UK Biobank Cohort.

Many researchers are beginning to adopt the use of wrist-worn accelerometers to objectively measure personal activity levels. Data from these devices are often used to summarise such activity in terms of averages, variances, exceedances, and patterns within a profile. In this study, we report the development of a clustering utilising the whole activity profile. This was achieved using the robust clustering technique of k-medoids applied to an extensive data set of over 90,000 activity profiles, collected as part of the UK Biobank study. We identified nine distinct activity profiles in these data, which captured both the pattern of activity throughout a week and the intensity of the activity: “Active 9 to 5”, “Active”, “Morning Movers”, “Get up and Active”, “Live for the Weekend”, “Moderates”, “Leisurely 9 to 5”, “Sedate” and “Inactive”. These patterns are differentiated by sociodemographic, socioeconomic, and health and circadian rhythm data collected by UK Biobank. The utility of these findings are that they sit alongside existing summary measures of physical activity to provide a way to typify distinct activity patterns that may help to explain other health and morbidity outcomes, e.g., BMI or COVID-19. This research will be returned to the UK Biobank for other researchers to use.

Robust clustering for assessing the spatiotemporal variability of groundwater quantity and quality

The long-term spatiotemporal assessment of groundwater resources through robust clustering techniques can be used to promote remediation measures for groundwater depletion and contamination. To fully understand the variability of groundwater quantity and quality due to anthropogenic activities and climate changes, a new ensemble clustering framework based on the Combining Multiple Clusters via Similarity Graph (COMUSA) method was developed. This new approach was applied and evaluated in the context of groundwater well systems on the Ghorveh-Dehgolan Plain (GDP), which is located in western Iran, for groundwater level (GWL) and 13 physicochemical parameters during four periods (the average of data from 1988–1990, 1997–1999, 2006–2008, and 2015–2017). The classification was confirmed by using the cluster validity index of the silhouette coefficient (SC), which indicated that the cluster ensemble method could improve the performance of individual clustering methods for groundwater quantity and quality by up to 12% and 20%, respectively. Piper plots, US Salinity Laboratory Staff (USSL) diagrams, and the pollution index of groundwater (PIG) were assessed for all clusters of physicochemical variables to analyse groundwater suitability for drinking and irrigation purposes. The results of the cluster ensemble showed that a critical pattern of groundwater depletion occurred in the western half of the GDP, while the eastern part was recognized as the most polluted zone on the plain. It could be concluded that the decline in GWL was not the only reason for the increase in groundwater quality variables, but other factors, such as noticeable cropland expansion and the overuse of chemical fertilizers and pesticides, were also influential factors related to these patterns. Taken together, the results of this study contribute to better recognizing the spatiotemporal changes in groundwater quantity and quality under the intense pressure of anthropogenic activities.

Noise robust image clustering based on reweighted low rank tensor approximation and $$l_{\frac{1}{2}}$$ regularization

In this paper, a noise robust tensor based image clustering approach is proposed which can also perform well in the presence of gross errors. Our major contribution is the improved submodule identification technique in noisy environment by incorporating three important improvements: better low rank representation using reweighted nuclear norm, $$l_\frac{1}{2}$$ regularization to accurately capture sparseness and an error term in the model for noise robustness. Reweighted nuclear norm is introduced in the clustering model to capture self-expressiveness property in a better manner. The $$l_\frac{1}{2}$$ norm regularization is applied in place of $$l_1$$ -norm to properly capture the correlation among data members. An error term is introduced into the model to separate noise and data, which brings a noise robust image clustering technique. Combined effect all the three factors results an accurate clustering method even under the presence of severe noise. The performance of the proposed method is tested on different datasets with varying amount of noise. It is found that the proposed method provides better classification accuracy in almost all conditions as compared with existing methods.

DeepPipes: Learning 3D pipelines reconstruction from point clouds

Pipes are the basic building block in many industrial sites like electricity and chemical plants. Although pipes are merely cylindrical primitives which can be defined by axis and radius, they often consist of additional components like flanges, valves, elbows, tees, etc. 3D pipes are typically dense, consisting of a wide range of topologies and geometries, with large self-occlusions. Thus, reconstruction of a coherent 3D pipe models from large-scale point clouds is a challenging problem. In this work we take a prior-based reconstruction approach which reduces the complexity of the general pipe reconstruction problem into a combination of part detection and model fitting problems. We utilize convolutional network to learn point cloud features and classify points into various classes, then apply robust clustering and graph-based aggregation techniques to compute a coherent pipe model. Our method shows promising results on pipe models with varying complexity and density both in synthetic and real cases.

Detection of Auction Fraud in Commercial Sites

Online auctions have become one of the most convenient ways to commit fraud due to a large amount of money being traded every day. Shill bidding is the predominant form of auction fraud, and it is also the most difficult to detect because it so closely resembles normal bidding behavior. Furthermore, shill bidding does not leave behind any apparent evidence, and it is relatively easy to use to cheat innocent buyers. Our goal is to develop a classification model that is capable of efficiently differentiating between legitimate bidders and shill bidders. For our study, we employ an actual training dataset, but the data are unlabeled. First, we properly label the shill bidding samples by combining a robust hierarchical clustering technique and a semi-automated labeling approach. Since shill bidding datasets are imbalanced, we assess advanced over-sampling, under-sampling and hybrid-sampling methods and compare their performances based on several classification algorithms. The optimal shill bidding classifier displays high detection and low misclassification rates of fraudulent activities.

Distributed robust data clustering in wireless sensor networks using diffusion moth flame optimization

Traditional K-Means based distributed clustering used in wireless sensor networks has limitation of getting stuck into local minima, thus many times results in giving inaccurate cluster partitions. To alleviate this drawback, evolutionary based robust distributed clustering techniques are proposed in this paper. These techniques have the capability to determine the global optima thus results in effective cluster partitioning. A moth flame optimization based method is proposed here which minimizes the intra-cluster distance to determine the optimal partition at every sensor node. A diffusion method of cooperation is employed by sharing the best moth position and corresponding fitness value (intra cluster distance) to the neighboring nodes. To introduce robustness, a weight based method for detection and removal of outliers is employed. In this method a weight based on volume and density is given to each data point for outlier detection, a larger weight is considered as an outlier. The simulation study is carried out on one synthetic and two real datasets. The performance of proposed approach is compared with diffusion particle swarm optimization (DPSO), diffusion whale optimization algorithm (DWOA), diffusion elephant herding optimization (DEHO) and distributed K-Means (DK-Means) in terms of Dunn’s index, Silhouette index and time complexity. The minimum average Euclidean deviation of proposed diffusion moth flame optimization is 7.16%, 3.25%, 5.24% and 21.70% lower compared to DPSO, DWOA, DEHO and DK-Means respectively for cook agronomy farm dataset.

Application of self-organizing map and fuzzy c-mean techniques for rockburst clustering in deep underground projects

One of the main concerns associated with deep underground constructions is the violent expulsion of rock induced by unexpected release of strain energy from surrounding rock masses that is known as rockburst. Rockburst hazard causes substantial damages to the foundation of the structure and equipment and can be a menace to the safety of workers. This study was intended to find the latent relationship between the rockburst-related parameters based on the compiled data samples from deep underground projects using two robust clustering techniques of self-organizing map (SOM) and fuzzy c-mean (FCM). The parameters of maximum tangential stress, uniaxial compressive strength, uniaxial tensile strength, and elastic energy index were considered as input parameters. SOM model could classify data samples into four distinct classes (clusters), and the rockburst intensities were identified precisely. FCM also proved its performance in clustering task with high convergence speed and acceptable accuracy. Having a comparison, the results of SOM and FCM models were compared with ones calculated from five empirical criteria of Russenes, Hoek, tangential stress, elastic energy index, and rock brittleness coefficient. At best, the empirical criteria of Hoek and tangential stress coefficient could predict rockburst intensity with the accuracy of 56.90%. By analyzing the SOM results as the best model, it was turned out that the maximum tangential stress around the openings has a crucial role in rockburst clustering and has the most influence on the occurrence of strong and moderate rockburst types. Hence, it was recommended as a possible solution to control these types of rockbursts by optimizing the diameter and shape of the underground openings.

Artificial intelligence for identifying hydrologically homogeneous regions: A state‐of‐the‐art regional flood frequency analysis

AbstractDue to the severity related to extreme flood events, recent efforts have focused on the development of reliable methods for design flood estimation. Historical streamflow series correspond to the most reliable information source for such estimation; however, they have temporal and spatial limitations that may be minimized by means of regional flood frequency analysis (RFFA). Several studies have emphasized that the identification of hydrologically homogeneous regions is the most important and challenging step in an RFFA. This study aims to identify state‐of‐the‐art clustering techniques (e.g., K‐means, partition around medoids, fuzzy C‐means, K‐harmonic means, and genetic K‐means) with potential to form hydrologically homogeneous regions for flood regionalization in Southern Brazil. The applicability of some probability density function, such as generalized extreme value, generalized logistic, generalized normal, and Pearson type 3, was evaluated based on the regions formed. Among all the 15 possible combinations of the aforementioned clustering techniques and the Euclidian, Mahalanobis, and Manhattan distance measures, the five best were selected. Several watersheds' physiographic and climatological attributes were chosen to derive multiple regression equations for all the combinations. The accuracy of the equations was quantified with respect to adjusted coefficient of determination, root mean square error, and Nash–Sutcliffe coefficient, whereas, a cross‐validation procedure was applied to check their reliability. It was concluded that reliable results were obtained when using robust clustering techniques based on fuzzy logic (e.g., K‐harmonic means), which have not been commonly used in RFFA. Furthermore, the probability density functions were capable of representing the regional annual maximum streamflows. Drainage area, main river length, and mean altitude of the watershed were the most recurrent attributes for modelling of mean annual maximum streamflow. Finally, an integration of all the five best combinations stands out as a robust, reliable, and simple tool for estimation of design floods.

Effective Kernel-Based Fuzzy Clustering Systems in Analyzing Cancer Database

The greatest challenge in high-dimensional medical cancer databases is to differentiate the available subtypes due to uncertainty associated with the objects of the database. Recently, mathematical algorithm-based diagnosing system plays an increasingly important role in analyzing the high-dimensional medical cancer databases. Particularly, fuzzy clustering technique has taken a major role in clustering medical cancer databases. The fuzzy clustering techniques are not robust with high-dimensional databases which have more similar objects. Therefore, this paper tries to propose robust fuzzy clustering algorithms for effective analysis of high-dimensional medical database. This paper introduces the objective function of proposed robust fuzzy clustering techniques by incorporating Laplacian kernel-induced distance, Canberra distance, possibilistic memberships, and fuzzy memberships. The proposed methods have been implemented with high-dimensional breast cancer database containing three subclasses which are the leading causes of cancer deaths in the world. Benchmark datasets have been used to evaluate the performance of the proposed methods, and this paper has shown the effectiveness of the proposed methods through clustering accuracy.

High Dimensional Cluster Analysis Using Path Lengths

A hierarchical scheme for clustering data is presented which applies to spaces with a high number of dimensions (). The data set is first reduced to a smaller set of partitions (multi-dimensional bins). Multiple clustering techniques are used, including spectral clustering; however, new techniques are also introduced based on the path length between partitions that are connected to one another. A Line-of-Sight algorithm is also developed for clustering. A test bank of 12 data sets with varying properties is used to expose the strengths and weaknesses of each technique. Finally, a robust clustering technique is discussed based on reaching a consensus among the multiple approaches, overcoming the weaknesses found individually.

Density-Based Multiscale Analysis for Clustering in Strong Noise Settings With Varying Densities

Finding meaningful clustering patterns in data can be very challenging when the clusters are of arbitrary shapes, different sizes, or densities, and especially when the data set contains high percentage (e.g., 80%) of noise. Unfortunately, most existing clustering techniques cannot properly handle this tough situation and often result in dramatically deteriorating performance. In this paper, a purposefully designed clustering algorithm called Density-Based Multiscale Analysis for Clustering (DBMAC)-II is proposed, which is an improved version of the latest strong-noise clustering algorithm DBMAC. DBMAC is proposed under the assumption that all clusters are homogeneous and cannot work well when the data set contains clusters of varying densities. DBMAC-II overcomes the limitation of DBMAC by executing the multiscale analysis iteratively and can conduct strong noise-robust clustering without any strict assumption on the shapes and densities of clusters. In DBMAC-II, each data point or object is mapped into a feature space using its ${r}$ -neighborhood statistics with different ${r}$ (radius) values, which is similar to DBMAC. In general, the higher the value of ${r}$ -neighborhood statistics, the more likely the object is considered as a “clustered” object. Instead of trying to find a single optimal ${r}$ value, a set of radius values appropriate for separating “clustered” objects and “noisy” objects is identified, using a formal statistical method for multimodality test, referred to as multiscale analysis. For clusters with varying densities, multiscale analysis is applied to extract the clusters with the highest density from the current data set iteratively. Moreover, a statistical uniformity test for measuring clustering tendency is used as the self-adaptive stopping criterion of the iteration. Comprehensive experimental studies on a series of challenging benchmark data sets demonstrate that DBMAC-II is not only superior to classical density-based clustering approaches, including DBSCAN, OPTICS, and HDBSCAN, but also can consistently outperform the latest strong-noise robust clustering techniques, such as Skinny-dip.