Unsupervised Data Research Articles

Unsupervised anomaly detection for manufacturing product images by significant feature space distance measurement

Using camera instruments to detect product anomalies is increasingly being applied in various manufacturing scenarios. Detection of anomaly product images and identification of anomaly areas can not only ensure product quality, but also help reveal potential problems in the manufacturing process. Unsupervised data environments and diverse engineering requirements are two challenges faced by modern manufacturing. To solve the problem of manufacturing product anomaly detection in the absence of anomaly data, and meet the requirements of model training cost, detection speed and accuracy, a fast deployable unsupervised image anomaly detection method is proposed without parameter training. The main idea is to design an efficient and training-free anomaly detection decoder. The encoder output is screened by non-parametric method, and the image anomaly score is calculated by the cumulative distance of the screened feature space. In addition, the proposed model is able to give the reference samples on which the prediction results are based to improve the interpretability of the predictions. Experiments are conducted with 15 real product image datasets, and the results show that the proposed method achieves performance improvement over other methods for both anomaly detection and segmentation.

Deep Learning-Based Equivalent Modelling of Hybrid RES Plant for Efficient, Repetitive Power System Transient Stability Studies

The paper presents the methodology for dynamic equivalent modelling of hybrid renewable energy source (HRES) plant with the aim of obtaining highly accurate time domain HRES plant power responses at any time during the year. The focus is on equivalent modelling for transient stability studies. Historical HRES plant production dataset and transmission network short-circuit fault statistical data, along with unsupervised data mining and deep learning techniques, represent a basis of the modelling procedure. Dynamic equivalent model (DEM) is developed in the form of deep Long Short Term Memory network. Real and imaginary parts of voltage at the point of common coupling (PCC) are model inputs, while real and reactive power at the PCC are model outputs. The paper also proposes a practical approach for selecting the adequate DEM at any time in a year based on the forecasted HRES plant operating scenario only. Model performance is evaluated on a large number of case studies using the HRES plant consisting of three renewable energy sources. The results have shown that only a few DEMs are required for representing the annual HRES plant dynamic performance and obtaining reliable transient system stability results.

Unsupervised damage assessment under varying ambient temperature based on an adjusted artificial neural network and new multivariate covariance-based distances

Temperature variability is one of the critical environmental conditions that causes confusing changes in structural properties and dynamic responses of bridges similar to damage. In this case, false alarms and mis-detection are among the major errors in health monitoring of such civil structures. High damage detectability is another significant challenge in bridge health monitoring. To deal with these issues, this article proposes an unsupervised damage assessment technique comprising two steps of data normalization and novelty detection. For the first step, an adjusted artificial neural network is considered to remove the effects of temperature variability from dynamic features (modal frequencies). This process is carried out by an auto-associative neural network by adjusting its hidden layer neurons through a new hyperparameter selection algorithm. Using normalized features obtained from the first step, this article proposes three multivariate covariance-based distances called linear dissimilarity analysis, multivariate Kullback–Leibler divergence, and multivariate Bregman distance to compute damage indices or novelty scores for damage assessment. The fundamental principles of these distances lie in three aspects: dividing the normalized features into segments, estimating the covariances of segmented feature sets, and incorporating the estimated covariances into the proposed distance measures. The major contributions of this article include proposing three non-parametric distance measures and developing an unsupervised data normalization framework via a new hyperparameter tuning algorithm for adjusting an artificial neural network. A concrete box-girder bridge is considered to verify the proposed approach, along with several comparative studies. Results show that the method presented here can mitigate severe temperature variability and increase damage detectability with superiority over some traditional and state-of-the-art damage assessment techniques.

Investigating Influence of Google-Play Application Titles on Success

The title (name) is the primary information related to a mobile (smartphone) application, as it describes its functions and services. An eye-catching title can entice customers to choose a certain application over others. Application development companies are well aware of this phenomenon and invest significant efforts in crafting their application titles with compelling keywords, phrases and topics in pursuit of higher installs. However, to the best of our knowledge, traditional literature that investigates the impact of application titles on success is limited. There may be only a few instances where scientific (data-analytical) approaches have been used to examine application titles. Moreover, these investigations of titles are dominated by supervised learning and traditional literature may lack any unsupervised (cluster) data analysis techniques to measure the impact of titles on application success. Therefore, this research work proposes an unsupervised data analysis approach based on multiple layers and algorithms. The initial layer clusters the application titles, the subsequent layer extracts various textual features from these clusters and the final layer refines the extracted attributes. In general, certain textual features in the titles are proven to be positively and negatively linked with the application installs. Verification of the results has confirmed that this proposed approach can successfully detect the most prominent features from application titles (textual data) that correlate with success.

Plasma Lipidomic Profiling Using Mass Spectrometry for Multiple Sclerosis Diagnosis and Disease Activity Stratification (LipidMS).

This investigation explores the potential of plasma lipidomic signatures for aiding in the diagnosis of Multiple Sclerosis (MS) and evaluating the clinical course and disease activity of diseased patients. Plasma samples from 60 patients with MS (PwMS) were clinically stratified to either a relapsing-remitting (RRMS) or a chronic progressive MS course and 60 age-matched controls were analyzed using state-of-the-art direct infusion quantitative shotgun lipidomics. To account for potential confounders, data were filtered for age and BMI correlations. The statistical analysis employed supervised and unsupervised multivariate data analysis techniques, including a principal component analysis (PCA), a partial least squares discriminant analysis (oPLS-DA) and a random forest (RF). To determine whether the significant absolute differences in the lipid subspecies have a relevant effect on the overall composition of the respective lipid classes, we introduce a class composition visualization (CCV). We identified 670 lipids across 16 classes. PwMS showed a significant increase in diacylglycerols (DAG), with DAG 16:0;0_18:1;0 being proven to be the lipid with the highest predictive ability for MS as determined by RF. The alterations in the phosphatidylethanolamines (PE) were mainly linked to RRMS while the alterations in the ether-bound PEs (PE O-) were found in chronic progressive MS. The amount of CE species was reduced in the CPMS cohort whereas TAG species were reduced in the RRMS patients, both lipid classes being relevant in lipid storage. Combining the above mentioned data analyses, distinct lipidomic signatures were isolated and shown to be correlated with clinical phenotypes. Our study suggests that specific plasma lipid profiles are not merely associated with the diagnosis of MS but instead point toward distinct clinical features in the individual patient paving the way for personalized therapy and an enhanced understanding of MS pathology.

A bearing remaining life prediction method under variable operating conditions based on cross-transformer fusioning segmented data cleaning

Bearing remaining useful life (RUL) prediction research based on deep learning mostly emphasizes model performance and effective feature vectors, overlooking different densities of outlier distributions in vibration signals at varying degradation stages. Moreover, forecasting models focus on capturing cross-time dependencies, ignoring the dependencies between different variables. To solve these problems, this paper proposes an unsupervised segmented data cleaning algorithm and a RUL prediction framework adaptable to variable operating conditions. The method consists of four steps: (1) Multi-domain feature extraction and selection establish a feature vector space reflecting degradation trends. (2) Segmented data cleaning divides degradation stages, using different penalty factors for outlier cleaning. (3) Cleaned vibration signals undergo a second round of multidomain feature engineering and degradation-stage division. (4) A two-stage Cross-Transformer model is used for RUL prediction. The method proposed has been validated on the prognostics and health management (PHM) bearing degradation dataset. In the constant condition prediction task, the root mean square error (RMSE) and mean absolute error (MAE) were improved to 1.88 and 5.78, respectively. In the variable condition prediction task, the proposed method outperformed existing methods, with an improvement of 59.10 % in RMSE, demonstrating strong generalization performance and practical application value.

Exposure to gestational diabetes mellitus increases subclinical inflammation mediated in part by obesity.

Gestational diabetes mellitus (GDM) is a frequent and serious complication of pregnancy, often associated with obesity. Metabolic dysfunction and metainflammation are evident in both obesity and GDM. In this cross-sectional study, we aimed at defining the direct contribution of the immune system in GDM, across the main metabolic tissues, specifically focussing on elucidating the roles of obesity and GDM to the clinical outcome. Using immunoassays and multicolour flow cytometry, cytokine profiles and immune cell frequencies were measured in maternal circulation and central metabolic tissues [placenta and visceral adipose tissue (VAT)] in GDM-diagnosed (n = 28) and normal glucose tolerant (n = 32) women undergoing caesarean section. Participants were sub-grouped as non-obese [body mass index (BMI) < 30 kg/m2] or obese (BMI ≥ 30 kg/m2). Unsupervised data analysis was performed on the flow cytometry data set to identify functional alterations. GDM obese participants had significantly elevated circulating IL-6 and IL-17A levels. GDM non-obese participants had elevated circulating IL-12p70, elevated placental IL-17A, and VAT IFN-γ production. Unsupervised clustering of immune populations across the three biological sites simultaneously, identified different NK- and T-cell phenotypes that were altered in NGT obese and GDM non-obese participants, while a classical tissue monocyte cluster was increased in GDM obese participants. In this study, there was significant evidence of subclinical inflammation, and significant alterations in clusters of NK cells, T cells, and tissue monocyte populations in GDM. While increased adiposity assimilates with increased inflammation in the non-pregnant state, this overt relationship may not be as evident during pregnancy and warrants further examination in future longitudinal studies.

SALAD: A split active learning based unsupervised network data stream anomaly detection method using autoencoders

Machine learning based intrusion detection systems monitor network data streams for cyber attacks. Challenges in this space include detecting unknown attacks, adapting to changes in the data stream such as changes in underlying behavior, the human cost of labeling data to retrain the machine learning model and the processing and memory constraints of a real-time data stream. Failure to manage the aforementioned factors could result in missed attacks, degraded detection performance, unnecessary expense or delayed detection times. This research proposes a new semi-supervised network data stream anomaly detection method, Split Active Learning Anomaly Detector (SALAD), which combines our novel Adaptive Anomaly Threshold and Stochastic Anomaly Threshold with Fading Factor methods. A novel Reconstruction Error based Distance from Threshold strategy is proposed and evaluated as part of an active stream framework to demonstrate reduction in labeling costs. The proposed methods are evaluated with the KDD Cup 1999, and UNSW-NB15 data sets, using the scikit-multiflow framework. Results demonstrated that the proposed SALAD method offered equivalent performance to full labeled and alternative Naïve Bayes (NB) and Hoeffding Adaptive Tree (HAT) methods, with a labeling budget of just 20%, significantly reducing the required human expertise to annotate the network data. Processing times of the SALAD method were demonstrated to be significantly lower than NB and HAT methods, allowing for greatly improved responsiveness to attacks occurring in real time.

Abstract 102: Microanatomy of Intracranial Human Thrombi by Single Nuclei Transcriptomics

Introduction: Microvascular dysfunction and inflammation are thought to underlie reperfusion failure following endovascular thrombectomy. Here we present the first study to produce a transcriptome-based cellular landscape of human intracranial thrombi highlighting the importance of cellular plasticity and intercellular communication at the site of occlusion. Methods: We performed single nuclei RNA-sequencing (snRNA-seq) on intracranial human thrombi harvested during revascularization. Unsupervised clustering, data visualization, differential expression and functional gene ontology/pathway analysis were utilized to deconvolve expression data into cell type-specific expression profiles. Thrombi snRNA-seq data were integrated with single cell RNA-seq data from symptomatic and asymptomatic human carotid plaques and healthy brain vasculature to contextualize thrombi in atherosclerotic disease, perform diffusion coefficient analysis, and identify ligand-receptor interactions. Results: Unbiased clustering identified six cellular populations in distinct effector states within the thrombus microenvironment. Functional gene ontology and pathway analysis revealed a monocyte subpopulation in a distinct endothelial-activation and neuroplasticity state, supporting the notion of cellular trans-differentiation and interconversion. Integration with carotid plaque data and diffusion coefficient analysis revealed shared transcriptional programs existing in a continuum, progressing from asymptomatic plaque to embolized thrombus. Ligand-receptor analysis identified macrophage mediated T-cell activation, pro-inflammatory response, immunological synapse stabilization, neuroplasticity, and monocyte trans-endothelial migration. Conclusion: Our novel application of snRNA-seq to thrombectomy specimens provides the highest resolution exploration of cellular differences and a best understanding of individual contextualized cell function to date. This detailed cellular landscape will facilitate mapping novel interventional targets with direct functional relevance and allow for greater personalization of post-reperfusion therapy.

MHCpLogics: an interactive machine learning-based tool for unsupervised data visualization and cluster analysis of immunopeptidomes.

The major histocompatibility complex (MHC) encodes a range of immune response genes, including the human leukocyte antigens (HLAs) in humans. These molecules bind peptide antigens and present them on the cell surface for T cell recognition. The repertoires of peptides presented by HLA molecules are termed immunopeptidomes. The highly polymorphic nature of the genres that encode the HLA molecules confers allotype-specific differences in the sequences of bound ligands. Allotype-specific ligand preferences are often defined by peptide-binding motifs. Individuals express up to six classical class I HLA allotypes, which likely present peptides displaying different binding motifs. Such complex datasets make the deconvolution of immunopeptidomic data into allotype-specific contributions and further dissection of binding-specificities challenging. Herein, we developed MHCpLogics as an interactive machine learning-based tool for mining peptide-binding sequence motifs and visualization of immunopeptidome data across complex datasets. We showcase the functionalities of MHCpLogics by analyzing both in-house and published mono- and multi-allelic immunopeptidomics data. The visualization modalities of MHCpLogics allow users to inspect clustered sequences down to individual peptide components and to examine broader sequence patterns within multiple immunopeptidome datasets. MHCpLogics can deconvolute large immunopeptidome datasets enabling the interrogation of clusters for the segregation of allotype-specific peptide sequence motifs, identification of sub-peptidome motifs, and the exportation of clustered peptide sequence lists. The tool facilitates rapid inspection of immunopeptidomes as a resource for the immunology and vaccine communities. MHCpLogics is a standalone application available via an executable installation at: https://github.com/PurcellLab/MHCpLogics.

Proteome and phospholipidome interrelationship of synovial fluid-derived extracellular vesicles in equine osteoarthritis: An exploratory ‘multi-omics’ study to identify composite biomarkers

Osteoarthritis causes progressive joint deterioration, severe morbidity, and reduced mobility in both humans and horses. Currently, osteoarthritis is diagnosed at late stages through clinical examination and radiographic imaging, hence it is challenging to address and provide timely therapeutic interventions to slow disease progression or ameliorate symptoms. Extracellular vesicles are cell-derived vesicles that play a key role in cell-to-cell communication and are potential sources for specific composite biomarker panel discovery. We here used a multi-omics strategy combining proteomics and phospholipidomics in an integral approach to identify composite biomarkers associated to purified extracellular vesicles from synovial fluid of healthy, mildly and severely osteoarthritic equine joints. Although the number of extracellular vesicles was unaffected by osteoarthritis, proteome profiling of extracellular vesicles by mass spectrometry identified 40 differentially expressed proteins (non-adjusted p < 0.05) in osteoarthritic joints associated with 7 significant canonical pathways in osteoarthritis. Moreover, pathway analysis unveiled changes in disease and molecular functions during osteoarthritis development. Phospholipidome profiling by mass spectrometry showed a relative increase in sphingomyelin and a decrease in phosphatidylcholine, phosphatidylinositol, and phosphatidylserine in extracellular vesicles derived from osteoarthritic joints compared to healthy joints. Unsupervised data integration revealed positive correlations between the proteome and the phospholipidome. Comprehensive analysis showed that some phospholipids and their related proteins increased as the severity of osteoarthritis progressed, while others decreased or remained stable. Altogether our data show interrelationships between synovial fluid extracellular vesicle-associated phospholipids and proteins responding to osteoarthritis pathology and which could be explored as potential composite diagnostic biomarkers of disease.

Robust object detection for autonomous driving based on semi-supervised learning

Deep learning based on labeled data has brought massive success in computer vision, speech recognition, and natural language processing. Nevertheless, labeled data is just a drop in the ocean compared with unlabeled data. How can people utilize the unlabeled data effectively? Research has focused on unsupervised and semi-supervised learning to solve such a problem. Some theoretical and empirical studies have proved that unlabeled data can help boost the generalization ability and robustness under adversarial attacks. However, current theoretical research on the relationship between robustness and unlabeled data limits its scope to toy datasets. Meanwhile, the visual models in autonomous driving need a significant improvement in robustness to guarantee security and safety. This paper proposes a semi-supervised learning framework for object detection in autonomous vehicles, improving the robustness with unlabeled data. Firstly, we build a baseline with the transfer learning of an unsupervised contrastive learning method—Momentum Contrast (MoCo). Secondly, we propose a semi-supervised co-training method to label the unlabeled data for retraining, which improves generalization on the autonomous driving dataset. Thirdly, we apply the unsupervised Bounding Box data augmentation (BBAug) method based on a search algorithm, which uses reinforcement learning to improve the robustness of object detection for autonomous driving. We present an empirical study on the KITTI dataset with diverse adversarial attack methods. Our proposed method realizes the state-of-the-art generalization and robustness under white-box attacks (DPatch and Contextual Patch) and black-box attacks (Gaussian noise, Rain, Fog, and so on). Our proposed method and empirical study show that using more unlabeled data benefits the robustness of perception systems in autonomous driving.

Applying Geo-Spatial Techniques to Determine the Temporal Variation and Conversion Pattern of Land-Use Surrounding Negombo Lagoon

Remote Sensing technology and Geographic Information Systems are the major partners in geospatial technology and they play a vital role in surface, subsurface and above earth’s studies. It provides a better platform for researchers specially in the fields of regional planning, disaster management, natural resources management etc. Observing land-use patterns in a particular area forms a basis for decision making in these fields. The study has the purpose of applying geo-spatial techniques to determine the temporal variation and conversion pattern of land-use surrounding Negombo lagoon. For that, the study has used Landsat imageries of 2003, 2009, 2015 and 2021 from the United States Geological Survey. ISO Cluster Unsupervised Classification method and data conversion tools were used to create initial layers for analysing the land-use in selected four years by using Arc GIS 10.4. software. Intersect and field calculators were used to get a better understanding of the variation of land-use through classified Landsat imageries of respective years. The results will support for measuring area of the conversion of different land-use into another and / or within the same category in a particular area throughout the time. The study concluded that, the used techniques could measure only the temporal variation of major land use types; water bodies, vegetation, built-up areas and cultivations but the minor categories of land uses were unable to detect. However, the better spatial and temporal resolution data enables to improve results to avoid such issues.

Comparative lipid composition and GC–MS fatty acid profiling of tropical African freshwater fishes

In dietary contexts, fish are renowned for the diverse array of unsaturated fatty acids they provide to the body. While data regarding the fatty acid composition of freshwater fish are readily available in most developed countries, research on African freshwater fish remains notably limited. In this study, we investigated the fatty acid composition of fifteen freshwater fish species from Africa, primarily Afrotropical regions. Lipids were extracted using a solvent mixture of methanol, water, and chloroform. Subsequently, total fatty acids were determined by saponification, and the fatty acid profile was analyzed via GC–MS following methylation in the presence of BF3 and methanol. The results revealed lipid levels in fish fillets ranging from 1.99 % for T. guineensis to 6.41 % for M. bananensis. Total fatty acid percentages of lipids were found to be 81.40 ± 9.15 % for S. nigripinnis, 83.75 ± 6.12 % for L. niloticus, and 92.99 ± 11.30 % for C. dageti. The fatty acid profiles of 15 properly identified fish species have been determined, with the majority being identified for the first time. The composition of health-important omega-3 fatty acids EPA and total DHA ranged from 4.82 ± 0.2 % for Cyprinus lepidotus to 17.43 ± 2.50 % for Mugil bananensis. Unsupervised multivariate data analysis of fatty acid profiles of the different fish species showed significant differences. When the fish were grouped into 4 categories based on their similarities, 12 discriminant fatty acids characteristic of each group were identified. Among these fatty acids are omega-6 acids such as γ-linolenic acid and arachidonic acid, and omega-3 7,10,13,16,19-docosapentaenoic acid (DPA). This research opens avenues for consumers to make informed dietary choices aligned with their specific fatty acid needs and preferences regarding fish consumption.

A Hybrid of Generative and Discriminative Models Based on the Gaussian-Coupled Softmax Layer.

Generative models offer advantageous characteristics for classification tasks, such as the availability of unsupervised data and calibrated confidence. In contrast, discriminative models have advantages in terms of their potential to outperform their generative counterparts and the simplicity of their model structures and learning algorithms. In this article, we propose a method to train a hybrid of discriminative and generative models in a single neural network (NN), which exhibits the characteristics of both models. The key idea is the Gaussian-coupled softmax layer, which is a fully connected layer with a softmax activation function coupled with Gaussian distributions. This layer can be embedded into an NN-based classifier and allows the classifier to estimate both the class posterior distribution and the input data distribution. We demonstrate that the proposed hybrid model can be applied to semi-supervised learning and confidence calibration.

Leveraging Unsupervised Data and Domain Adaptation for Deep Regression in Low-Cost Sensor Calibration.

Air quality monitoring is becoming an essential task with rising awareness about air quality. Low-cost air quality sensors are easy to deploy but are not as reliable as the costly and bulky reference monitors. The low-quality sensors can be calibrated against the reference monitors with the help of deep learning. In this article, we translate the task of sensor calibration into a semi-supervised domain adaptation problem and propose a novel solution for the same. The problem is challenging, because it is a regression problem with a covariate shift and label gap. We use histogram loss instead of mean-squared or mean absolute error (MAE), which is commonly used for regression, and find it useful against covariate shift. To handle the label gap, we propose the weighting of samples for adversarial entropy optimization. In experimental evaluations, the proposed scheme outperforms many competitive baselines, which are based on semi-supervised and supervised domain adaptation, in terms of R2 score and MAE. Ablation studies show the relevance of each proposed component in the entire scheme.

Guided principal component analysis (GPCA): a simple method for improving detection of a known analyte.

There is increasing interest in the application of Raman spectroscopy in a medical setting, ranging from supporting real-time clinical decisions e.g. surgical margins to assisting pathologists with disease classification. However, there remain a number of barriers for adoption in the medical setting due to the increased complexity of probing highly heterogeneous, dynamic biological materials. This inherent challenge can also limit the deployment of higher level analytical approaches such as Artificial Intelligence (AI) including convolutional neural networks (CNN), as there is a lack of a ground truth required for training purposes i.e. in complex clinical samples. Principal component analysis (PCA) is an unsupervised data reduction approach (orthogonal linear transformation) that has been used extensively in spectroscopy for 30+ years, due to its capability to simplify analysis of complex spectroscopic data. However, due to PCA being unsupervised features will inherently appear mixed and their rank may vary between experiments. Here we propose Guided PCA (GPCA), a simple approach that allows PCA to be guided with spectral data to ensure a consistent rank of a key target moiety by the inclusion of a reference (guiding) spectrum to the data set. This simplifies analysis, increases robustness of PCA analysis and improves quantification and the limits of detection and decreases RMSE.

An unsupervised data mining-based framework for evaluation and optimization of operation strategy of HVAC system

The evaluation and optimization of existing operation strategies are significant for energy-saving in heating, ventilation, and air conditioning (HVAC) systems. The personalized characteristics of HVAC systems, however, pose a challenge in developing the baseline. This paper proposes an unsupervised data mining-based framework for the evaluation and optimization of HVAC system operation strategies. The framework involves several steps: firstly, data preprocessing is performed to clean and reduce the raw data. Then, an improved clustering method is used to divide the external conditions of system operation, followed by an association rule algorithm to mine the evaluation baseline for operation strategies. Finally, the low-energy efficiency operating strategies can be identified and optimized using the baseline. Validity of the proposed framework is confirmed by practical cases and data. The results indicate that the mined operation evaluation baseline effectively identifies low-energy efficiency operating strategies that can be improved, resulting in energy savings of 6.9 %. Notably, optimizing the operation number of chillers and water pumps has the best energy-saving potential. Furthermore, compared to model-based operation evaluation methods, this data-driven approach offers advantages such as low computational costs and quick execution.

Lithofacies Identification from Wire-Line Logs Using an Unsupervised Data Clustering Algorithm

Stratigraphic identification from wire-line logs and core samples is a common method for lithology classification. This traditional approach is considered superior, despite its significant financial cost. Artificial neural networks and machine learning offer alternative, cost-effective means for automated data interpretation, allowing geoscientists to extract insights from data. At the same time, supervised and semi-supervised learning techniques are commonly employed, requiring a sufficient amount of labeled data to be generated through manual interpretation. Typically, there are abundant unlabeled geophysical data while labeled data are scarcer. Supervised and semi-supervised techniques partially address the cost issue. An underutilized class of machine-learning-based methods, unsupervised data clustering, can perform consonant classification by grouping similar data without requiring known results, presenting an even more cost-effective solution. In this study, we examine a state-of-the-art unsupervised data clustering algorithm called piecemeal clustering to identify lithofacies from wire-line logs, effectively addressing these challenges. The piecemeal clustering algorithm groups similar wire-log signatures into clusters, determines the number of clusters present in the data, and assigns each signature to one of the clusters, each of which represents a lithofacies. To evaluate the performance, we tested the algorithm on publicly released data from ten wells drilled in the Hugoton and Panoma fields of southwest Kansas and northwest Oklahoma, respectively. The data consist of two major groups: marine and non-marine facies. The study herein is centered around addressing two fundamental research questions regarding the accuracy and practicality of the piecemeal clustering algorithm. The algorithm successfully identified nine distinct clusters in our dataset, aligning with the cluster count observed in previously published works employing the same data. Regarding mapping accuracy, the results were notable, with success rates of 81.90% and 45.20% with and without considering adjacent facies, respectively. Further detailed analysis of the results was conducted for individual types of facies and independently for each well. These findings suggest the algorithm’s precision in characterizing the geological formations. To assess its performance, a comprehensive comparative analysis was conducted, encompassing other data clustering algorithms, as well as supervised and semi-supervised machine learning techniques. Notably, the piecemeal clustering algorithm outperformed alternative data clustering methods. Furthermore, despite its unsupervised nature, the algorithm demonstrated competitiveness by yielding results comparable to, or even surpassing, those obtained through supervised and semi-supervised techniques.

Children in All Policies (CAP) 2030 Citizen Science for Climate Change Resilience: a cross-sectional pilot study engaging adolescents to study climate hazards, biodiversity and nutrition in rural Nepal.

Young people will suffer most from climate change yet are rarely engaged in dialogue about it. Citizen science offers a method for collecting policy-relevant data, whilst promoting awareness and capacity building. We tested the feasibility and acceptability of engaging Nepalese adolescents in climate change and health-related citizen science. We purposively selected 33 adolescents from two secondary schools in one remote and one relatively accessible district of Nepal. We contextualised existing apps and developed bespoke apps to survey climate hazards, waste and water management, local biodiversity, nutrition and sociodemographic information. We analysed and presented quantitative data using a descriptive analysis. We captured perceptions and learnings via focus group discussions and analysed qualitative data using thematic analysis. We shared findings with data collectors using tables, graphs, data dashboards and maps. Adolescents collected 1667 biodiversity observations, identified 72 climate-change related hazards, and mapped 644 geolocations. They recorded 286 weights, 248 heights and 340 dietary recalls. Adolescents enjoyed learning how to collect the data and interpret the findings and gained an appreciation of local biodiversity which engendered 'environmental stewardship'. Data highlighted the prevalence of failing crops and landslides, revealed both under- and over-nutrition and demonstrated that children consume more junk foods than adults. Adolescents learnt about the impacts of climate change and the importance of eating a diverse diet of locally grown foods. A lack of a pre-established sampling frame, multiple records of the same observation and spurious nutrition data entries by unsupervised adolescents limited data quality and utility. Lack of internet access severely impacted feasibility, especially of apps which provide online feedback. Citizen science was largely acceptable, educational and empowering for adolescents, although not always feasible without internet access. Future projects could improve data quality and integrate youth leadership training to enable climate-change advocacy with local leaders.