Model-based Clustering Research Articles

Model-Based Clustering of Categorical Data Based on the Hamming Distance

A model-based approach is developed for clustering categorical data with no natural ordering. The proposed method exploits the Hamming distance to define a family of probability mass functions to model the data. The elements of this family are then considered as kernels of a finite mixture model with an unknown number of components. Conjugate Bayesian inference has been derived for the parameters of the Hamming distribution model. The mixture is framed in a Bayesian nonparametric setting, and a transdimensional blocked Gibbs sampler is developed to provide full Bayesian inference on the number of clusters, their structure, and the group-specific parameters, facilitating the computation with respect to customary reversible jump algorithms. The proposed model encompasses a parsimonious latent class model as a special case when the number of components is fixed. Model performances are assessed via a simulation study and reference datasets, showing improvements in clustering recovery over existing approaches. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work.

Bayesian clustering of high-dimensional data via latent repulsive mixtures

SUMMARY Model-based clustering of moderate or large dimensional data is notoriously difficult. We propose a model for simultaneous dimensionality reduction and clustering by assuming a mixture model for a set of latent scores, which are then linked to the observations via a Gaussian latent factor model. This approach was recently investigated by Chandra et al. (2023). The authors use a factor-analytic representation and assume a mixture model for the latent factors. However, performance can deteriorate in the presence of model misspecification. Assuming a repulsive point process prior for the component-specific means of the mixture for the latent scores is shown to yield a more robust model that outperforms the standard mixture model for the latent factors in several simulated scenarios. The repulsive point process must be anisotropic to favour well-separated clusters of data, and its density should be tractable for efficient posterior inference. We address these issues by proposing a general construction for anisotropic determinantal point processes. We illustrate our model in simulations as well as a plant species co-occurrence dataset.

An autoencoder-based confederated clustering leveraging a robust model fusion strategy for federated unsupervised learning

Concerns related to data privacy, security, and ethical considerations become more prominent as data volumes continue to grow. In contrast to centralized setups, where all data is accessible at a single location, model-based clustering approaches can be successfully employed in federated settings. However, this approach to clustering in federated settings is still relatively unexplored and requires further attention. As federated clustering deals with remote data and requires privacy and security to be maintained, it poses particular challenges as well as possibilities. While model-based clustering offers promise in federated environments, a robust model aggregation method is essential for clustering rather than the generic model aggregation method like Federated Averaging (FedAvg). In this research, we proposed an autoencoder-based clustering method by introducing a novel model aggregation method FednadamN, which is a fusion of Adam and Nadam optimization approaches in a federated learning setting. Therefore, the proposed FednadamN adopted the adaptive learning rates based on the first and second moments of gradients from Adam which offered fast convergence and robustness to noisy data. Furthermore, FednadamN also incorporated the Nesterov-accelerated gradients from Nadam to further enhance the convergence speed and stability. We have studied the performance of the proposed Autoencoder-based clustering methods on benchmark datasets and using the novel FednadamN model aggregation strategy. It shows remarkable performance gain in federated clustering in comparison to the state-of-the-art.

Toward understanding autism heterogeneity: Identifying clinical subgroups and neuroanatomical deviations.

Autism spectrum disorder ("autism") is a neurodevelopmental condition characterized by substantial behavioral and neuroanatomical heterogeneity. This poses significant challenges to understanding its neurobiological mechanisms and developing effective interventions. Identifying phenotypically more homogeneous subgroups and shifting the focus from average group differences to individuals is a promising approach to addressing heterogeneity. In the present study, we aimed to parse clinical and neuroanatomical heterogeneity in autism by combining clustering of clinical features with normative modeling based on neuroanatomical measures (cortical thickness [CT] and subcortical volume) within the Autism Brain Imaging Data Exchange data sets (N autism = 861, N nonautistic individuals [N NAI] = 886, age range = 5-64). First, model-based clustering was applied to autistic symptoms as measured by the Autism Diagnostic Observation Schedule to identify clinical subgroups of autism (N autism = 499). Next, we ran normative modeling on CT and subcortical parcellations (N autism = 690, N NAI = 744) and examined whether clinical subgrouping resulted in increased neurobiological homogeneity within the subgroups compared to the entire autism group by comparing their spatial overlap of neuroanatomical deviations. We further investigated whether the identified subgroups improved the accuracy of autism classification based on the neuroanatomical deviations using supervised machine learning with cross-validation. Results yielded two clinical subgroups primarily differing in restrictive and repetitive behaviors (RRBs). Both subgroups showed increased homogeneity in localized deviations with the high-RRB subgroup showing increased volume deviations in the cerebellum and the low-RRB subgroup showing decreased CT deviations predominantly in the postcentral gyrus and fusiform cortex. Nevertheless, substantial within-group heterogeneity remained highlighted by the lack of improvement of the classifier's performance when distinguishing between the subgroups and NAI. Future research should aim to further reduce heterogeneity incorporating additional neuroanatomical clustering in even larger samples. This will eventually pave the way for more tailored behavioral interventions and improving clinical outcomes. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Derivation of outcome-dependent dietary patterns for low-income women obtained from survey data using a supervised weighted overfitted latent class analysis.

Poor diet quality is a key modifiable risk factor for hypertension and disproportionately impacts low-income women. Analyzing diet-driven hypertensive outcomes in this demographic is challenging due to the complexity of dietary data and selection bias when the data come from surveys, a main data source for understanding diet-disease relationships in understudied populations. Supervised Bayesian model-based clustering methods summarize dietary data into latent patterns that holistically capture relationships among foods and a known health outcome but do not sufficiently account for complex survey design. This leads to biased estimation and inference and lack of generalizability of the patterns. To address this, we propose a supervised weighted overfitted latent class analysis (SWOLCA) based on a Bayesian pseudo-likelihood approach that integrates sampling weights into an exposure-outcome model for discrete data. Our model adjusts for stratification, clustering, and informative sampling, and handles modifying effects via interaction terms within a Markov chain Monte Carlo Gibbs sampling algorithm. Simulation studies confirm that the SWOLCA model exhibits good performance in terms of bias, precision, and coverage. Using data from the National Health and Nutrition Examination Survey (2015-2018), we demonstrate the utility of our model by characterizing dietary patterns associated with hypertensive outcomes among low-income women in the United States.

B-170 Age-Specific reference intervals for ethanolamine plasmalogen species in red blood cells using liquid chromatography tandem mass spectrometry

Abstract Background Plasmalogens are critical membrane structural components that are mainly generated by de novo synthesis starting in peroxisomes. Hence, patients with defects in peroxisome biogenesis (PBD) exhibit markedly reduced plasmalogen levels. Plasmalogen ratios are traditionally measured by gas chromatography-mass spectrometry (GC-MS); however, this method entails a lengthy sample extraction and derivatization and does not report concentrations of individual plasmalogen species. We have developed a robust and easy-to-implement liquid chromatography-tandem mass spectrometry (LC-MS/MS) method to quantify the 18 most abundant ethanolamine plasmalogen (PlsEtn) species in packed red blood cells (RBCs). However, no reference intervals have been published for individual PlsEtn. Here, we describe the establishment of age-specific reference intervals for individual PlsEtn species and their total values (16:0, 18:0, 18:1 species, and total plasmalogens). Methods Plasmalogens were extracted using methanol containing two labeled internal standards, shaking for one hour at room temperature. Chromatographic separation was performed using an Acquity Premier BEH C18 UPLC column with a binary gradient of 5 mM ammonium acetate in water:methanol (15:85) and 5 mM ammonium acetate in methanol. Analysis was performed using a XEVO TQ-XS Mass Spectrometer with Ultra-High Performance Liquid Chromatography (Waters) in multiple reaction monitoring mode. Eighteen PlsEtn species were quantified using four commercially available standards; additionally, totals were calculated for 16:0, 18:0 or 18:1 species, and for total plasmalogens. Reference intervals were established using 376 RBCs from self-reported healthy volunteers and de-identified clinical samples referred for unrelated testing (182 females and 194 males; range 0 to 88 years). Data was analyzed using the R programming language. The study was approved by the Institutional Review Board of the University of Utah. Results Initial age groups were identified using a model-based clustering algorithm followed by iterative Harris-Boyd analysis. Finally, the adjacent groups were merged if their means differed by less than 10%. Once the final age groups were partitioned, data in each individual age group were analyzed using parametric or non-parametric statistics to determine reference intervals (95%, with 90%confidence intervals). PlsEtn species displayed the lowest concentration in the first few months of life, which increased in childhood until adolescence or adulthood (depending on PlsEtn). For most species, the concentrations increased over time reaching a plateau between 18 and 48 years of age, and then starting to decrease. The total values followed the same trend, with neonates showing significantly lower values compared to other age groups. Conclusions We applied a novel statistical approach to identify age groups and determine age-specific reference intervals for 18 individual PlsEtn in RBC and their totals. Lacking previously published data, this study is critical for supporting test implementation in clinical laboratories.

A Biomarker Signature Associated With Future Functional Decline And Outcomes On The Frailty-Disability Cascade

Abstract Background Biological ageing is a dynamic and heterogeneous process that occurs due to dysregulation at the molecular, cellular, physiological and functional level, manifesting in functional decline. We investigated whether biomarker data across a range of biological systems predicted incident functional decline and frailty-disability cascade outcomes. Methods Secondary data analyses were performed on n=4,961 participants aged 50+ years from waves 1-5 of The Irish Longitudinal Study on Ageing (TILDA). We examined 18 blood biomarkers representing key systems that become dysregulated with ageing (inflammatory, cardiovascular, metabolic, renal, cognitive, musculoskeletal and senescence). Model-based clustering classified participants as belonging to one of three biomarker signatures - low, medium and high risk. These biomarker signatures were utilized to predict 4-year functional decline (grip strength, timed up and go test (TUG) and gait speed) using mixed effect models; 8-year frailty and disability using logistic regression models; and 12-year mortality using Cox-Proportional Hazard models. All models were adjusted for age, sex, education, smoking history, lipid lowering and anti-hypertensive medications. Results The prevalence of the low, medium and high-risk groups was 58.5%, 9.2% and 32.3%. Medium and high-risk biomarker signatures had progressively increased hazard of 12-year mortality (HR:1.54, 95%CI:1.13-2.10) and (HR:1.89, 95%CI:1.53-2.34) respectively. Participants with the high-risk signature also had significantly higher odds of 8-year incident frailty (OR:2.1, 95%CI:1.60-2.74) and disability (OR:1.72, 95%CI:1.30-2.29), with the medium-risk group showing attenuated associations. The biomarker signatures had a dose response relationship with functional decline; with the high-risk group having lower grip strength, slower walking speed and higher time to complete the TUG over 4-year follow-up (p<0.01). Findings were replicated in a US cohort. Conclusion We report a unique biomarker signature that was consistently associated with mortality, incident frailty and disability, and functional decline. This signature may be a useful early indicator and risk stratification tool to predict future functional decline and frailty-disability cascade outcomes.

Association between betrayal trauma and typologies of anger and aggression

Anger and aggression are related forms of externalizing phenomena that are distinct in that anger is a form of emotion, whereas aggression is a behavior. Both anger and aggression (as well as their combination) function as responses to environmental conditions, one of the most common of which is trauma. Recent research has highlighted the association between trauma high in betrayal and externalizing problems, although this research has not been examined with respect to typologies of anger and aggression specifically. In this study, we examined the association between different forms of trauma and typologies of anger and aggression in a sample of women and men recruited online (N = 364) using model-based cluster analysis and multinomial logistic regression. Results suggest that trauma with a high degree of betrayal was associated with a typology of implicit anger characterized by intense anger but controlled aggression. Findings suggest the importance of understanding the impact of betrayal on anger and aggression.

Dense SNP-based analyses complement forensic anthropology biogeographical ancestry assessments

Identification of unidentified human remains (UHRs) is crucial yet challenging, especially with traditional forensic techniques. Forensic anthropological examinations can yield ancestry estimations; however, the utility of these estimates is limited by the data points that can be collected from partial remains, complexities of admixture, and variation of phenotypic expression due to environmental effects. While it is generally known that anthropological estimates can be imprecise, the performance of these methods has not been studied at scale. Genome-wide SNP testing is an orthogonal approach for estimating ancestry and offers a unique opportunity to measure the magnitude of anthropological ancestry misattribution. Genomic ancestry inference leverages principal component analysis (PCA) and model-based clustering approaches. This study compares anthropologically determined ancestry with those estimated using genome-wide SNP markers. A dataset of 611 UHR samples with publicly available ancestry assessments from National Missing and Unidentified Persons System (NamUs) was analyzed. The genetic ancestry approach, validated against reference population samples, offers robust ancestry calculations for major population groups. Inconsistency between anthropological and genomic ancestry assignments were observed, particularly for admixed populations. Although forensic anthropological examinations remain valuable, their limitations emphasize the need for refinement and enhancement through the augmentation of SNP-based analyses. Further validation studies are crucial to define the uncertainty associated with both anthropological and genome-based ancestry estimates to resolve cases and aid law enforcement investigations. Additionally, current policy and practices for reporting ancestry for UHRs should be revisited to reduce potential misinformation.

Clinical phenotypes of sarcoidosis using cluster analysis: a Spanish population-based cohort study.

Sarcoidosis is a clinically heterogenous disease. The objective of this study is the identification of clinical phenotypes using cluster analysis. A model-based clustering relaying on 19 clinical variables was performed in a retrospective cohort of 342 sarcoidosis patients, diagnosed and followed-up from 1999 to 2019 in a tertiary hospital at Northern Spain. Chi-square test and ANOVA were used to compare categorical and continuous variables among groups. Two-sample t-tests and the partition of Pearson's chi-square statistic were used in pairwise comparisons. The Wasfi severity score was calculated and compared among clusters. Cluster analysis identified five groups: C1 (16.1%), C2 (14.3%), C3 (24.3%), C4 (5.0%), and C5 (40.4%). Lung involvement was predominant, ranging from 55.1% (C2) to 100% (C1 and C4). Extrapulmonary involvement was significantly higher in C2 (96.4%) and C3 (98.0%). A significant lower FEV1 percent predicted was detected in C5 (90.5±21.8) versus C1 (102.0±22.9), C3 (102.3±17.6) and C4 (105.8±20.8). The cluster 5 had a lower FVC percent predicted (96.6±18.9) than others, ranging from 108.1±18.0 (C3) to 111.5±21.7 (C4). The prescription of systemic glucocorticoids and non-corticosteroid immunosuppressants was higher in the clusters 1, 3 and 5. Chronicity rates were higher in C3 (31.3%) and C5 (32.6%) compared to C1 (9.1%) and C4 (0%), as well as the Wasfi severity score values. Five phenotypes with different clinical and prognostic characteristics are proposed in our study. Cluster analysis can be a useful tool for identifying clinical patterns in a disease as heterogeneous as sarcoidosis and optimising its management.

Data-driven subclassification of ANCA-associated vasculitis: model-based clustering of a federated international cohort

Antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis is a heterogenous autoimmune disease. While traditionally stratified into two conditions, granulomatosis with polyangiitis (GPA) and microscopic polyangiitis (MPA), the subclassification of ANCA-associated vasculitis is subject to continued debate. Here we aim to identify phenotypically distinct subgroups and develop a data-driven subclassification of ANCA-associated vasculitis, using a large real-world dataset. In the collaborative data reuse project FAIRVASC (Findable, Accessible, Interoperable, Reusable, Vasculitis), registry records of patients with ANCA-associated vasculitis were retrieved from six European vasculitis registries: the Czech Registry of ANCA-associated vasculitis (Czech Republic), the French Vasculitis Study Group Registry (FVSG; France), the Joint Vasculitis Registry in German-speaking Countries (GeVas; Germany), the Polish Vasculitis Registry (POLVAS; Poland), the Irish Rare Kidney Disease Registry (RKD; Ireland), and the Skåne Vasculitis Cohort (Sweden). We performed model-based clustering of 17 mixed-type clinical variables using a parsimonious mixture of two latent Gaussian variable models. Clinical validation of the optimal cluster solution was made through summary statistics of the clusters' demography, phenotypic and serological characteristics, and outcome. The predictive value of models featuring the cluster affiliations were compared with classifications based on clinical diagnosis and ANCA specificity. People with lived experience were involved throughout the FAIRVASVC project. A total of 3868 patients diagnosed with ANCA-associated vasculitis between Nov 1, 1966, and March 1, 2023, were included in the study across the six registries (Czech Registry n=371, FVSG n=1780, GeVas n=135, POLVAS n=792, RKD n=439, and Skåne Vasculitis Cohort n=351). There were 2434 (62·9%) patients with GPA and 1434 (37·1%) with MPA. Mean age at diagnosis was 57·2 years (SD 16·4); 2006 (51·9%) of 3867 patients were men and 1861 (48·1%) were women. We identified five clusters, with distinct phenotype, biochemical presentation, and disease outcome. Three clusters were characterised by kidney involvement: one severe kidney cluster (555 [14·3%] of 3868 patients) with high C-reactive protein (CRP) and serum creatinine concentrations, and variable ANCA specificity (SK cluster); one myeloperoxidase (MPO)-ANCA-positive kidney involvement cluster (782 [20·2%]) with limited extrarenal disease (MPO-K cluster); and one proteinase 3 (PR3)-ANCA-positive kidney involvement cluster (683 [17·7%]) with widespread extrarenal disease (PR3-K cluster). Two clusters were characterised by relative absence of kidney involvement: one was a predominantly PR3-ANCA-positive cluster (1202 [31·1%]) with inflammatory multisystem disease (IMS cluster), and one was a cluster (646 [16·7%]) with predominantly ear-nose-throat involvement and low CRP, with mainly younger patients (YR cluster). Compared with models fitted with clinical diagnosis or ANCA status, cluster-assigned models demonstrated improved predictive power with respect to both patient and kidney survival. Our study reinforces the view that ANCA-associated vasculitis is not merely a binary construct. Data-driven subclassification of ANCA-associated vasculitis exhibits higher predictive value than current approaches for key outcomes. European Union's Horizon 2020 research and innovation programme under the European Joint Programme on Rare Diseases.

Model-based multifacet clustering with high-dimensional omics applications.

High-dimensional omics data often contain intricate and multifaceted information, resulting in the coexistence of multiple plausible sample partitions based on different subsets of selected features. Conventional clustering methods typically yield only one clustering solution, limiting their capacity to fully capture all facets of cluster structures in high-dimensional data. To address this challenge, we propose a model-based multifacet clustering (MFClust) method based on a mixture of Gaussian mixture models, where the former mixture achieves facet assignment for gene features and the latter mixture determines cluster assignment of samples. We demonstrate superior facet and cluster assignment accuracy of MFClust through simulation studies. The proposed method is applied to three transcriptomic applications from postmortem brain and lung disease studies. The result captures multifacet clustering structures associated with critical clinical variables and provides intriguing biological insights for further hypothesis generation and discovery.

Leafiness-LiDAR index and NDVI for identification of temporal patterns in super-intensive almond orchards as response to different management strategies

The use of super-intensive orchards is a growing trend in fruit production. The present study aims to improve management of these cropping systems by focusing on how agronomic decisions impact orchard dynamics in the short to medium term and by providing a decision-support approach based on stable temporal patterns from previous seasons. A multitemporal study using remote sensing and LiDAR was conducted in a commercial almond orchard over four growing seasons (2019–2022) to determine the optimal timing of image acquisition for variable pre-harvest treatments. A model-based clustering (mclust) was applied to optimal Sentinel-2 NDVI maps and apparent soil electrical conductivity (ECa) data, interpolated to the pixel centroids of Sentinel-2 image grids, to delineate potential management zones (PMZs). The leafiness-LiDAR index (LLI), a leaf area index (LAI) estimator, was obtained as ground truth after summer pruning and before harvesting, showing a significant influence of fertigation and pruning on the LAI, with summer pruning particularly influencing orchard dynamics. The optimal time for NDVI mapping was found to be two months after summer pruning in productive years and two weeks after in unproductive years. The delineated PMZs were consistent across seasons and corresponded to significant LAI differences. This method could contribute to improving resource management and sustainability in super-intensive commercial orchards.

Using high-density SNP data to unravel the origin of the Franches-Montagnes horse breed

BackgroundThe Franches-Montagnes (FM) is the last native horse breed of Switzerland, established at the end of the 19th century by cross-breeding local mares with Anglo-Norman stallions. We collected high-density SNP genotype data (Axiom™ 670 K Equine genotyping array) from 522 FM horses, including 44 old-type horses (OF), 514 European Warmblood horses (WB) from Sweden and Switzerland (including a stallion used for cross-breeding in 1990), 136 purebred Arabians (AR), 32 Shagya Arabians (SA), and 64 Thoroughbred (TB) horses, as introgressed WB stallions showed TB origin in their pedigrees. The aim of the study was to ascertain fine-scale population structures of the FM breed, including estimation of individual admixture levels and genomic inbreeding (FROH) by means of Runs of Homozygosity.ResultsTo assess fine-scale population structures within the FM breed, we applied a three-step approach, which combined admixture, genetic contribution, and FROH of individuals into a high-resolution network visualization. Based on this approach, we were able to demonstrate that population substructures, as detected by model-based clustering, can be either associated with a different genetic origin or with the progeny of most influential sires. Within the FM breed, admixed horses explained most of the genetic variance of the current breeding population, while OF horses only accounted for a small proportion of the variance. Furthermore, we illustrated that FM horses showed high TB admixture levels and we identified inconsistencies in the origin of FM horses descending from the Arabian stallion Doktryner. With the exception of WB, FM horses were less inbred compared to the other breeds. However, the relatively few but long ROH segments suggested diversity loss in both FM subpopulations. Genes located in FM- and OF-specific ROH islands had known functions involved in conformation and behaviour, two traits that are highly valued by breeders.ConclusionsThe FM remains the last native Swiss breed, clearly distinguishable from other historically introgressed breeds, but it suffered bottlenecks due to intensive selection of stallions, restrictive mating choices based on arbitrary definitions of pure breeding, and selection of rare coat colours. To preserve the genetic diversity of FM horses, future conservation managements strategies should involve a well-balanced selection of stallions (e.g., by integrating OF stallions in the FM breeding population) and avoid selection for rare coat colours.

Phenogrouping heart failure with preserved or mildly reduced ejection fraction using electronic health record data

BackgroundHeart failure (HF) with preserved or mildly reduced ejection fraction includes a heterogenous group of patients. Reclassification into distinct phenogroups to enable targeted interventions is a priority. This study aimed to identify distinct phenogroups, and compare phenogroup characteristics and outcomes, from electronic health record data.Methods2,187 patients admitted to five UK hospitals with a diagnosis of HF and a left ventricular ejection fraction ≥ 40% were identified from the NIHR Health Informatics Collaborative database. Partition-based, model-based, and density-based machine learning clustering techniques were applied. Cox Proportional Hazards and Fine-Gray competing risks models were used to compare outcomes (all-cause mortality and hospitalisation for HF) across phenogroups.ResultsThree phenogroups were identified: (1) Younger, predominantly female patients with high prevalence of cardiometabolic and coronary disease; (2) More frail patients, with higher rates of lung disease and atrial fibrillation; (3) Patients characterised by systemic inflammation and high rates of diabetes and renal dysfunction. Survival profiles were distinct, with an increasing risk of all-cause mortality from phenogroups 1 to 3 (p < 0.001). Phenogroup membership significantly improved survival prediction compared to conventional factors. Phenogroups were not predictive of hospitalisation for HF.ConclusionsApplying unsupervised machine learning to routinely collected electronic health record data identified phenogroups with distinct clinical characteristics and unique survival profiles.

A New Look at the Dirichlet Distribution: Robustness, Clustering, and Both Together

AbstractCompositional data have peculiar characteristics that pose significant challenges to traditional statistical methods and models. Within this framework, we use a convenient mode parametrized Dirichlet distribution across multiple fields of statistics. In particular, we propose finite mixtures of unimodal Dirichlet (UD) distributions for model-based clustering and classification. Then, we introduce the contaminated UD (CUD) distribution, a heavy-tailed generalization of the UD distribution that allows for a more flexible tail behavior in the presence of atypical observations. Thirdly, we propose finite mixtures of CUD distributions to jointly account for the presence of clusters and atypical points in the data. Parameter estimation is carried out by directly maximizing the maximum likelihood or by using an expectation-maximization (EM) algorithm. Two analyses are conducted on simulated data to illustrate the effects of atypical observations on parameter estimation and data classification, and how our proposals address both aspects. Furthermore, two real datasets are investigated and the results obtained via our models are discussed.

Model-based clustering with missing not at random data

Model-based clustering with missing not at random data

Explainable finite mixture of mixtures of bounded asymmetric generalized Gaussian and Uniform distributions learning for energy demand management

We introduce a mixture of mixtures of bounded asymmetric generalized Gaussian and uniform distributions. Based on this framework, we propose model-based classification and model-based clustering algorithms. We develop an objective function for the minimum message length (MML) model selection criterion to discover the optimal number of clusters for the unsupervised approach of our proposed model. Given the crucial attention received by Explainable AI (XAI) in recent years, we introduce a method to interpret the predictions obtained from the proposed model in both learning settings by defining their boundaries in terms of the crucial features. Integrating Explainability within our proposed algorithm increases the credibility of the algorithm’s predictions since it would be explainable to the user’s perspective through simple If-Then statements using a small binary decision tree. In this paper, the proposed algorithm proves its reliability and superiority to several state-of-the-art machine learning algorithms within the following real-world applications: fault detection and diagnosis (FDD) in chillers, occupancy estimation and categorization of residential energy consumers.

A Bayesian hierarchical hidden Markov model for clustering and gene selection: Application to kidney cancer gene expression data.

We introduce a Bayesian approach for biclustering that accounts for the prior functional dependence between genes using hidden Markov models (HMMs). We utilize biological knowledge gathered from gene ontologies and the hidden Markov structure to capture the potential coexpression of neighboring genes. Our interpretable model-based clustering characterized each cluster of samples by three groups of features: overexpressed, underexpressed, and irrelevant features. The proposed methods have been implemented in an R package and are used to analyze both the simulated data and The Cancer Genome Atlas kidney cancerdata.

Finding Outliers in Gaussian Model-based Clustering

Finding Outliers in Gaussian Model-based Clustering