Robust Clustering Approach Research Articles

An Objective Detection of Separation Scenario in Tropical Cyclone Trajectories Based On Ensemble Weather Forecast Data

AbstractIn ensemble weather forecast, tropical cyclone (TC) tracks sometimes group together into trajectories parting away from each other. The goal of this study is to propose an objective method, based on a robust clustering approach, to detect such separation scenarios in the Japan Meteorological Agency Meso‐scale Ensemble Prediction System (MEPS) for three TCs: “Dolphin” (2020), “Nepartak” (2021), and “Meari” (2022). Taking advantage of the independence of the density‐based spatial clustering of applications with noise algorithm to the prior choice of the number of clusters, we first describe an objective way to calculate the aggregation distance, by searching the most frequent Euclidean distance between all the tracks. The clustering is then applied to the forecasted tracks, for each initialization time of MEPS (every 6 hr). Separation scenarios exist when the number of clusters is greater than one.

Disentangling the heterogeneity of brain atrophy in Alzheimer’s Disease

AbstractBackgroundAlzheimer’s disease (AD) is the most common cause of dementia. In AD patients, structural MRI analyses have identified various brain atrophy patterns, which are likely the result of differences in regional pathology distribution and variable presence of co‐pathologies.MethodWe included 496 ADNI participants and applied our robust collaborative clustering approach (RCC) to the MRI scans of the Aβ+ participants. RCC removes noise while simultaneously clustering subjects and features. Group comparisons were performed applying Kruskal‐Wallis analyses for quantitative variables and chi‐square tests for categorical variables (Table 1). Power transformations were used to normalize the distribution of quantitative variable for linear regression analyses. APOE ε4 presence and white matter hyperintensity (WMH) volume were included as predictors. P‐values <0.05 were considered statistically significant.ResultWe identified three distinct atrophy clusters among the Aβ+ participants (Table 1). A “limbic predominant” atrophy cluster showed greater cingulate and right hippocampus, inferior and middle temporal lobe atrophy than the cognitively unremarkable (CU) Aβ‐ reference group (Figure 1). The diffuse atrophy cluster showed greater parieto‐occipital‐temporal atrophy. The “hippocampus sparing” atrophy cluster showed a similar neocortical atrophy pattern except for less atrophy in the temporal lobe. The limbic predominant cluster participants had a higher APOE ε4 prevalence than the hippocampal sparing cluster participants (Table 1, 2). The three clusters had higher baseline ADAS‐Cog13 and flortaucipir SUVR binding in all the Braak staging‐defined areas than the CU Aβ‐ reference group (Table 3). The diffuse atrophy cluster had higher ADAS‐Cog13 scores, florbetapir composite scores, and flortaucipir SUVRs in all the Braak staging‐defined areas than the two other atrophy clusters. The hippocampal sparing atrophy cluster participants were older and had higher WMH volumes than the limbic predominant atrophy cluster and the CU Aβ‐ reference group participants.ConclusionOur clustering method discovered three reproducible atrophy patterns in AD, including typical and atypical presentations. This finding likely reflects different combinations of co‐pathology leading to the development of AD, some of which are captured with MRI. Mainly, WMH contributed to the heterogeneity of AD and the atypical neocortical variant.

Functional Impairment in Individuals Exposed to Violence Based on Electronical Forensic Medical Record Mining and Their Profile Identification: Controlled Observational Study.

Little is known about the functional consequences of violence when directly assessed as a primary outcome, and even less about how consistently these consequences are evaluated in a judicial context. The World Health Organization (WHO) highlighted the importance of a functional approach to health in 2001 with the release of the International Classification of Functioning, Disability, and Health (ICF). In most European countries, forensic physicians assess individuals exposed to violence to evaluate the outcomes of violence, providing certified medical evidence for magistrates' sentencing decisions. This evaluation involves a mix of objective, subjective, and contextual elements, such as reported symptoms of fear, pain, and details of the assault. Quantifying these subjective elements with scales could enhance their interpretation and application in a judicial context. This study aims to (1) characterize and (2) assess 6 scales measuring subjective elements of functional impairment among individuals exposed to violence. We conducted a retrospective study that included individuals exposed to violence examined in a French department of forensic medicine over 12 months. A typology of violence encountered in medical settings was built based on the mining of electronic health records and the use of pattern recognition algorithms. The optimal number of violence types was determined using a robust and stable clustering approach, involving sample resampling and a multimetric scheme. Patients were then paired according to their homogeneous profiles, and the intra- and interrater reproducibility of the scales was evaluated. All pain, fear, and life threat scales were significantly associated with higher functional impairment, suggesting that these measures contribute to the overall assessment of functional impairment. The intra- and interrater reproducibility of scales among similar situations of violence was measured, ranging from mild to good, with coefficients of concordance between 0.46-0.66 and 0.43-0.66, respectively. Individuals reporting intimate partner violence showed higher scores in both fear and perception of a life threat during the assault and medical interview, while individuals reporting battery by multiple unknown assailants presented higher scores only in perception of a life threat during the assault. We identified 5 remarkably stable profiles of situations of violence, consistent with clinical practice. Pain, fear, and life threat scales were related to functional impairment according to expert knowledge and demonstrated fair reproducibility under real-life conditions for similar situations of violence. Subjective elements related to functional impairment in individuals exposed to violence can be quantified using Likert scales during medical interviews.

Robust collaborative clustering approach with adaptive local structure learning

Different from one-sided clustering technique based on sample similarity, collaborative clustering captures the sample clustering structure and feature clustering structure by mining the duality between samples and features. Non-negative Matrix Tri-Factorization (NMTF) has become one of the most commonly used frameworks for collaborative clustering due to its excellent geometric significance. Unfortunately, NMTF-based approaches generally suffer from the following drawbacks: (1) Based on a fixed and fallible affinity matrix to explore the spatial intrinsic geometry structure, which limits the spatial exploration ability of the model. (2) Based on the L2 loss function to measure the reconstruction error, which weakens the robustness of the model. (3) Based on strict constraints to preserve the orthogonality of the factor matrix, which increases the burden of model solving. To overcome the aforementioned shortcomings, we propose a novel NMTF-based unsupervised collaborative clustering approach. In particular, we first propose a new adaptive local structure learning strategy to enhance the spatial exploration ability of the model. Secondly, we adopt the L1 loss function to improve the robustness of the model. Thirdly, we relax the orthogonal constraints of the factor matrix as a regularization term to reduce the burden of model solving. Finally, we propose a special four-step iterative algorithm to solve the corresponding model, and prove the convergence of the algorithm. The numerical performance on 3 synthetic data sets and 8 benchmark databases demonstrates that the proposed approach outperforms the state-of-the-art collaborative clustering algorithm.

Robust multi-view subspace clustering based on consensus representation and orthogonal diversity

The main purpose of multi-view subspace clustering is to reveal the intrinsic low-dimensional architecture of data points according to their multi-view characteristics. Exploring the potential relationship from views is one of the most essential research focuses of the multi-view task. To better utilize the complementary and consistency information from distinct views, we propose a novel robust subspace clustering approach based on consensus representation and orthogonal diversity (RMSCCO). A novel defined orthogonality term is adopted to improve the diversity and decrease the redundance of learning subspace representation. The consensus representation and subspace learning are integrated into one unified framework to characterize the consistency from views. The grouping-enhanced representation is utilized to maintain the local geometric architecture in the original data space. The ℓ2,1-norm regularizer constraint to the noise is applied to improve the robustness. Finally, an optimization algorithm is exploited to solve RMSCCO with the Alternating Direction Method of Multipliers (ADMM). Extensive experimental results on six challenging datasets demonstrate that our approach has accomplished highly qualified performance.

A copula based topology preserving graph convolution network for clustering of single-cell RNA-seq data.

Annotation of cells in single-cell clustering requires a homogeneous grouping of cell populations. There are various issues in single cell sequencing that effect homogeneous grouping (clustering) of cells, such as small amount of starting RNA, limited per-cell sequenced reads, cell-to-cell variability due to cell-cycle, cellular morphology, and variable reagent concentrations. Moreover, single cell data is susceptible to technical noise, which affects the quality of genes (or features) selected/extracted prior to clustering. Here we introduce sc-CGconv (copula based graph convolution network for single clustering), a stepwise robust unsupervised feature extraction and clustering approach that formulates and aggregates cell-cell relationships using copula correlation (Ccor), followed by a graph convolution network based clustering approach. sc-CGconv formulates a cell-cell graph using Ccor that is learned by a graph-based artificial intelligence model, graph convolution network. The learned representation (low dimensional embedding) is utilized for cell clustering. sc-CGconv features the following advantages. a. sc-CGconv works with substantially smaller sample sizes to identify homogeneous clusters. b. sc-CGconv can model the expression co-variability of a large number of genes, thereby outperforming state-of-the-art gene selection/extraction methods for clustering. c. sc-CGconv preserves the cell-to-cell variability within the selected gene set by constructing a cell-cell graph through copula correlation measure. d. sc-CGconv provides a topology-preserving embedding of cells in low dimensional space.

Disentangling tau and brain atrophy cluster heterogeneity across the Alzheimer's disease continuum.

IntroductionNeuroimaging heterogeneity in dementia has been examined using single modalities. We evaluated the associations of magnetic resonance imaging (MRI) atrophy and flortaucipir positron emission tomography (PET) clusters across the Alzheimer's disease (AD) spectrum.MethodsWe included 496 Alzheimer's Disease Neuroimaging Initiative participants with brain MRI, flortaucipir PET scan, and amyloid beta biomarker measures obtained. We applied a novel robust collaborative clustering (RCC) approach on the MRI and flortaucipir PET scans. We derived indices for AD‐like (SPARE‐AD index) and brain age (SPARE‐BA) atrophy.ResultsWe identified four tau (I–IV) and three atrophy clusters. Tau clusters were associated with the apolipoprotein E genotype. Atrophy clusters were associated with white matter hyperintensity volumes. Only the hippocampal sparing atrophy cluster showed a specific association with brain aging imaging index. Tau clusters presented stronger clinical associations than atrophy clusters. Tau and atrophy clusters were partially associated.ConclusionsEach neuroimaging modality captured different aspects of brain aging, genetics, vascular changes, and neurodegeneration leading to individual multimodal phenotyping.

A robust clustering approach for extending the lifetime of wireless sensor networks in an optimized manner with a novel fitness function

The huge load on the cluster head (CH) nodes in the clustered Wireless Sensor Network (WSN) is still a problem that causes premature death for those overloaded nodes. In this paper, we have practically proved that CH-load reduction has the largest effect on enhancing the energy efficiency of the clustering algorithms compared to some other energy efficiency- enhancing factors. So, this paper also proposes a model that divides the load of the CH role into small parts/values, in an optimized manner, and allocates these values to the whole nodes in the cluster. Therefore, using the Wireless Power Transfer (WPT) strategy, each member node in the cluster will transfer a specific amount of energy (equal to a part of the CH load that is assigned to the node) to the CH node which will also bear a suitable and a calculated part of the load instead of the whole load. So, the lifetime of the WSN is enhanced. Our simulation results show that our proposed model achieves high lifetime improvement over Leach and K-means clustering algorithms respectively.

Exploring nonnegative and low-rank correlation for noise-resistant spectral clustering

Clustering has been extensively explored in pattern recognition and data mining in order to facilitate various applications. Due to the presence of data noise, traditional clustering approaches may become vulnerable and unreliable, thereby degrading clustering performance. In this paper, we propose a robust spectral clustering approach, termed Non-negative Low-rank Self-reconstruction (NLS), which simultaneously a) explores the nonnegative low-rank properties of data correlation as well as b) adaptively models the structural sparsity of data noise. Specifically, in order to discover the intrinsic correlation among data, we devise a self-reconstruction approach to jointly consider the nonnegativity and low-rank property of data correlation matrix. Meanwhile, we propose to model data noise via a structural norm, i.e., lp,2-norm, which not only naturally conforms to genuine patterns of data noise in real-world situations, but also provides more adaptivity and flexibility to different noise levels. Extensive experiments on various real-world datasets illustrate the advantage of the proposed robust spectral clustering approach compared to existing clustering methods.

Habitat Evolution Imaging Biomarkers to Assess Early Response and Predict Treatment Outcomes in Oropharyngeal Cancer

The incidence of oropharyngeal cancer (OPC) has been rapidly increasing in recent years. There is an unmet need for reliable prognostic biomarkers in OPC. This work aims to develop an imaging signature to assess early response and predict outcomes of OPC based on serial FDG-PET and CT imaging analysis. We propose a novel spatiotemporal habitat evolution-based method to analyze the complex response patterns. First, we used a robust two-stage clustering approach to partition the primary tumor and involved lymph nodes into spatially distinct subregions (i.e., habitats) based on the local image intensity and entropy maps. Second, we proposed 27 quantitative image features to characterize the serial volumetric change of the habitats and peritumor/nodal parenchyma at both regional and single-voxel levels between baseline and mid-treatment imaging. We then evaluated the reproducibility of the proposed features with respect to multiple delineations and image registration and removed redundant features with linear correlation above 0.95. Given the remaining features, we developed an imaging signature to predict progression-free survival (PFS) by fitting an L1-regularized Cox regression model in a training cohort of 81 patients treated with concurrent radiation and chemotherapy. Finally, we tested the imaging signature in an independent validation cohort of 81 patients with similar clinical features and treatment. We identified 3 phenotypically distinct subregions based on PET and CT imaging, which were metabolically active and heterogeneous (habitat 1), enhancing and heterogeneous (habitat 2), metabolically inactive and homogeneous (habitat 3). All 3 habitats showed high concordance across the two cohorts with in-group proportion ranging from 0.97 to 0.99. After removing redundant and less stable features (intra-class coefficient < 0.75), 21 features remained. The final Cox model consisted of 3 habitat evolution-based image features. This imaging signature predicted 2-year PFS with a reasonably good accuracy, AUC: 0.74, 95% CI: 0.69-0.78 (training) and 0.70, 95% CI: 0.65-0.74 (validation), and significantly outperformed traditional imaging response metrics such as tumor volume, SUVmax and MTV (AUC: 0.52-0.58). When tested in the validation cohort, the imaging signature stratified patients into high vs low-risk groups with distinct 2-year PFS (64% vs 90%; P = 0.003). On multivariable analysis, the imaging signature was independently associated with PFS (hazard ratio: 5.2, 95% CI: 1.6-17.5, P = 0.008) adjusting for clinical and pathologic factors including stage, HPV status, and smoking history. The proposed habitat evolution-based imaging signature allows reliable assessment of early response and accurate prediction of outcomes in OPC. If validated, it can help identify patients who may benefit from de-intensification therapy or adjuvant immunotherapy.

Tumor Subregion Evolution-Based Imaging Features to Assess Early Response and Predict Prognosis in Oropharyngeal Cancer.

The incidence of oropharyngeal squamous cell carcinoma (OPSCC) has been rapidly increasing. Disease stage and smoking history are often used in current clinical trials to select patients for deintensification therapy, but these features lack sufficient accuracy for predicting disease relapse. Our purpose was to develop an imaging signature to assess early response and predict outcomes of OPSCC. Methods: We retrospectively analyzed 162 OPSCC patients treated with concurrent chemoradiotherapy, equally divided into separate training and validation cohorts with similar clinical characteristics. A robust consensus clustering approach was used to spatially partition the primary tumor and involved lymph nodes into subregions (i.e., habitats) based on 18F-FDG PET and contrast CT imaging. We proposed quantitative image features to characterize the temporal volumetric change of the habitats and peritumoral/nodal tissue between baseline and midtreatment. The reproducibility of these features was evaluated. We developed an imaging signature to predict progression-free survival (PFS) by fitting an L1-regularized Cox regression model. Results: We identified 3 phenotypically distinct intratumoral habitats: metabolically active and heterogeneous, enhancing and heterogeneous, and metabolically inactive and homogeneous. The final Cox model consisted of 4 habitat evolution-based features. In both cohorts, this imaging signature significantly outperformed traditional imaging metrics, including midtreatment metabolic tumor volume for predicting PFS, with a C-index of 0.72 versus 0.67 (training) and 0.66 versus 0.56 (validation). The imaging signature stratified patients into high-risk versus low-risk groups with 2-y PFS rates of 59.1% versus 89.4% (hazard ratio, 4.4; 95% confidence interval, 1.4-13.4 [training]) and 61.4% versus 87.8% (hazard ratio, 4.6; 95% confidence interval, 1.7-12.1 [validation]). The imaging signature remained an independent predictor of PFS in multivariable analysis adjusting for stage, human papillomavirus status, and smoking history. Conclusion: The proposed imaging signature allows more accurate prediction of disease progression and, if prospectively validated, may refine OPSCC patient selection for risk-adaptive therapy.

Robust low-rank kernel multi-view subspace clustering based on the Schatten p-norm and correntropy

Multi-view subspace clustering, which studies the similarities and differences among data in multiple views, is an efficient clustering problem. However, to deal with problems that have non-linear structures and non-Gaussian noise in multi-view data, existing clustering methods relax the original problem convexly. The solutions that are generated by these convex relaxations are not the optimal solutions to the original problem. To overcome this deficiency, this paper presents a robust low-rank kernel multi-view subspace clustering approach that combines the non-convex Schatten p-norm (0<p≤1) regularizer with the “kernel trick”, which can efficiently deal with problems that have non-linear structures in multi-view data via non-convex methods. In addition, the correntropy is introduced into our model, which is a robust measure of the corruptions that are caused by non-Gaussian noise. Moreover, our method can learn the joint subspace representation of all views. Because it learns a low-rank kernel mapping, the data in the feature space are both low-rank and self-expressed. The optimization problems are solved efficiently via an iterative algorithm (HQ-ADMM). This algorithm can ensure that each iteration has closed-form solutions, which simplifies the optimization problems substantially. Experimental comparisons on five real-world datasets demonstrate that the proposed algorithm outperforms state-of-the-art multi-view subspace clustering algorithms.

A reweighting approach to robust clustering

An iteratively reweighted approach for robust clustering is presented in this work. The method is initialized with a very robust clustering partition based on an high trimming level. The initial partition is then refined to reduce the number of wrongly discarded observations and substantially increase efficiency. Simulation studies and real data examples indicate that the final clustering solution has both good properties in terms of robustness and efficiency and naturally adapts to the true underlying contamination level.

Bank branch operational performance: A robust multivariate and clustering approach

This paper proposes a multi-step procedure that integrates robust methods, clustering analysis and data envelopment analysis (DEA) to identify bank branch managerial clusters and to study efficiency performance. By applying robust techniques based on principal component analysis, we look for (1) the detection of influential branches, i.e., exhibiting extreme operating behaviors, and (2) the clustering of branches based on operating characteristics. Our premise is that influential branches affect both the clustering and the determination of efficiency performance. The application of the procedure yields various aggregate influential-based branch profiles along with cluster profiles. These aggregate profiles provide valuable insights on the determinants of branch efficiency performance and operating patterns. Using the profiles as contextual information, DEA input-oriented slack-based models are applied to study branch efficiency performance from meta-frontier and cluster-frontier perspectives. Branch performance is characterized in terms of influential-based and cluster profiles, and efficiency designations. This allows for the understanding of how efficiency and peer selection are affected by influential branches, and how the profiles can be used to inform design decisions.

Visual hierarchical clustering of supply chain using growing hierarchical self-organising map algorithm

The study identifies a need for efficient and robust visual clustering approach that can potentially deal with complex supply chain clustering problems. Based on the underlying philosophy of group technology, a growing hierarchical self-organising map algorithm (GHSOM) is proposed to identify a lower two-dimension visual clustering map that can effectively address supply chain clustering problems. The proposed approach provides optimal solutions by decomposing a large-sized supply chain problem into independent, small, manageable problems. It facilitates simple decision-making by exploring similar clusters that are represented by the neighbouring branches in the GHSOM map structure. Unlike other approaches in literature, the proposed approach can further attain good topological ordered representations of the various work order families, to be processed by clusters of supply units along with information on hierarchical sub-cell formation as identifiable from the visually navigable map. The proposed approach has been successfully applied on 16 benchmarked problems. The performance of GHSOM based on grouping efficacy measure outperformed the best results in literature.

Robust Dynamic Clustering Approach to Improve Quality of Service in Wireless Sensor Network

Robust Dynamic Clustering Approach to Improve Quality of Service in Wireless Sensor Network

A robust fiber clustering method for tract refinement

Diffusion tensor imaging fiber tracking (DTI-FT) could delineate neuronal fiber tracts, and it has been widely accepted in clinical practices. Due to the two reasons of image noise and low resolution of DTI, DTI-FT would easily lead to false propagations. In this paper, we propose a robust fiber clustering (FC) approach to diminish false fibers from one fiber tract. Our method consists of two stages: First, the optimized FT using single-tensor fiber assignment continuous tracking (FACT) is implemented to demonstrate one fiber tract; second, each curved fiber in the fiber tract is mapped to a point by kernel principal component analysis (KPCA), and then the point clouds of fiber tract are labeled by hierarchical clustering which could distinguish false fibers from true fibers in one tract. In our experiment, the corticospinal tract (CST) in one case of human data in vivo was used to validate our method. Compared with the k-most similar fibers method, our method showed more reliable capability in decreasing the false fibers in one tract. In conclusion, our method could effectively optimize the visualization of fiber bundles and would help a lot in the field of fiber tract evaluation.

TCLUST: Trimming Approach of Robust Clustering Method

TCLUST is a method in statistical clustering technique which is based on modification of trimmed k-means clustering algorithm. It is called “crisp” clustering approach because the observation is can be eliminated or assigned to a group. TCLUST strengthen the group assignment by putting constraint to the cluster scatter matrix. The emphasis in this paper is to restrict on the eigenvalues, λ of the scatter matrix. The idea of imposing constraints is to maximize the log-likelihood function of spurious-outlier model. A review of different robust clustering approach is presented as a comparison to TCLUST methods. This paper will discuss the nature of TCLUST algorithm and how to determine the number of cluster or group properly and measure the strength of group assignment. At the end of this paper, R-package on TCLUST implement the types of scatter restriction, making the algorithm to be more flexible for choosing the number of clusters and the trimming proportion.

Forecasting Simultaneously High‐Dimensional Time Series: A Robust Model‐Based Clustering Approach

ABSTRACTThis paper considers the problem of forecasting high‐dimensional time series. It employs a robust clustering approach to perform classification of the component series. Each series within a cluster is assumed to follow the same model and the data are then pooled for estimation. The classification is model‐based and robust to outlier contamination. The robustness is achieved by using the intrinsic mode functions of the Hilbert–Huang transform at lower frequencies. These functions are found to be robust to outlier contamination. The paper also compares out‐of‐sample forecast performance of the proposed method with several methods available in the literature. The other forecasting methods considered include vector autoregressive models with ∕ without LASSO, group LASSO, principal component regression, and partial least squares. The proposed method is found to perform well in out‐of‐sample forecasting of the monthly unemployment rates of 50 US states. Copyright © 2013 John Wiley &amp; Sons, Ltd.

Robust constrained fuzzy clustering

It is well-known that outliers and noisy data can be very harmful when applying clustering methods. Several fuzzy clustering methods which are able to handle the presence of noise have been proposed. In this work, we propose a robust clustering approach called F-TCLUST based on trimming a fixed proportion of observations that are (“impartially”) determined by the data set itself. The proposed approach also considers an eigenvalue ratio constraint that makes it a mathematically well-defined problem and serves to control the allowed differences among cluster scatters. A computationally feasible algorithm is proposed for its practical implementation. Some guidelines about how to choose the parameters controlling the performance of the fuzzy clustering procedure are also given.