Search

Advancements in deep learning and other approaches are making synthetic speech more natural-sounding. These advancements enable people to train a speech synthesizer with a target voice, resulting in a model that can imitate someone's voice with excellent accuracy. To overcome these challenges, in this work, we are presenting a fusion of Spectrogram and Deep Convolutional Neural Network (DeepCNN) for detecting fake audio. The work has been implemented in three phases: Spectrogram generation, feature extraction & reduction, and classification. In the first step, Erlang spectrograms have been generated using the proposed Erlang filter bank. These generated spectrograms are then passed to the Residual Network (ResNet11) for feature extraction, and later these extracted features are fed to the Linear Discriminant Analysis (LDA) for dimensionality reduction. In the last phase, LDA-optimized features were used to train the four machine learning classifiers, one by one, eXtreme Gradient Boosting (XGboost), Naïve Bayes (NB), KNN (K-Nearest Neighbor), and Random Forest (RF). The proposed work has been evaluated on the complete Fake or Real (FoR) dataset for detecting fake audio in different scenarios. The combination of Erlang spectrogram-ResNet11-LDA with XGBoost has achieved a minimum 0% and 1.4% EER for FoR original and FoR 2sec. [ABSTRACT FROM AUTHOR]

Introduction: Machine learning-based power transformer fault diagnosis methods often grapple with the challenge of imbalanced fault case distributions across different categories, potentially degrading diagnostic accuracy. To address this issue and enhance the accuracy and operational efficiency of power transformer fault diagnosis models, this paper presents a novel fault diagnosis model that integrates Neighborhood Component Analysis (NCA) and k-Nearest Neighbor (KNN) learning, with the incorporation of correction factors. Methods: The methodology begins by introducing a correction factor into the objective function of the NCA algorithm to reduce the impact of sample imbalance on model training. We derive a sample parameter correlation quantization matrix from oil chromatography fault data using association rules, which serves as the initial value for the NCA algorithm's training metric matrix. The metric matrix obtained from training is then applied to perform a mapping transformation on the input data for the KNN classifier, thereby reducing the distance between similar samples and enhancing KNN classification performance. Hyperparameter tuning is achieved through the Bayesian optimization algorithm to identify the model parameter set that maximizes test set accuracy. Results: Analysis of the transformer fault case library reveals that the model proposed in this paper reduces diagnostic time by nearly half compared to traditional machine learning diagnosis models. Additionally, the accuracy for minority sample classes is improved by at least 15% compared to other models. Discussion: The integration of NCA and KNN with correction factors not only mitigates the effects of sample imbalance but also significantly enhances the operational efficiency and diagnostic accuracy of power transformer fault diagnosis. The proposed model's performance improvements highlight the potential of this approach for practical applications in the field of power transformer maintenance and diagnostics. [ABSTRACT FROM AUTHOR]

Classification of adenocarcinoma (AC) and squamous cell carcinoma (SCC) poses significant challenges for cytopathologists, often necessitating clinical tests and biopsies that delay treatment initiation. To address this, we developed a machine learning-based approach utilizing resected lung-tissue microbiome of AC and SCC patients for subtype classification. Differentially enriched taxa were identified using LEfSe, revealing ten potential microbial markers. Linear discriminant analysis (LDA) was subsequently applied to enhance inter-class separability. Next, benchmarking was performed across six different supervised-classification algorithms viz. logistic-regression, naïve-bayes, random-forest, extreme-gradient-boost (XGBoost), k-nearest neighbor, and deep neural network. Noteworthy, XGBoost, with an accuracy of 76.25%, and AUROC (area-under-receiver-operating-characteristic) of 0.81 with 69% specificity and 76% sensitivity, outperform the other five classification algorithms using LDA-transformed features. Validation on an independent dataset confirmed its robustness with an AUROC of 0.71, with minimal false positives and negatives. This study is the first to classify AC and SCC subtypes using lung-tissue microbiome. [ABSTRACT FROM AUTHOR]

This study introduces a hybrid data assimilation method that significantly improves the predictive accuracy of the time-dependent Susceptible-Exposed-Asymptomatic-Infected-Quarantined-Removed (SEAIQR) model for epidemic forecasting. The approach integrates real-time Ensemble Kalman Filtering (EnKF) with the K-Nearest Neighbors (KNN) algorithm, combining dynamic real-time adjustments with pattern recognition techniques tailored to the specific dynamics of epidemics. This hybrid methodology overcomes the limitations of single-model predictions in the face of increasingly complex transmission pathways in modern society. Numerical experiments conducted using COVID-19 case data from Xi'an, Shaanxi Province, China (December 9, 2021, to January 8, 2022) demonstrate a marked improvement in forecasting accuracy relative to traditional models and other data assimilation approaches. These findings underscore the potential of the proposed method to enhance the accuracy and reliability of predictive models, providing valuable insights for future epidemic forecasting and disease control strategies. [ABSTRACT FROM AUTHOR]

Power load forecasting is an important part of modern smart grid operation management. Accurate forecasting guides the efficient and stable operation of the power system. In this paper, a fuzzy C-means clustering algorithm and an improved locally weighted linear regression model are proposed for short-term power load forecasting. First, the fuzzy C-means clustering algorithm is used to cluster the power load. Make the power consumption behavior of load data of the same category similar and use the power consumption load data of the same category as the training sample. Then, to solve the problem of large calculation and insufficient fitting of the locally weighted linear regression model, the k-nearest neighbor range constraint is introduced into the model for daily load forecasting. Finally, the effectiveness of the method is verified by a simulation example. Experimental results show that this method can effectively improve the accuracy and generalization ability of power load forecasting compared with other methods. [ABSTRACT FROM AUTHOR]

The objective was to utilize nine machine learning (ML) methods to predict the prognosis of antibody positive autoimmune encephalitis (AE) patients. The encephalitis data from the Global Burden of Disease (GBD) study is analyzed to reflect the disease burden of encephalitis. This study included 187 patients with AE. 121 patients as training set and 67 patients as validation set. Decision trees (DT), random forest (RF), extreme gradient boosting (XGBoost), k-nearest neighbor (KNN), support vector machine (SVM), naive bayes (NB), neural network (NN), light gradient boosting machine (LGBM), and logistic regression (LR) are ML methods used to construct predictive models. The constructed models were validated for discrimination, calibration and clinical applicability using validation set data. Shapley additive explanation (SHAP) analysis was used to explain the model. The number of encephalitis worldwide deaths, incidence and prevalence is increasing every year from 2010 to 2021. The training set included 121 patients with AE. Univariate analysis and LASSO screening identified six variables. The results of constructing models using 9 ML methods showed RF had the highest accuracy (0.860), followed by XGBoost (0.826), with F1 scores of 0.844 and 0.807, respectively. Validation set data showed good discrimination, calibration and clinical applicability of the model. The SHAP values of infection, CSF monocyte percentage, and prealbumin were 0.906, 0.790, and 0.644, respectively. As a rare disease, the sample size of this study is relatively small. The model constructed using RF and XGBoost has good performance, good discrimination, calibration, clinical applicability, and interpretability. • Reported the latest encephalitis data from the Global Burden of Disease Study • Build models for predicting the prognosis of patients with AE • Nine ML algorithms are used to construct models. • Shapley additive explanation analysis model [ABSTRACT FROM AUTHOR]

Customer turnover is a crucial issue in banking since maintained profitability depends on keeping clients. This work aims to categorize consumer turnover in banks by using a new ensemble approach combining many machine learning methods, hence enhancing churn prediction models. Using a comprehensive dataset including demographic, financial, and behavioral data—such as credit score, account balance, tenure, and activity levels—the study employs the goal variable revealing if a customer has left the bank. The study starts with univariate, bivariate, and multivariate feature exploration and subsequently uses the Interquartile Range (IQR) approach to identify outliers thereby improving the data quality. Five models—K-Nearest Neighbors, Support Vector Classifier, Random Forest, Decision Tree, and XGBoost—a Voting Classifier ensemble—are used to estimate project churn. Building upon all the strengths of each model, this approach improves the prediction of classification and provides a balanced and highly robust classification system. The applied approaches are K-Nearest Neighbors (KNN), Support Vector Classifier (SVC), Random Forest, Decision Tree, and XGBoost within a Voting Classifier configuration. The performance of the Voting Classifier without SMOTE yields the following results: Accuracy: 0.87, precision: 0.87, recall: 0.80, and F1-Score: 0.87. The proposed model that extend the base model using SMOTE (Synthetic Minority Over-sampling Technique), yields a higher prediction accuracy of 0.90, precision of 0.90, recall of 0.90 and F1-Score of 0.90. This enhancement is proving the efficiency of SMOTE to handle the class imbalance problem in order to render the churn prediction more balanced and reliable system. The proposed approach assures a reliable solution to the strategies to retain the customers in the banking organisations. [ABSTRACT FROM AUTHOR]

This study aims to identify the geometry of weld beads from point cloud data to enhance robotic grinding planning. Robotic grinding is widely used in industry due to the hazardous working conditions and health risks associated with manual operations. However, certain grinding tasks require precise recognition of the work area for effective path planning. The processing and identification of weld bead data from point clouds remain active research challenges because of varying geometries. By using a precise single-line laser vision system, many studies have focused on fitting the base metal surface to extract weld beads. Nevertheless, advanced machine learning algorithms offer the potential for more efficient classification of weld bead points. This study compares the performance of three classification algorithms, random forest, k-nearest neighbors, and support vector machine, in identifying weld bead profiles and positions. Both flat and curved base metal surfaces were analyzed with the proposed inputs from the filtered point data of the scanned line profiles. These inputs included (1) the vertical distance to the next point, (2) the vertical distance to the highest point, (3) the horizontal distance to the highest point, (4) the point height relative to the scanner datum, and (5) the slope to the next point for each point on the profiles. The proposed inputs enhanced performance, and the results show that the random forest algorithm achieved the best accuracy of 98.98% for peripheral welds and 98.78% for parallel welds. The standard deviations of the measured maximum heights were 0.0129 mm for peripheral welds and 0.1613 mm for parallel welds, with respective maximum heights of 2.354 mm and 3.375 mm. Furthermore, a grinding error of just 0.066 mm was achieved, underscoring the effectiveness of these algorithms in accurately identifying weld bead geometries for robotic grinding applications. [ABSTRACT FROM AUTHOR]

Electric vehicles (EVs) rely heavily on lithium-ion battery packs as essential energy storage components. However, inconsistencies in cell characteristics and operating conditions can lead to imbalanced state of charge (SOC) levels, resulting in reduced capacity and accelerated degradation. This study presents an active cell balancing method optimized for both charging and discharging scenarios, aiming to equalize SOC across cells and improve overall pack performance. The proposed system includes two balancing strategies: a charging balance that redistributes excess charge from high-SOC cells to maximize capacity, and a discharging balance that addresses low-SOC cells to extend discharge duration. Experimental results confirm that this method effectively reduces SOC disparities, enhancing both charging and discharging capacities. Additionally, to accurately predict battery lifespan and remaining useful life (RUL), seven machine learning models are evaluated using R-squared (R2) and Mean Absolute Error (MAE) metrics. Among these, k-nearest Neighbors and Random Forest models deliver the highest accuracy, achieving R2 values of 0.996 and above with low MAE, demonstrating strong predictive capability. The integration of active balancing and RUL prediction enables a feedback loop where balanced SOC levels promote battery health, and RUL predictions inform optimal balancing strategies. This comprehensive approach advances EV battery management, enhancing lifespan and reliability through proactive balancing and predictive insights. [ABSTRACT FROM AUTHOR]

Hypervelocity breakup events contribute to the rapidly growing population of space debris orbiting Earth. To ensure a safe environment for future space missions, it is crucial to accurately characterize resulting fragments, which constitute a major portion of the orbital population but are often too small for conventional tracking methods. Currently, the Space Surveillance Network tracks satellites and debris within a specific altitude range and releases the data publicly as Two-Line Elements, though these datasets are limited in both size and accuracy. To supplement Two-Line Element data, this paper presents a novel Deep Neural Network approach to estimate orbital elements of hypervelocity fragments resulting from simulated satellite explosions. It employs initial conditions collected from realistic terrestrial explosions considered to be relative to 5 different detonation points on the polar orbit of the weather satellite, NOAA-16, to model the initial states of debris fragments immediately following an explosion. The debris trajectories are then propagated using a high-precision semi-analytic integration method, Analytic Continuation, considering J 2 − J 6 zonal gravitational terms and Earth's atmospheric drag perturbations. The collected data is used to train a Deep Neural Network for each of the orbital elements. Finally, a testing data set is used to validate the results, finding that this technique accurately estimates most of the orbital elements, but is not suited to predict the true anomaly of the fragments. Therefore, K-Nearest Neighbor regression is utilized instead to predict the nature of the final orbital element. • Two-Line Elements alone are unreliable and often inaccurate. • Terrestrial data used to represent satellite debris from explosions. • Propagating satellite debris using perturbed Analytic Continuation. • Deep Neural Networks estimate orbital elements except true anomaly. • K-Nearest Neighbors implemented to estimate true anomaly. [ABSTRACT FROM AUTHOR]

Currently, the accurate prediction of tunnel boring machine (TBM) performance remains a considerable challenge due to the complex interactions between the TBM and rock mass. In this study, the research work is based on part of a metro tunnel project that covers 2,083.94 m. The Gaussian mixture model (GMM) and K-nearest neighbor algorithm (KNN) are used to classify and predict the rock mass drillability in the TBM excavation process. Drillability indexes are introduced to cluster and classify the rock mass, including the penetration (P), field penetration index (FPI), torque penetration index (TPI), and specific energy (SE). Statistical characteristics of the drillability indexes were analyzed, and it was found that their distributions did not conform to the normal distribution, with large variation coefficients. Clustering analysis was then conducted on the TPI and FPI within the training group using the Gaussian mixture model , and six drillability categories of rock mass were classified. Subsequently, the mapping relationship between the cutterhead speed, advance speed, total advance force, and cutterhead torque in the training group and the drillability of rock mass was established based on the KNN classification model. It was revealed that when the K-value is set to 4, the model has high macro - F 1, macro - P , and macro - R. Validated by the testing group data, this method has been proven to be feasible and effective. The research results indicate that this method can effectively classify and predict the drillability of tunneling surrounding rock mass in shield construction, particularly when the rock mass at the shield face is uniform and homogeneous. This provides a theoretical basis and technical support for safe and efficient shield tunneling. [ABSTRACT FROM AUTHOR]

This paper presents a machine learning (ML) methodology grounded in quantitative structure–property relationship (QSPR) principles for the prediction of corrosion inhibition efficiency (CIE) values, explicitly focusing on pyridazine inhibitor compounds. The training phase incorporates the kernel density estimation (KDE) function to generate virtual samples, aiming to enhance the prediction accuracy of the ML model. The study evaluates the performance of three models, namely gradient boosting (GB), random forest (RF), and k-nearest neighbor (KNN). The results exhibit a substantial enhancement in predictive ability following the incorporation of virtual samples. Specifically, coefficient of determination (R2) values for GB, RF, and KNN models increase from − 0.33 to 0.97, − 0.20 to 0.96, and − 0.17 to 0.95, respectively, with the addition of 1000 virtual samples. Correspondingly, root mean square error (RMSE) values for each model experience a significant decrease, reducing from 9.20 to 1.57, 9.07 to 1.81, and 8.60 to 2.12, respectively. This augmentation enhances the correlation between features and targets, resulting in more accurate predictions and eliminating the necessity for feature selection. Furthermore, it implies resilience to model variations, eliminating the need for model selection. The proposed methodology is a crucial link between theoretical research and experimental synthesis, providing a reliable and accurate prediction tool. This tool proves instrumental in efficiently designing and exploring corrosion inhibitor candidates, thereby contributing to the advancement of effective corrosion inhibition strategies. [ABSTRACT FROM AUTHOR]

Cognition is crucial to brain function, and accurately classifying cognitive load is essential for understanding the psychological processes across tasks. This paper innovatively combines functional near-infrared spectroscopy (fNIRS) with eye tracking technology to delve into the classification of cognitive load at the neurocognitive level. This integration overcomes the limitations of a single modality, addressing challenges such as feature selection, high dimensionality, and insufficient sample capacity. We employ fNIRS-eye tracking technology to collect neural activity and eye tracking data during various cognitive tasks, followed by preprocessing. Using the maximum relevance minimum redundancy algorithm, we extract the most relevant features and evaluate their impact on the classification task. We evaluate the classification performance by building models (naive Bayes, support vector machine, K-nearest neighbors, and random forest) and employing cross-validation. The results demonstrate the effectiveness of fNIRS-eye tracking, the maximum relevance minimum redundancy algorithm, and machine learning techniques in discriminating cognitive load levels. This study emphasizes the impact of the number of features on performance, highlighting the need for an optimal feature set to improve accuracy. These findings advance our understanding of neuroscientific features related to cognitive load, propelling neural psychology research to deeper levels and holding significant implications for future cognitive science. [ABSTRACT FROM AUTHOR]

Working memory plays an important role in cognitive science and is a basic process for learning. While working memory is limited in regard to capacity and duration, different cognitive tasks are designed to overcome these difficulties. This study investigated information flow during a novel visual working memory task in which participants respond to exaggerated and normal pictures. Ten healthy men (mean age 28.5 ± 4.57 years) participated in two stages of the encoding and retrieval tasks. The electroencephalogram (EEG) signals are recorded. Moreover, the adaptive directed transfer function (ADTF) method is used as a computational tool to investigate the dynamic process of visual working memory retrieval on the extracted event-related potentials (ERPs) from the EEG signal. Network connectivity and P300 sub-components (P3a, P3b, and LPC) are also extracted during visual working memory retrieval. Then, the nonparametric Wilcoxon test and five classifiers are applied to network properties for features selection and classification between exaggerated-old and normal-old pictures. The Z-values of Ge is more distinctive rather than other network properties. In terms of the machine learning approach, the accuracy, F1-score, and specificity of the k-nearest neighbors (KNN), classifiers are 81%, 77%, and 81%, respectively. KNN classifier ranked first compared with other classifiers. Furthermore, the results of in-degree/out-degree matrices show that the information flows continuously in the right hemisphere during the retrieval of exaggerated pictures, from P3a to P3b. During the retrieval of visual working memory, the networks associated with attentional processes show greater activation for exaggerated pictures compared to normal pictures. This suggests that the exaggerated pictures may have captured more attention and thus required greater cognitive resources for retrieval. [ABSTRACT FROM AUTHOR]

Chillers are one of the biggest energy consumption devices in HVAC systems. Abnormal operation may undermine the performance, energy efficiency, and environment. This study comprehensively explores the hybrid applications of deep convolutional neural network (CNN) in chiller fault diagnosis and fault feature extraction. Unlike the computer vision where the feature locations are fixed, for chiller fault, it can be changed. The effect of feature sequence on diagnostic performance is carefully investigated, and found that it depends on the size and number of the convolution kernels. Small and large number of kernels may extract fine and enough features for the model to counter the location change and maintain the basic characteristics of the faults. 1-D CNN is further studied as a hierarchical feature extractor combined with the traditional machine learning, like k-nearest neighbor (KNN), decision tree (DT), and random forest (RF), to build a hybrid strategy. It is found that the highest accuracy of 99.85% is achieved by RF plus 1-D CNN with an accuracy of 100% for refrigerant leakage. Fine and clear features from deeper structure are most favorable for the weak learner like DT, but may harm the information diversity of RF and lower its accuracy. [ABSTRACT FROM AUTHOR]

Human factors contribute in some way to almost 93% of road crashes. Because the unavailability of good pre-crash data, the contribution of human factors to safety-critical events (SCEs) and the prediction of crashes using real-world data is lightly researched. This study provides predictive accuracy by harnessing unique real-world naturalistic driving study (NDS) data, which includes dynamic pre-crash information about driving behavior and performance. After cleaning and preprocessing, a final subsample (N = 9,237) was used and split into training and test samples. For consistent comparison of variables' importance in statistical and machine learning (ML) models, the dominance analysis uncovered the most important predictors used by the ordered Probit model. Next, three non-parametric supervised ML methods, because promising prediction performance and cost-effectiveness, including Naïve Bayes, K-Nearest Neighbors, and Gradient Boosting Tree (GBT) were used. The overall out-of-sample prediction accuracy for the ordered Probit model was 85.75% which was lower than all three ML methods. The GBT showed the highest (91.23%) out-of-sample prediction accuracy. The availability of pre-crash naturalistic data helps significantly improve the prediction accuracy of SCEs as cumulative importance for all available human factors in the GBT classifier was 94%. For practical applications, refer to the article. [ABSTRACT FROM AUTHOR]

Currently, web services‐based applications have an important presence in public and private organizations. The vulnerabilities that these types of applications may have pose an inherent potential risk to the business model of these organizations. These applications have the inherent risk of being used by organizations in such a way that their activity is affected and they become the main entry point for attackers who want to breach their security. The main barrier to this type of attack are web application firewalls (WAF), which are responsible for processing Hypertext Transfer Protocol requests between clients and web servers, classifying them and rejecting malicious requests. This type of (WAF) applications, for the most part, have regular expressions that correspond to general rules and allow detecting malicious requests that follow a pattern contained in them. However, due to the knowledge of these rules by attackers, it is easy to circumvent security and to impersonate a malicious request by an innocuous request. Therefore, in this article, we present a study of different models based on artificial intelligence techniques as Naïve Bayes, k‐nearest neighbors, support vector machines, and linear regression to test their effectiveness in detecting malicious requests from a synthetic dataset containing more than 100,000 requests. The results obtained show that the implementation of these methods optimize the detection of malicious requests obtaining results between 92% and 99% of success in their classification. [ABSTRACT FROM AUTHOR]

COVID-19, originating in Wuhan, China, in December 2019 declared a pandemic by the World Health Organization on March 11, 2020, rapidly spread worldwide, significantly impacting human life and the health sector. This study aims to develop a WHO (World Health Organization) oriented disease severity prediction model using laboratory and demographic data from COVID-19 patients upon admission to Marmara University Hospital. The relationship between oxygen and intensive care needs with laboratory results on the data set was analyzed using K-nearest neighbor, Bagging, Random Forest and Decision Tree machine learning methods. The dataset's unbalanced class distribution was balanced using the SMOTE method, and the impact of data multiplication on classification performance was evaluated. In the data set without SMOTE, the patient's oxygen requirement during the first hospitalization was estimated with 16 features at 91.67% accuracy, the oxygen requirement at hospitalization with 18 features at 91.96%, and the intensive care need at hospitalization with 12 features at 92.17% accuracy. After SMOTE data multiplication, an increase of 6%, 24% and 21% was observed in F1-Score values, respectively. This study significantly contributes to the field by utilizing machine learning methods on patient data, essential for COVID-19 diagnosis, monitoring, and clinical management through required laboratory tests. makine öğrenmesi laboratuvar verisi prognoz [ABSTRACT FROM AUTHOR]

Developmental coordination disorder is a neurodevelopmental disorder characterized by a marked deterioration in the development of motor and coordination skills that significantly impairs with daily activity and academic performance. Accurate diagnosis of the disease requires extensive testing and long-term observations. These tests and observations can be time-consuming, expensive, incomplete, inaccurate, and subjective. EEG is a strong candidate to overcome these limitations. In this study, an EEG-based deep learning model is presented to support experts in detecting developmental coordination disorder in children from EEG signals. The dataset consists of EEG signals recorded from 16 children without developmental coordination disorder and 16 children with developmental coordination disorder. By calculating the power spectral density values of frequencies between 1-49 hertz of EEG signals using periodogram, welch, and multitaper methods, 49 feature vectors were extracted for each of the spectral analysis methods. Then, the performances of support vector machine (SVM), random forest (RF), k-nearest neighbors (kNN) and longshort-term memory (LSTM) algorithms were compared. After the experiments, the model integrating welch and LSTM algorithm showed the highest performance. The proposed deep learning model achieved promising performance with 0.984 sensitivity, 0.959 specificity, 0.962 precision, 0.973 f1-score, 0.944 Matthews correlation coefficient, and 97.20% accuracy. [ABSTRACT FROM AUTHOR]

Internet of Things (IoT) refers to the infrastructures that connect smart devices to the Internet, operating autonomously. This connectivity makes it possible to harvest vast quantities of data, creating new opportunities for the emergence of unprecedented knowledge. To ensure IoT securit, various approaches have been implemented, such as authentication, encoding, as well as devices to guarantee data integrity and availability. Among these approaches, Intrusion Detection Systems (IDS) is an actual security solution, whose performance can be enhanced by integrating various algorithms, including Machine Learning (ML) and Deep Learning (DL), enabling proactive and accurate detection of threats. This study proposes to optimize the performance of network IDS using an ensemble learning method based on a voting classification algorithm. By combining the strengths of three powerful algorithms, Random Forest (RF), K-Nearest Neighbors (KNN), and Support Vector Machine (SVM) to detect both normal behavior and different categories of attack. Our analysis focuses primarily on the NSL-KDD dataset, while also integrating the recent Edge-IIoT dataset, tailored to industrial IoT environments. Experimental results show significant enhancements on the Edge-IIoT and NSL-KDD datasets, reaching accuracy levels between 72% to 99%, with precision between 87% and 99%, while recall values and F1-scores are also between 72% and 99%, for both normal and attack detection. Despite the promising results of this study, it suffers from certain limitations, notably the use of specific datasets and the lack of evaluations in a variety of environments. Future work could include applying this model to various datasets and evaluating more advanced ensemble strategies, with the aim of further enhancing the effectiveness of IDS. [ABSTRACT FROM AUTHOR]

In the rapidly evolving landscape of television advertising, optimizing ad schedules to maximize viewer engagement and revenue has become significant. Traditional methods often operate in silos, limiting the potential insights gained from broader data analysis due to concerns over privacy and data sharing. This article introduces a novel approach that leverages Federated Learning (FL) to enhance TV ad schedule optimization, combining the strengths of local optimization techniques with the power of global Machine Learning (ML) models to uncover actionable insights without compromising data privacy. It combines linear programming for initial ads scheduling optimization with ML—specifically, a K-Nearest Neighbors (KNN) model—for predicting ad spot positions. Taking into account the diversity and the difficulty of the ad-scheduling problem, we propose a prescriptive-predictive approach in which first the position of the ads is optimized (using Google's OR-Tools CP-SAT) and then the scheduled position of all ads will be the result of the optimization problem. Second, this output becomes the target of a predictive task that predicts the position of new entries based on their characteristics ensuring the implementation of the scheduling at large scale (using KNN, Light Gradient Boosting Machine and Random Forest). Furthermore, we explore the integration of FL to enhance predictive accuracy and strategic insight across different broadcasting networks while preserving data privacy. The FL approach resulted in 8750 ads being precisely matched to their optimal category placements, showcasing an alignment with the intended diversity objectives. Additionally, there was a minimal deviation observed, with 1133 ads positioned within a one-category variance from their ideal placement in the original dataset. [ABSTRACT FROM AUTHOR]

In recent years, gait-based emotion recognition has been widely applied in the field of computer vision. However, existing gait emotion recognition methods typically rely on complete human skeleton data, and their accuracy significantly declines when the data is occluded. To enhance the accuracy of gait emotion recognition under occlusion, this paper proposes a Multi-scale Suppression Graph Convolutional Network (MS-GCN). The MS-GCN consists of three main components: Joint Interpolation Module (JI Moudle), Multi-scale Temporal Convolution Network (MS-TCN), and Suppression Graph Convolutional Network (SGCN). The JI Module completes the spatially occluded skeletal joints using the (K-Nearest Neighbors) KNN interpolation method. The MS-TCN employs convolutional kernels of various sizes to comprehensively capture the emotional information embedded in the gait, compensating for the temporal occlusion of gait information. The SGCN extracts more non-prominent human gait features by suppressing the extraction of key body part features, thereby reducing the negative impact of occlusion on emotion recognition results. The proposed method is evaluated on two comprehensive datasets: Emotion-Gait, containing 4227 real gaits from sources like BML, ICT-Pollick, and ELMD, and 1000 synthetic gaits generated using STEP-Gen technology, and ELMB, consisting of 3924 gaits, with 1835 labeled with emotions such as "Happy," "Sad," "Angry," and "Neutral." On the standard datasets Emotion-Gait and ELMB, the proposed method achieved accuracies of 0.900 and 0.896, respectively, attaining performance comparable to other state-of-the-art methods. Furthermore, on occlusion datasets, the proposed method significantly mitigates the performance degradation caused by occlusion compared to other methods, the accuracy is significantly higher than that of other methods. [ABSTRACT FROM AUTHOR]

The development of big data analysis technology has brought new development opportunities to the production and management of various industries. Through the mining and analysis of various data in the operation process of enterprises by big data technology, the internal associated data of the enterprises and even the entire industry can be obtained. As a common method for large-scale data statistical analysis, clustering technology can effectively mine the relationship within massive heterogeneous multidimensional data, complete unlabeled data classification, and provide data support for various model analysis of big data. Common big data density clustering methods are time-consuming and easy to cause errors in data density allocation, which affects the accuracy of data clustering. Therefore we propose a novel large data density peak clustering based on sparse auto-encoder and data space meshing via evidence probability distribution. Firstly, the sparse auto-encoder in deep learning is used to achieve feature extraction and dimensionality reduction for input high-dimensional data matrix through training. Secondly, the data space is meshed to reduce the calculation of the distance between the sample data points. When calculating the local density, not only the density value of the grid itself, but also the density value of the k nearest neighbors are considered, which reduces the influence of the subjective selection truncation distance on the clustering results and improves the clustering accuracy. The grid density threshold is set to ensure the stability of the clustering results. Using the K-nearest neighbor information of the sample points, the transfer probability distribution strategy and evidence probability distribution strategy are proposed to optimize the distribution of the remaining sample points, so as to avoid the joint error of distribution. The experimental results show that the proposed algorithm has higher clustering accuracy and better clustering performance than other advanced clustering algorithms on artificial and real data sets. [ABSTRACT FROM AUTHOR]

Microplastics (MPs) are exudated fragments and fibers from environmental plastic refuse in the lower millimeter to micrometer range with some intentionally created particles and degraded exudates from anthropogenic sources. These particles are often hard to identify due to extensive weathering and material heterogeneity from additives. Leveraging the ability of machine learning (ML) models, which train on features from selected polymer classes, can aid in determining particle identity despite heterogeneity. Herein, a 1,800 spectral training dataset was employed for model training (900 from µ-FTIR and 900 from µ-Raman) using data from common MP polymers (synthetically or naturally derived) and plastics. A threefold battery; synthetic data (SD), mixed synthetic data (MSD), and real-world data from an FTIR Library of Plastic Particles sourced from the Environment (FLOPP-E) and a [Raman] Spectral Library of Plastic Particles aged in the Environment (SLOPP-E); was administered for reliability. Firstly, the SD test determined subspace k-nearest neighbors (SKNN) and wide neural network (WNN) as champions (µ-FTIR and µ-Raman, respectively) with an accuracy of, 99%/100% and 98%/100% (χ2 = 31.99/69, p =.0024/ <.0001). Secondly, the MSD test served as a progenitor to the multi-class prediction from µ-Raman's SKNN showing consistency across 5 replicates (H =.25—6.59, p =.156—.993). And thirdly, the real-world test exhibited a loss in accuracy rate with only the champion, SKNN, retaining ~ 73% and ~ 49% of the correct predictions. Investigation into the probable causes of misprediction via colorant additives led to the discovery of white and blue as the most unpredictable across both databases. Lastly, an investigation of 2 samples revealed confounding colorant additives (SKNN's predictions) as copper phthalocyanine (C2. Blue Fiber) and a derivative from the family of diketo-pyrrolo-pyrroles (Polyester 12. Red Fiber). [ABSTRACT FROM AUTHOR]

With the rise of the Internet of Things (IoT) and Fifth Generation (5G) networks, real-time health monitoring has become more efficient, enabling continuous tracking of cardiac health. Electrocardiograms (ECG) are essential for diagnosing cardiovascular diseases, but classifying ECG signals presents challenges due to their high-dimensional features. Standard machine learning classifiers often struggle to achieve high accuracy, and IoT devices face resource constraints that hinder real-time processing. This paper introduces a novel methodology that combines Particle Swarm Optimization (PSO)-based feature selection with machine learning classifiers, such as K-Nearest Neighbors (KNN), Random Forest (RF), Decision Trees (DT), and Support Vector Machines (SVM). The proposed method uses the accuracy of the machine learning classifiers as the fitness function in the PSO algorithm. This ensures that the selected features are optimal and well-suited for the classifier, improving both classification accuracy and computational efficiency. The approach was validated using the MIT-BIH Arrhythmia dataset, achieving 98% accuracy with PSOSVM and 84% without PSO-based feature selection. The dimensionality of the ECG dataset is reduced from 4000 features to 888, improving classification accuracy and computational efficiency. These results outperform current machine learning and deep learning methods, demonstrating the effectiveness of the proposed approach for arrhythmia detection. This research provides a scalable solution for IoT-enabled, 5G-powered health monitoring systems, enhancing both classification performance and real-time processing in resource-constrained environments. [ABSTRACT FROM AUTHOR]

As the prevalence of Internet of Things (IoT) devices increases, Cyber incidents are also increasing significantly. These Cyber incidents are mainly caused by various attacks, such as Distributed Denial of Service (DDoS), Denial of Service (DoS), intrusions, and web-based attacks. This type of attacks can severely impact valuable IoT system resources, compromise stored data, and lead to substantial financial losses if not adequately mitigated. Detecting these attacks within network traffic is complex and requires intelligent Intrusion Detection Systems (IDS). This paper proposes a Machine Learning (ML) based hybrid IDS model for edge computing networks. The feature selection process employs the 'Recursive Feature Elimination technique' (RFE) combined with 'Random Forest' (RF) to identify optimal features for attack detection. The Hybrid IDS model integrates 'Random Forest' (RF), 'Decision Tree' (DT), 'Extra Tree' (ET), and 'K-Nearest Neighbor' (KNN) algorithms. The Hybrid IDS model is evaluated on four datasets: 'CIC-IDS-2017', 'NSL-KDD', 'UNSW-NB15', and 'CSE-CIC-IDS-2018'. The results of the proposed model show maximum prediction accuracy of 99.92%, 99.89%, 99.50%, and 99.13%, and F1-score values obtained are 99.95%, 99.90%, 99.23%, and 99.13% on 'CIC-IDS-2017', 'NSL-KDD', 'UNSW-NB15', and 'CSE-CIC-IDS2018' datasets, respectively. The experimental results clearly demonstrate that the proposed model performs better than the models reported in the existing studies. [ABSTRACT FROM AUTHOR]

COVID19 otherwise called Severe Acute Respiratory Syndrome Corona virus-2 is an infectious illness. Another transmittable infection called Pneumonia is mainly attributable to infection because of bacteria in the alveoli of the lungs. Once a diseased lung tissue has infection, it elevates excretion in it. Specialists conduct health examinations and identify the patient through ultrasound, biopsy, or Chest X-ray of lungs to identify whether the patient has these diseases. Incorrect treatment, misdiagnosis, and if the disease was disregarded will result in the fatality. The development of Deep Learning and neutrosophic set (NS) supports the decision-making procedure of professionals to identify patients with this disease. NS is a prolongation of the fuzzy set and classical theories. The NS determines three memberships such as T, I and F. T, I, and F display the degree of truth, the false, and the indeterminacy membership, correspondingly. This enables a more nuanced representation of contradiction, uncertainty, and ambiguity within the dataset, allowing superior handling of imprecise and complex data. This study develops a new Deep learning with Neutrosophic Set-Based k-Nearest Neighbors Classifier for disease detection (DLNSKNN-DD) technique. The major purpose of the DLNSKNN-DD method is to identify the existence of virus pneumonia and COVID-19. In the DLNSKNN-DD technique, the feature extraction from the medical images is carried out by residual network (ResNet50v2). Moreover, the parameter tuning of the ResNetv2 model is done using Adadelta optimizer. The DLNSKNN-DD technique exploits NSKNN model for classification purposes. The performance evaluation of the DLNSKNN-DD algorithm can be assessed on medicinal image dataset. The experimental outcomes underlined the effectual recognition results of the DLNSKNN-DD technique on the identification of diseases [ABSTRACT FROM AUTHOR]

This study evaluates flood susceptibility in the Fez-Meknes region of Morocco by comparing the performance of five machine learning (ML) models using 14 environmental variables. The selected models, including random forest (RF), support vector machine (SVM), K-Nearest neighbor (KNN), recursive partitioning and regression trees (RPART), and logistic regression (LR), were assessed for prediction accuracy and enhanced with partial dependence plots (PDP) and local interpretable model-agnostic explanations (LIME) to increase interpretability. Results indicate that the RF model outperforms other models, achieving a high prediction accuracy with an AUC of 96%, low mean absolute error (MAE) of 0.26, and root mean squared error (RMSE) of 0.31, along with strong Nash Sutcliffe efficiency (NSE) and correlation coefficient (R²). Through PDP and LIME, the primary factors influencing flood susceptibility were identified as proximity to rivers, drainage density, slope, normalized difference vegetation index (NDVI), terrain roughness index (TRI), and land use and land cover (LULC). These findings highlight the potential of interpretable ML models to enhance flood risk assessment, providing valuable insights for urban planning and flood mitigation strategies in vulnerable regions. [ABSTRACT FROM AUTHOR]

Skin cancer is most top three critical kinds of cancer due to damaged DNA, which is cause death. This damaged DNA begins cells for growing uncontrollably and currently it can be obtaining improved quickly. It is several researches on the computerized examination of malignancy from the skin cancer image. But, study of these images are very difficult taking several troublesome issues such as light reflections on the skin surface, differences from the color illumination, sizes of lesions, and distinct shapes. Thus, the outcome, evidential automatic detection of skin cancer are appreciated for developing the accuracy and efficiency of pathologists at the beginning phases. This manuscript develops a Stacked Ensemble Machine Learning based Skin Cancer Detection and Classification (SEML-SKCDC) approach. The presented SEML-SKCDC technique majorly aims to offer ensemble of three ML models for skin cancer classification. In the presented SEML-SKCDC technique, median filtering and contrast enhancement is performed at the pre-processing stage. To generate feature vectors, the honey badger algorithm (HBA) with EfficientNet method has been exploited in this work. At last, an ensemble of k-nearest neighbor (KNN), random forest (RF), and feed forward neural network (FFNN) approaches are applied for skin cancer classification. The simulation evaluation of the SEML-SKCDC system on skin cancer database depicts the developments of the SEML-SKCDC algorithm with recent methods. [ABSTRACT FROM AUTHOR]

In the evolving landscape of the Internet of Things (IoT), effective intrusion detection is paramount for maintaining security and data integrity. This study introduces a hybrid heuristic technique utilizing artificial intelligence for enhancing intrusion detection systems (IDS) in IoT environments. By integrating various machine learning models, the research focuses on training, tuning, and validating a sequential neural network to predict intrusion occurrences based on extensive data analysis. The methodology involves modelling, which starts with training machine learning algorithms to predict labels from features, tuning the models to meet organizational requirements, and validating them using holdout data. Key machine learning techniques explored include logistic regression, k-nearest neighbors (KNN), naive Bayes, support vector machines (SVM), decision trees, random forests, and neural networks. Each technique's applicability to classification tasks, particularly binary and multivariate scenarios, is discussed in the context of enhancing IDS capabilities. A sequential neural network model, comprising multiple dense and dropout layers, was developed and trained with 148,033 parameters to achieve high accuracy and robustness. The architecture's effectiveness in learning intricate patterns associated with malicious activities while avoiding overfitting is emphasized. The study demonstrates the model's proficiency in binary classification tasks, which is critical for distinguishing between normal and anomalous behaviors in IoT systems. The results indicate that the neural network, optimized using the hybrid heuristic approach, shows a significant reduction in validation loss and a steady improvement in accuracy over multiple epochs. Despite initial overfitting signs, the model maintains high performance on unseen data, underscoring the importance of ongoing model assessment and tuning. [ABSTRACT FROM AUTHOR]

Early identification of hypertension is crucial to prevent its serious complications, which can lead to devastating health effects by threatening lifestyle quality and significantly increasing premature mortality. This study aims to evaluate the effectiveness of machine learning techniques in predicting the presence of hypertension from an unbalanced dataset consisting of 4,363 records and 35 features. To balance the dataset, we employed the synthetic minority over-sampling technique (SMOTE) algorithm. In addition, to select the most relevant features, we used ant colony optimization. Next, we applied various algorithms, including logistic regression (LR), K-nearest neighbors (KNNs), support vector machine (SVM), extra trees (ETs), and AdaBoost (AB). We also evaluated the optimization of hyperparameters using two methods: Bayesian optimization (BO) and particle swarm optimization (PSO). The results reveal that the combination of AB with BO demonstrated superior performance, with an accuracy of 97.60%, a recall of 98.93%, and a precision of 98.59%. This research emphasizes the potential of machine learning techniques for anticipating hypertension and highlights the importance of optimization techniques in improving predictive models' performance. [ABSTRACT FROM AUTHOR]

In modern times, social networks have become ubiquitous platforms facilitating widespread information dissemination, resulting in significant daily data generation. This increase in data production encompasses a wide range of user-generated content, which in turn promotes the proliferation of fraudulent users creating fake profiles and engaging in deceptive activities. This article aims to address this challenge by employing machine learning algorithms to accurately identify fake profiles. The research involves a thorough analysis of various user behaviors, engagement metrics, and content attributes within social platforms. The primary goal is to develop robust models capable of effectively detecting deceptive profiles by meticulously examining user activities and content characteristics. The study explores the application of robust methodologies such as K-means and K-medoids clustering, alongside supervised machine learning classifiers including K-nearest neighbors (KNN), support vector machine (SVM), Bernoulli Naïve Bayes (NB), logistic regression, and linear support vector classification (SVC), specifically tailored for the detection of fake profiles. [ABSTRACT FROM AUTHOR]

Introduction: Motor Imagery (MI) Electroencephalography (EEG) signals are non-stationary and dynamic physiological signals which have low signal-to-noise ratio. Hence, it is difficult to achieve high classification accuracy. Although various machine learning methods have already proven useful to that effect, the use of many features and ineffective EEG channels often leads to a complex structure of classifier algorithms. State-of-the-art studies were interested in improving classification performance with complex feature extraction and classification methods by neglecting detailed EEG channel and feature investigation in predicting MI tasks from EEGs. Here, we investigate the effects of the statistically significant feature selection method on four different feature domains (time-domain, frequency-domain, time-frequency domain, and non-linear domain) and their two different combinations to reduce the number of features and classify MI-EEG features by comparing low-dimensional matrices with well-known machine learning algorithms. Methods: Our main goal is not to find the best classifier performance but to perform feature and channel investigation in MI task classification. Therefore, the detailed investigation of the effect of EEG channels and features is implemented using a statistically significant feature distribution on 22 EEG channels for each feature set separately. We used the BCI Competition IV Dataset IIa and 288 samples per person. A total of 1,364 MI-EEG features were analyzed in this study. We tested nine distinct classifiers: Decision tree, Discriminant analysis, Logistic regression, Naive Bayes, Support vector machine, k-Nearest neighbor, Ensemble learning, Neural networks, and Kernel approximation. Results: Among all feature sets considered, classifications performed with non-linear and combined feature sets resulted in a maximum accuracy of 63.04% and 47.36% for binary and multiple MI task predictions, respectively. The ensemble learning classifier achieved the maximum accuracy in almost all feature sets for binary and multiple MI task classifications. Discussion: Our research thus shows that the statistically significant feature-based feature selection method significantly improves the classification performance with fewer features in almost all feature sets, enabling detailed and effective EEG channel and feature investigation. [ABSTRACT FROM AUTHOR]

The accurate and timely assessment of wheat freshness is not only a complex scientific endeavor but also a critical aspect of grain storage safety. This study introduces an innovative approach for evaluating wheat freshness by integrating machine learning algorithms with Biophoton Analytical Technology (BPAT). Initially, spontaneous ultraweak photon emissions from wheat are measured, and various statistical descriptors are derived to construct a feature vector. Particle Swarm Optimization (PSO) is then utilized to determine the optimal parameters for the Support Vector Machine (SVM). To validate the efficacy of the proposed method, additional machine learning techniques such as K-Nearest Neighbors (KNN), Multilayer Perceptron (MLP), and decision trees are employed. Experimental results demonstrate that both the machine learning algorithms and input features significantly influence model performance. Notably, using only central tendency factor features yields commendable recognition outcomes, eliminating the need for variability factor features. This research offers a novel perspective on the quantitative evaluation of wheat freshness. [ABSTRACT FROM AUTHOR]

Currently, pain assessment using electroencephalogram signals and machine learning methods in clinical studies is of great importance, especially for those who cannot express their pain. Since newborns are among the high-risk group and always experience pain at the beginning of birth, in this research, the severity of newborns has been investigated and evaluated. Other studies related to the annoyance of newborns have used the EEG signal of newborns alone; therefore, in this study, the intensity of newborn pain was measured using the electroencephalogram signal of 107 infants who were stimulated by the heel lance in three levels: no pain, low pain and moderate pain were recorded as a single trial and evaluated. The support vector machine (SVM), K-Nearest Neighbors (KNN) and Ensemble bagging classifiers were trained using the K-fold cross-validation method and features of the brain's time-frequency domain. The results were obtained with accuracies of 72.8 ± 2, 84.4 ± 1.3 and 82.9 ± 1.6%, respectively. Also, in examining the problem of distinguishing pain and no pain, the electroencephalogram signal of 74 infants was evaluated, and similar to the three-class mode, with the 10-fold validation method, we reached the highest accuracy of 100% in Bagging classifier and 98.6 ± 0.1 accuracy in KNN and SVM classifiers. [ABSTRACT FROM AUTHOR]

Blasting excavation is widely used in mining, tunneling and construction industries, but it leads to produce ground vibration which can seriously damage the urban communities. The peak particle velocity (PPV) is one of main indicators for determining the extent of ground vibration. Owing to the complexity of blasting process, there is controversy over which parameters will be considered as the inputs for empirical equations and machine learning (ML) algorithms. According to current researches, the burden has controversial impact on the blast-induced ground vibration. To judge whether the burden affects blast-induced ground vibration, the data of ground vibration considering burden have been recorded at the Wujiata coal mine. Correlation coefficient is used to analyze the relationship between variables, the correlation between the distance from blasting center to monitored point (R) and peak particle velocity (PPV) is greatest and the value of correlation coefficient is − 0.67. This study firstly summarizes the most common empirical equations, and a new empirical equation is established by dimension analysis. The new equation shows better performance of predicting PPV than most other empirical equations by regression analysis. Secondly, the machine learning is confirmed the applicability of predicting PPV. Based on the performance assessments, regression error characteristic curve and Uncertainty analysis in the first round of predicting PPV, the random forest (RF) and K-Nearest Neighbors (KNN) show better performance than other four machine learning algorithms. Then, in the second round, based on the artithmetic optimization algorithm (AOA), the optimized random forest (AOA-RF) model as the most accurate model compared with the optimized K-Nearest Neighbors (AOA-KNN) presented in the literature. Finally, the points of predicted PPV which have been informed of danger are marked based on Chinese safety regulations for blasting. [ABSTRACT FROM AUTHOR]

The development of data analysis techniques and intelligent systems has had a considerable impact on education, and has seen the emergence of the field of educational data mining (EDM). The Early Warning System (EWS) has been of great use in predicting at-risk students or analyzing learners' performance. Our project concerns the development of an early warning system that takes into account a number of socio-cultural, structural and educational factors that have a direct impact on a student's decision to drop out of school. We have worked on an original database dedicated to this issue, which reflects our approach of seeking exhaustiveness and precision in the choice of dropout indicators. The model we built performed very well, particularly with the K-Nearest Neighbor (KNN) algorithm, with an accuracy rate of over 99.5% for the training set and over 99.3% for the test set. The results are visualized using a Django application we developed for this purpose, and we show how this can be useful for educational planning. [ABSTRACT FROM AUTHOR]

Background: Heart failure (HF) and atrial fibrillation (AF) usually coexist and are associated with a poorer prognosis. This study aimed to develop a model to predict in-hospital mortality in patients with HF combined with AF. Methods: Patients with HF and AF were obtained from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database from 2008 to 2019. Feature selection was based on the Mann-Whitney U test and the least absolute shrinkage and selection operator (LASSO) regression model. Random Forest, eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LGBM), K-Nearest Neighbor (KNN) models, and their stacked model (the stacking ensemble model) were established. The area under of the curve (AUC) with 95% confidence interval (CI), sensitivity, specificity, as well as accuracy were applied to assess the performance of the predictive models. Results: A total of 5,998 patients with HF combined with AF were included, of which 4,198 patients were assigned to the training set and 1,800 to the testing set (7:3). Among these 4,198 patients, 624 (14.86%) died in-hospital and 3,574 (85.14%) survived. Twenty-two features were used to construct the predictive model. Among these four single models, the AUC was 0.747 (95%CI: 0.717–0.777) for the Random Forest model, 0.755 (95%CI: 0.725–0.785) for the XGBoost model, 0.754 (95%CI: 0.724–0.784) for the LGBM model, and 0.746 (95%CI: 0.716–0.776) for the KNN model in the testing set. The stacking ensemble model had the highest AUC compared to the four single models, with AUCs of 0.837 (95%CI: 0.821–0.852) and 0.768 (95%CI: 0.740–0.796) for the training set and testing set, respectively. Conclusion: The stacking ensemble model showed a good predictive effect in predicting in-hospital mortality in patients with HF combined with AF and may provide clinicians with a reference tool for early identification of mortality risk. [ABSTRACT FROM AUTHOR]

Background: Psoriatic arthritis (PSA) is an inflammatory joint disease associated with psoriasis (PSO) that can be easily missed. Existing PSA screening tools ignore objective serologic indicators. The aim of this study was to develop a disease screening model and the Psoriatic Arthritis Inflammation Index (PSAII) based on serologic data to enhance the efficiency of PSA screening. Method: A total of 719 PSO and PSA patients from the National Health and Nutrition Examination Survey (NHANES) (as training set and test set) and 135 PSO and PSA patients who were seen at The First Affiliated Hospital of Zhejiang Chinese Medical University (as external validation set) were selected, 31 indicators for these patients were collected as potential input features for the model. Least Absolute Shrinkage and Selection Operator (LASSO) was used to identify PSA-related features. Five models of logistic regression (LR), random forest, k-nearest neighbor, gradient augmentation and neural network were developed in the training set using quintuple cross validation. And we developed PSAII based on the results of LASSO regression and weights of logistic model parameters. All performance metrics are derived on the test set and the external validation set. Results: Five variables were selected to build models, including age, lymphocyte percentage, neutrophil count, eosinophilic count, and C-reactive protein. In all established models, the LR model performed the best, with an Area Under Curve (AUC) of 0.87 (95% confidence interval (CI): 0.83-0.90) on the test set; on the external validation set the AUC was 0.82 (95%CI: 0.74-0.90). The PSAII formula was PSAII = percentage of lymphocytes × C-reactive protein/(neutrophil count × eosinophilic count × 10). The AUC of PSAII in the test is 0.93 (95%CI: 0.88-0.97), and the cutoff value is 18. The AUC of the external validation set is 0.81 (95%CI: 0.72-0.89). Conclusions: This study developed and validated five models to assist screening for PSA by analyzing serum data from NHANES and Chinese populations. The LR model demonstrated the best performance. We created PSAII for PSA screening. However, the high false positive rate of PSAII makes it necessary to combine it with other PSA screening tools when applied. [ABSTRACT FROM AUTHOR]

Early prediction of pest occurrences can enhance crop production, reduce input costs, and minimize environmental damage. Advances in machine learning algorithms facilitate the development of efficient pest alert systems. Furthermore, ensemble algorithms help in the utilization of several models rather than being dependent on a single model. This study introduces a dynamic ensemble model with absolute log error (ALE) and logistic error functions using four machine learning models—artificial neural networks (ANNs), support vector regression (SVR), k-nearest neighbors (kNN), and random forests (RF). Various abiotic factors such as minimum and maximum temperature, rainfall, and morning and evening relative humidity were incorporated into the model as exogenous variables. The proposed algorithms were compared with fixed-weighted and unweighted ensemble methods, and candidate machine learning models, using the pest population data for yellow stem borer (YSB) from two regions of India. Error metrics include the root mean square log error (RMSLE), root relative square error (RRSE), and median absolute error (MDAE), along with the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) algorithm. This study concluded that the proposed dynamic ensemble algorithm demonstrated better predictive accuracy in forecasting YSB infestation in rice crops. [ABSTRACT FROM AUTHOR]

Ransomware poses a significant threat to Android devices, presenting a pressing concern in the realm of malware. While there has been extensive research on malware detection, distinguishing between various malware categories remains a challenge. Notably, ransomware often disguises its behavior to resemble less harmful forms of malware like adware, evading conventional security measures. Therefore, there is a critical need for advanced malware category detection techniques to elucidate specific behaviors unique to each malware type and bolster detection efficacy. This paper aims to enhance Android ransomware detection by investigating the optimal combination of static features (such as permissions, intents, and API calls) and dynamic features (captured from network traffic flow) that contribute to minimize false negatives when applying supervised machine learning classification. Our research also aims to discern the pivotal features essential for accurate ransomware detection. To this end, we propose a model integrating feature selection techniques and employing various machine learning classifiers, including decision trees, k-nearest neighbors, random forest, gradient boosting, and bagging. The model was implemented in Python, and its evaluation was conducted with and without k-fold validation to offer a broader range of explored behaviours. Our findings highlight the efficacy of combining network-Permission and network-API features, exhibiting superior ransomware detection rates compared to other feature combinations. Moreover, our model achieved recall scores of 99.2 and 97% before and after employing cross-validation, respectively. We also identified 6 API features, 27 network features, and 18 permission features as the most useful ones for ransomware detection in Android. [ABSTRACT FROM AUTHOR]

This study proposes an AI-based framework to predict the quality of results from the deep drawing process without requiring considerable amounts of actual data. A few-shot learning method with a Siamese network was employed for training with limited experimental data and virtual simulation data. As the base of the Siamese network, three network architectures, fully connected network (FCN), convolutional neural network (CNN), and long short-term memory (LSTM), were considered. By increasing the ratio of the experimental data, the best accuracy for each network was determined. Finally, the K-nearest neighbor algorithm was trained to predict the drawing quality using the embedding vector from the Siamese network. In all networks, it was confirmed that the accuracy of the classification algorithm improved as more experimental data were added. At this time, the classification accuracy of the CNN was 100%, which was significantly superior to the 77.8% of the FCN and 88.9% of the LSTM. Finally, the AI-based framework proposed in this paper can be effectively applied to various press forming processes as a method for predicting the quality of deep drawing molded products. [ABSTRACT FROM AUTHOR]

The agricultural sector contributes significantly to greenhouse gas emissions, which cause global warming and climate change. Numerous mathematical models have been developed to predict the greenhouse gas emissions from agriculture. However, the database utilized for prediction has thousands of missing values and imbalanced data records due to various factors, such as environmental disasters, sensor failure, maintenance issues, and budgetary constraints. Many researchers have either completely discarded records with missing values or have only partially addressed the issue, leading to less precise predictions. This study proposes a machine learning-based ensemble approach that uses multiple imputation by chained equations (MICE), K-nearest neighbors (KNN), and MissForest techniques to impute the missing values by reducing uncertainties and improves robustness in predictions. Furthermore, Synthetic Minority Oversampling Technique (SMOTE R) addresses data imbalance in CO2 emissions prediction by generating synthetic data to balance the target variable's distribution in the regression problem. Thereby the proposed SMOTE R and ensemble imputation approach aims to improve prediction accuracy and reliability by tackling data completeness and distribution issues simultaneously. The experiments are conducted on the GRACEnet database shows that the proposed approach outperforms the existing work in measure of R2, MAE, and RMSE metrics. [ABSTRACT FROM AUTHOR]

In the realm of wireless sensor networks (WSNs), precise localization of sensor nodes and attack detection has emerged as a significant challenge. Accurate localization of sensor nodes and detection of attacks is crucial for enhancing the network lifetime and preventing the data from unauthorized access. However, prior research has encountered various obstacles, such as high localization error, more time consumption, and poor detection performance. To overcome these difficulties, this research proposes a novel Recurrent Crossover-based Dynamic Differential (RC-DD) algorithm for effective localization and attack detection in WSN. The proposed model utilizes a weighted K-Nearest Neighbor (KNN) with Mahalanobis distance for accurately localizing the position of the sensor nodes within the network. Moreover, the Recurrent Neural Network (RNN) is employed for effectively detect multiple attacks, and Dynamic Differential Annealed Optimization (DDAO) with crossover strategy is implemented for fine-tuning the RNN’s hyperparameter, thereby enhancing the attack detection accuracy. The experimental outcome shows that the proposed RC-DD framework effectively localizes the sensor nodes and detects the attacks with a higher accuracy of 98.9% compared to existing methodologies. Additionally, the proposed framework minimizes localization error and energy consumption, underscoring its potential to enhance the security and lifetime of the network. [ABSTRACT FROM AUTHOR]

Screening of an appropriate enhanced oil recovery (EOR) technique is important for the maximum reservoir recovery and economics of an oil exploration and production project. The conventional approaches for selecting EOR are often time-consuming and may fail to produce satisfactory results for the complex reservoirs. This paper focuses on implementing of selected five supervised machine learning (ML) techniques for accurate screening of EOR methods for complex reservoirs. Two of them are never used before for EOR screening. A global database of 358 successful EOR projects is collected of which 176 are used in the present study to train Neural Network (NN), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Gaussian Naive Bayes, and Random Forest (RF) classifier and tested their efficiencies. The result shows that the RF classification technique predicts the most suitable EOR with an accuracy value of 0.91 which is the highest accuracy among all other techniques. The study also ranks the reservoir and fluid parameters depending on their influences on the EOR screening. Viscosity is found to be the most impacting factor in the selection of an EOR technique with a feature importance of 34.6%. [ABSTRACT FROM AUTHOR]

Diabetes is a prevalent and costly condition, with early diagnosis pivotal in mitigating ‎its progression and complications. The diagnostic process often contends with data ‎ambiguity and decision uncertainty, adding complexity to achieving definitive ‎outcomes. This study addresses the diabetes diagnostic challenge through data mining ‎and machine learning techniques. It involves training various machine learning ‎algorithms and conducting statistical analysis on a dataset comprising 520 patients, ‎encompassing both normal and diabetic cases, to discern influential features.‎ Incorporating 17 features as classifier inputs, this research evaluates the diagnostic ‎performance using four reputable techniques: support vector machine (SVM), random ‎forest (RF), multi-layer perceptron (MLP), and k-nearest neighbor (kNN). The outcomes ‎underscore the SVM model's superior performance, boasting accuracy, specificity, and ‎sensitivity values of 98.78±1.96%, 99.28±1.63%, and 97.32±2.45%, ‎respectively, across 50 iterations. The findings establish SVM as the preferred method ‎for diabetes diagnosis.‎ This study highlights the efficacy of data mining and machine learning models in ‎diabetes diagnosis. While these methods exhibit respectable predictive accuracy, their ‎integration with a physician's assessment promises even better patient outcomes.‎ [ABSTRACT FROM AUTHOR]

Technological advancements have significantly progressed from the inception of computers and the internet to the rise of artificial intelligence (AI), profoundly impacting various sectors such as business, education, and healthcare. Among these advancements, ChatGPT has emerged as a prominent conversational AI tool, facilitating tasks such as sentiment prediction and sector-specific decision-making. This study aims to evaluate ChatGPT's performance in analyzing public sentiment using data from X (formerly Twitter) and to propose an ethical framework for its sector-specific applications. By combining the BERT transformer model with BiLSTM, Random Forest, and K-Nearest Neighbor algorithms, the study achieves a hybrid approach to sentiment prediction, demonstrating superior performance with an accuracy of 86.7%. Furthermore, the study explores ethical considerations, such as fairness, accountability, and transparency, to address ChatGPT's adoption across industries. The findings offer insights into both the technical and ethical dimensions of ChatGPT's integration into business, education, and healthcare sectors, emphasizing its potential for sustainable and responsible use. [ABSTRACT FROM AUTHOR]

Background: The rapid onset of COVID-19 placed immense strain on many already overstretched healthcare systems. The unique physiological changes in pregnancy, amplified by the complex effects of COVID-19 in pregnant women, rendered prioritization of infected expectant mothers more challenging. This work aims to use state-of-the-art machine learning techniques to predict whether a COVID-19-infected pregnant woman will be admitted to ICU (Intensive Care Unit). Methods: A retrospective study using data from COVID-19-infected women admitted to one hospital in Astana and one in Shymkent, Kazakhstan, from May to July 2021. The developed machine learning platform implements and compares the performance of eight binary classifiers, including Gaussian naïve Bayes, K-nearest neighbors, logistic regression with L2 regularization, random forest, AdaBoost, gradient boosting, eXtreme gradient boosting, and linear discriminant analysis. Results: Data from 1292 pregnant women with COVID-19 were analyzed. Of them, 10.4% were admitted to ICU. Logistic regression with L2 regularization achieved the highest F1-score during the model selection phase while achieving an AUC of 0.84 on the test set during the evaluation stage. Furthermore, the feature importance analysis conducted by calculating Shapley Additive Explanation values points to leucocyte counts, C-reactive protein, pregnancy week, and eGFR and hemoglobin as the most important features for predicting ICU admission. Conclusions: The predictive model obtained here may be an efficient support tool for prioritizing care of COVID-19-infected pregnant women in clinical practice. [ABSTRACT FROM AUTHOR]

Point cloud registration is pivotal across various applications, yet traditional methods rely on unordered point clouds, leading to significant challenges in terms of computational complexity and feature richness. These methods often use k-nearest neighbors (KNN) or neighborhood ball queries to access local neighborhood information, which is not only computationally intensive but also confines the analysis within the object's boundary, making it difficult to determine if points are precisely on the boundary using local features alone. This indicates a lack of sufficient local feature richness. In this paper, we propose a novel registration strategy utilizing ordered point clouds, which are now obtainable through advanced depth cameras, 3D sensors, and structured light-based 3D reconstruction. Our approach eliminates the need for computationally expensive KNN queries by leveraging the inherent ordering of points, significantly reducing processing time; extracts local features by utilizing 2D coordinates, providing richer features compared to traditional methods, which are constrained by object boundaries; compares feature similarity between two point clouds without keypoint extraction, enhancing efficiency and accuracy; and integrates image feature-matching techniques, leveraging the coordinate correspondence between 2D images and 3D-ordered point clouds. Experiments on both synthetic and real-world datasets, including indoor and industrial environments, demonstrate that our algorithm achieves an optimal balance between registration accuracy and efficiency, with registration times consistently under one second. [ABSTRACT FROM AUTHOR]