Search

This study calls for a broadening of the perspective on academic success. While passing exams is an essential objective of higher education, it should not overshadow another important objective which is the development of students' skills, such as becoming curious, autonomous and reflective in the learning process. This study used Academic Performance in Exams (APE) and Deep Approach to Learning (DAL) as measures related to these two objectives. The aim was to identify and compare the factors that may influence APE and DAL. The study was conducted on first-year students (2011) at a French university. It was based on a random forest algorithm and took into account a wide range of factors belonging to different dimensions: demographics, social background, educational background, context of the educational programme, behavioural engagement, social environment, psychological and cognitive characteristics. The results show that the most important factors in predicting APE are the educational programme undertaken, student's educational background and parents' occupation. DAL was not found to be an important factor in APE. Regarding the prediction of DAL, the results point to the predominant weight of intrinsic motivation and the important weight of elaborated epistemic beliefs. In contrast, demographics and behavioural engagement were found to have negligible weight in predicting both APE and DAL. These findings raise questions about the type of success that is valued in the first year of university and call for reflection on assessment methods. They also allow the identification of levers that teachers can activate to support first year students. [ABSTRACT FROM AUTHOR]

Heart diseases are a major cause of death worldwide, highlighting the need for early detection. The electrocardiogram (ECG) records the heart's electrical activity using electrodes. Our research focuses on the ECG data to diagnose heart disorders, particularly arrhythmias. We utilized the MIT-BIH arrhythmia dataset for comparative analysis of various machine learning techniques, including random forest, K-Nearest Neighbor, and Decision Tree, along with deep learning algorithms like Long short-term memory and Convolutional Neural Networks. This required employing various preprocessing methods like filtering and normalization and feature selection techniques such as chi-square and sequential feature selectors to improve the performance of heart disease prediction. Therefore, hybrid machine and deep learning models are proposed, and the results reveal that hybrid models perform better than conventional models. [ABSTRACT FROM AUTHOR]

The evaluation of balance and postural stability holds significant importance in both medical rehabilitation and daily life. However, the clinical method is hindered by the inconvenience of immobility and relatively high costs associated with the force platforms. Wearable sensors, such as accelerometers, have emerged as an alternative solution, overcoming the limitations of traditional force platforms. Thus, the purpose of this study is to utilize data obtained from a low-cost, portable, small-sized IMU (specifically an accelerometer) to predict indicators derived from force platform devices. A miniaturized and portable acceleration test equipment was proposed. Together with the random forest algorithm, our classification method achieved classification results with accuracy, recall, precision, f1-score, and specificity scores above 95%, This study provides a more portable and highly accurate tool for assessing balance ability. [ABSTRACT FROM AUTHOR]

This study presents a new tool for predicting the compressive strength of alkali-activated concrete (AAC) based on its binder mineralogy. It was made using a machine-learning (ML) framework. To achieve this challenging task, the authors collected 45 data sources from the literature to build a data set of 809 samples that included nine effective features, such as binder content, alkaline-to-binder ratio, binder chemical compositions (CaO, SiO2 , Al2O3 , and MgO contents), NaOH molarity and percentage in alkaline solution, age, and compressive strength. To assess the accuracy of the prediction tool, the authors trained and evaluated the data set using the most relevant ML methods: Lasso regression, random forest regression, decision tree, AdaBoost, extreme gradient boosting (XGB), and long short-term memory with recurrent neural network (LSTM-RNN). An experimental program was also conducted using the ML approaches to further validate the accuracy of the predictions. Overall, the XGB and LSTM-RNN methods were observed to significantly outperform the other methods in terms of accuracy when predicting compressive strength. Particularly impressive results were seen, with R2 values of 0.93 and 0.96 for compressive strength prediction being recorded. Further analysis of the binder mineralogy showed that increasing calcite content percentage led to an increase in AAC compressive strength, whereas increasing silicates in the binder mineralogy caused a decrease in AAC compressive strength due to the shortage of calcites. The Shapley additive explanations (SHAP) analysis revealed that calcite and silicate had the highest SHAP values for the AAC compressive strength. In contrast, the Al2O3 and MgO percentages had only a minor impact on the compressive strength of AAC. [ABSTRACT FROM AUTHOR]

Speech enables clear and powerful idea transmission. The human voice, rich in tone and emotion, holds unique beauty and significance in daily life. Vocal pitches vary by gender and are influenced by emotions and languages. While people naturally perceive these nuances, machines often struggle to capture these subtle distinctions. Machines may struggle to detect these nuances, but people effortlessly perceive them. This project aims to use various machine learning (ML) and deep learning (DL) techniques to reliably determine an individual's gender from a corpus of Bengali conversations. Our dataset comprises 3185 Bengali speeches, with 1100 delivered by males, 1035 by women, and 1050 by those who identify as third gender. We employed six distinct feature extraction techniques to examine the audio data: roll-off, spectral centroid, chroma-stft, spectral bandwidth, zero crossing rate, and Mel-frequency cepstral coefficients (MFCC). Extreme gradient boosting (XGBoost), support vector machines (SVM), Knearest neighbors (KNN), decision trees classifier (DTC), and random forest (RF) were employed as the five ML algorithms to comprehensively analyze the dataset. For a full study, we also included 1D convolutional neural networks (CNN) from the DL area. The 1D CNN performed extraordinarily well, exceeding the accuracy of all other algorithms with a stunning 99.37%. [ABSTRACT FROM AUTHOR]

Purpose: In this study, we investigated the performance of deep learning (DL) models to differentiate between normal and glaucomatous visual fields (VFs) and classify glaucoma from early to the advanced stage to observe if the DL model can stage glaucoma as Mills criteria using only the pattern deviation (PD) plots. The DL model results were compared with a machine learning (ML) classifier trained on conventional VF parameters. Methods: A total of 265 PD plots and 265 numerical datasets of Humphrey 24–2 VF images were collected from 119 normal and 146 glaucomatous eyes to train the DL models to classify the images into four groups: normal, early glaucoma, moderate glaucoma, and advanced glaucoma. The two popular pre-trained DL models: ResNet18 and VGG16, were used to train the PD images using five-fold cross-validation (CV) and observed the performance using balanced, pre-augmented data (n = 476 images), imbalanced original data (n = 265) and feature extraction. The trained images were further investigated using the Grad-CAM visualization technique. Moreover, four ML models were trained from the global indices: mean deviation (MD), pattern standard deviation (PSD) and visual field index (VFI), using five-fold CV to compare the classification performance with the DL model's result. Results: The DL model, ResNet18 trained from balanced, pre-augmented PD images, achieved high accuracy in classifying the groups with an overall F1-score: 96.8%, precision: 97.0%, recall: 96.9%, and specificity: 99.0%. The highest F1 score was 87.8% for ResNet18 with the original dataset and 88.7% for VGG16 with feature extraction. The DL models successfully localized the affected VF loss in PD plots. Among the ML models, the random forest (RF) classifier performed best with an F1 score of 96%. Conclusion: The DL model trained from PD plots was promising in differentiating normal and glaucomatous groups and performed similarly to conventional global indices. Hence, the evidence-based DL model trained from PD images demonstrated that the DL model could stage glaucoma using only PD plots like Mills criteria. This automated DL model will assist clinicians in precision glaucoma detection and progression management during extensive glaucoma screening. [ABSTRACT FROM AUTHOR]

The lake-floodplain wetlands are characterized by high biodiversity, difficult access, and significant environmental changes. Traditional remote sensing mapping methods struggle to generate consistent time-series data on wetland vegetation communities. Current research has endeavoured to address this issue through the application of deep learning methodologies. However, a significant limitation of these models is their reliance on a substantial volume of training samples, which contradicts the difficulty and high cost of obtaining samples from the lake-floodplain wetlands. Whether it is possible to construct a transferable deep learning model under small sample conditions and apply it to the mapping of lake-floodplain wetlands is an urgent issue that needs to be addressed. To solve this problem, this study first constructed a deep neural network (DNN) designed specifically for mapping complex lake-floodplain wetland vegetation under conditions of limited sample size. Subsequently, using 2021 as a reference year, a novel histogram threshold method was proposed to identify the unchanged samples for the target transfer years of 2019, 2020, 2022, and 2023. Finally, annual wetland vegetation mapping was performed in Poyang Lake using DNN and sample transfer learning (STL). The results showed that high-quality annual time-series data of wetland vegetation can be generated using the constructed DNN and STL, with all overall accuracies exceeding 80%. The histogram threshold method, which combines SAD and NDVI indicators of key phenological period, can effectively solve the problem of difficulty in determining the unchanged samples in transfer learning for heterogeneous lake wetlands. Furthermore, the performance of STL based on the constructed DNN model was significantly superior to those based on support vector machine and random forest algorithms for mapping annual wetland vegetation communities using limited training samples. This study demonstrates that the effective application of DNN and STL will be highly beneficial for long-term monitoring of vegetation in lake-floodplain wetlands, particularly where sample availability is limited. [ABSTRACT FROM AUTHOR]

Skin cancer is considered globally as the most fatal disease. Most likely all the patients who received wrong diagnosis and low-quality treatment die early. Though if it is detected in the early stages the patient has fairly good chance and the aforementioned diseases can be cured. Consequently, diagnostic identification and management of the patient at this level becomes a rather enormous task. This paper offers a cutting-edge hybrid deep learning approach of better segmentation and classification of skin lesions. The proposed method incorporates three key stages: preprocessing, segmentation of lesions, and classification of lesions. By the stage of preprocessing, a morphology-based technique takes out hair so as to enhance the segmentation precision to use the cleansed images for subsequent analysis. Segmentation cuts off the lesion from the surrounding skin, giving the classification phase a dedicated area of interest and the ability to clear the background noise that may affect classification rates. The isolation enables the model to better analyze anatomical lesion features in order to achieve accurate benign and malignant classifications. Using ResUNet++, the cutting-edge deep learning architecture, we achieved accurate lesion segmentation. Next, we will modify and use an AlexNet-Random Forest (AlexNet-RF) based classifier for robust lesion classification. The proposed hybrid deep learning model is intensively validated on the Ham10000 data set which is one of the most popular datasets for skin lesions analysis. The obtained results show that the utilized approach, compared to the previous ones, is more effective, giving better segmentation and classification results. This method takes advantage of ResUNet++ strong classification skill and modified AlexNet-Random Forest robustness for more accurate segmentation. There is a high probability that ResUNet++, which is highly proficient at medical image segmentation, can produce better segmentation of lesions than the simpler models. The composition of AlexNet's extraction of features with Random Forest ability to reduce overfitting possibly may be more precise in the classification when compared to using only one model. [ABSTRACT FROM AUTHOR]

Tidal flat ecosystems, are under steady decline due to anthropogenic pressures including sea level rise and climate change. Monitoring and managing these coastal systems requires accurate and up-to-date mapping. Sediment characteristics and macrozoobenthos are major indicators of the environmental status of tidal flats. Field monitoring of these indicators is often restricted by low accessibility and high costs. Despite limitations in spectral contrast, integrating remote sensing with deep learning proved efficient for deriving macrozoobenthos and sediment properties. In this study, we combined deep-learning features derived from Sentinel-2 images and Object-Based Image Analysis (OBIA) to explicitly include spatial aspects in the prediction of tsediment and macrozoobenthos properties of tidal flats , as well as the distribution of four benthic species. The deep-learning features extracted from a convolutional autoencoder model were analysed with OBIA to include spatial, textural, and contextual information. Object sets of varying sizes and shapes based on the spectral bands and/or the deep-learning features, served as the spatial units. These object sets and the field-collected points were used to train the Random Forest prediction model. Predictions were made for the tidal basins Pinkegat and Zoutkamperlaag in the Dutch Wadden Sea for 2018 to 2020. The overall prediction scores of the environmental variables ranged between 0.31 and 0.54. The species-distribution prediction model achieved accuracies ranging from 0.54 to 0.68 for the four benthic species). There was an average improvement of 21% points on predictions using objects with deep learning features compared to the pixel-based predictions with just the spectral bands. The mean spatial unit that captured the patterns best ranged between 0.3 ha and 13 ha for the different variables. Overall, using both OBIA and deep-learning features consistently improved the predictions, making it a valuable combination for monitoring these important environmental variables of coastal regions. [ABSTRACT FROM AUTHOR]

Urban waterfront areas, which are essential natural resources and highly perceived public areas in cities, play a crucial role in enhancing urban environment. This study integrates deep learning with human perception data sourced from street view images to study the relationship between visual landscape features and human perception of urban waterfront areas, employing linear regression and random forest models to predict human perception along urban coastal roads. Based on aesthetic and distinctiveness perception, urban coastal roads in Xiamen were classified into four types with different emphasis and priorities for improvement. The results showed that: 1) the degree of coastal openness had the greatest influence on human perception while the coastal landscape with a high green visual index decreases the distinctiveness perception; 2) the random forest model can effectively predict human perception on urban coastal roads with an accuracy rate of 87% and 77%; 3) The proportion of low perception road sections with potential for improvement is 60.6%, among which the proportion of low aesthetic perception and low distinctiveness perception road sections is 10.5%. These findings offer crucial evidence regarding human perception of urban coastal roads, and can provide targeted recommendations for enhancing the visual environment of urban coastal road landscapes. [ABSTRACT FROM AUTHOR]

This paper presents a method for estimating arousal and emotional valence levels using non-contact environmental sensing, addressing challenges such as discomfort from long-term device wear and privacy concerns associated with facial image analysis. We employed environmental data to develop machine learning models, including Random Forest, Gradient Boosting Decision Trees, and the deep learning model CNN-LSTM, and evaluated their accuracy in estimating emotional states. The results indicate that decision tree-based methods, particularly Random Forest, are highly effective for estimating emotional states from environmental data. [ABSTRACT FROM AUTHOR]

In tree-based algorithms like random forest and deep forest, due to the presence of numerous inefficient trees and forests in the model, the computational load increases and the efficiency decreases. To address this issue, in the present paper, a model called Automatic Deep Forest Shrinkage (ADeFS) is proposed based on shrinkage techniques. The purpose of this model is to reduce the number of trees, enhance the efficiency of the gcforest, and reduce computational load. The proposed model comprises four steps. The first step is multi-grained scanning, which carries out a sliding window strategy to scan the input data and extract the relations between features. The second step is cascade forest, which is structured layer-by-layer with a number of forests consisting of random forest (RF) and completely random forest (CRF) within each layer. In the third step, which is the innovation of this paper, shrinkage techniques such as LASSO and elastic net (EN) are employed to decrease the number of trees in the last layer of the previous step, thereby decreasing the computational load, and improving the gcforest performance. Among several shrinkage techniques, elastic net (EN) provides better performance. Finally, in the last step, the simple average ensemble method is employed to combine the remaining trees. The proposed model is evaluated by Monte Carlo simulation and three real datasets. Findings demonstrate the superior performance of the proposed ADeFS-EN model over both gcforest and RF, as well as the combination of RF with shrinkage techniques. [ABSTRACT FROM AUTHOR]

Analyzing habitat conditions and mapping habitat structures are crucial for monitoring ecosystems and implementing effective conservation measures, especially in the context of declining open grassland ecosystems in Europe. The marsh fritillary (Euphydryas aurinia), an endangered butterfly species, depends heavily on specific habitat conditions found in these grasslands, making it vulnerable to environmental changes. To address this, we conducted a comprehensive habitat suitability analysis within the Hainich National Park in Thuringia, Germany, leveraging very high-resolution (VHR) airborne, red-green-blue (RGB), and color-infrared (CIR) remote sensing data and deep learning techniques. We generated habitat suitability models (HSM) to gain insights into the spatial factors influencing the occurrence of E. aurinia and to predict potential habitat suitability for the whole study site. Through a deep learning classification technique, we conducted biotope mapping and generated fine-scale spatial variables to model habitat suitability. By employing various modeling techniques, including Generalized Additive Models (GAM), Generalized Linear Models (GLM), and Random Forest (RF), we assessed the influence of different modeling parameters and pseudo-absence (PA) data generation on model performance. The biotope mapping achieved an overall accuracy of 81.8%, while the subsequent HSMs yielded accuracies ranging from 0.69 to 0.75, with RF showing slightly better performance. The models agree that homogeneous grasslands, paths, hedges, and areas with dense bush encroachment are unsuitable habitats, but they differ in their identification of high-suitability areas. Shrub proximity and density were identified as important factors influencing the occurrence of E. aurinia. Our findings underscore the critical role of human intervention in preserving habitat suitability, particularly in mitigating the adverse effects of natural succession dominated by shrubs and trees. Furthermore, our approach demonstrates the potential of VHR remote sensing data in mapping small-scale butterfly habitats, offering applicability to habitat mapping for various other species. [ABSTRACT FROM AUTHOR]

Performing simulated snoring (SS) is a commonly used method to evaluate the source of snoring in obstructive sleep apnea (OSA). SS sounds is considered as a potential biomarker for OSA. SS sounds can be easily recorded, which is a cost-effective method for prescreening purposes. This study aimed to validate the performance of artificial intelligence (AI) models using SS sounds for OSA diagnosis. All participants underwent full-night polysomnography (PSG) monitoring at the sleep center. SS sounds of the participants were recorded during the laryngoscopy procedure. The audio data were processed via Python, and relevant features were extracted. OSA diagnostic models were developed using three machine learning (ML) algorithms and one deep learning (DL) algorithm. The diagnostic performance was evaluated by multiple indicators. A total of 465 participants were included. For the support vector machine algorithm, the accuracy values at apnea-hypopnea index (AHI) levels of 5, 15, and 30 per hour were 0.914, 0.887, and 0.807, respectively. For the K-nearest neighbor algorithm, the accuracy values at AHI levels of 5, 15, and 30 per hour were 0.896, 0.872, and 0.756, respectively. For the random forest algorithm, the accuracy values at AHI levels of 5, 15, and 30 per hour were 0.905, 0.881, and 0.804, respectively. For the audio spectrogram transformer algorithm, the accuracy values at AHI levels of 5, 15, and 30 per hour were 0.926, 0.887, and 0.830, respectively. Our study demonstrates that DL models can effectively screen and identify OSA with commendable performance. In addition, the identification ability of the DL models was better than that of any of the ML models. • We can employ simulated snoring sound features to develop machine learning and deep learning predictive models for OSA screening. [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]

The continuous evolution of deep learning has garnered significant attention in spectroscopy. This study focuses on identifying wheat flour, presenting a more efficient and accurate method by combining two-dimensional correlation spectroscopy (2D-COS) and deep learning techniques. A data set of 316 near-infrared (NIR) spectral samples of four types of wheat flour was collected. By applying three disparate 2D-COS techniques, i.e., synchronous, asynchronous, and integrated, we crafted 948 2D-COS images. These images, obtained by transforming the original one-dimensional spectra into 2D representations, offer richer information for deep learning analysis. The study introduced an 18-layer residual network incorporating a convolutional attention mechanism, specifically tailored for the 2D-COS analysis of wheat flour, aimed at enhancing the model's discriminative capabilities by refining the residual neural network's structure. Achieving an unprecedented recognition accuracy of 100% through methodical optimization and rigorous training on the synchronous 2D-COS data set of wheat flour, the proposed model is a testament to the efficacy of deep learning in spectroscopic analysis. To further exhibit the confluence of 2D-COS with deep learning, t-distributed stochastic neighbor embedding was employed to visualize the distinctive 2D-COS features within the deep learning architecture. Additionally, the model's performance was juxtaposed with prevailing NIR spectral recognition methods, including random forest, gradient boosting decision tree, and artificial neural network. This comparison cemented the proposed approach's superiority in wheat flour categorization. The findings of this study not only introduce a novel and efficient solution for wheat flour quality analysis but also underscore the significant potential of deep learning techniques in spectroscopy applications. [ABSTRACT FROM AUTHOR]

Soil salinization affects agricultural productivity and ecosystem health in Xinjiang, especially in arid areas. The region's complex topography and limited agricultural data emphasize the pressing need for effective, large-scale monitoring technologies. Therefore, 1044 soil samples were collected from arid farmland in northern Xinjiang, and the potential effectiveness of soil salinity monitoring was explored by combining environmental variables with Landsat 8 and Sentinel-2. The study applied four types of feature selection algorithms: Random Forest (RF), Competitive Adaptive Reweighted Sampling (CARS), Uninformative Variable Elimination (UVE), and Successive Projections Algorithm (SPA). These variables are then integrated into various machine learning models—such as Ensemble Tree (ETree), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and LightBoost—as well as deep learning models, including Convolutional Neural Networks (CNN), Residual Networks (ResNet), Multilayer Perceptrons (MLP), and Kolmogorov–Arnold Networks (KAN), for modeling. The results suggest that fertilizer use plays a critical role in soil salinization processes. Notably, the interpretable model KAN achieved an accuracy of 0.75 in correctly classifying the degree of soil salinity. This study highlights the potential of integrating multi-source remote sensing data with deep learning technologies, offering a pathway to large-scale soil salinity monitoring, and thereby providing valuable support for soil management. [ABSTRACT FROM AUTHOR]

The rise in cyber security issues has caused significant harm to tech world and thus society in recent years. Intrusion detection systems (IDSs) are crucial for the detection and the mitigation of the increasing risk of cyber attacks. False and disregarded alarms are a common problem for traditional IDSs in high-bandwidth and large-scale network systems. While applying learning techniques to intrusion detection, researchers are facing challenges mainly due to the imbalanced training sets and the high dimensionality of datasets, resulting from the scarcity of attack data and longer training periods, respectively. Thus, this leads to reduced efficiency. In this research study, we propose a strategy for dealing with the problems of imbalanced datasets and high dimensionality in IDSs. In our efficient and novel framework, we integrate an oversampling strategy that uses Generative Adversarial Networks (GANs) to overcome the difficulties introduced by imbalanced datasets, and we use the Random Forest (RF) importance algorithm to select a subset of features that best represent the dataset to reduce the dimensionality of a training dataset. Then, we use three deep learning techniques, Multi-Layer Perceptron (MLP), Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM), to classify the attacks. We implement and evaluate this proposed framework on the CICIDS2017 dataset. Experimental results show that our proposed framework outperforms state-of-the-art approaches, vastly improving DL model detection accuracy by 98% using CNN. [ABSTRACT FROM AUTHOR]

This study aims to accurately estimate vertical tire forces on racing tires of specific stiffness using acceleration, pressure, and speed data measurements from a test rig. A hybrid model, termed Random Forest Assisted Deep Neural Network (RFADNN), is introduced, combining a novel deep learning framework with the Random Forest Algorithm to enhance estimation accuracy. By leveraging the Temporal Convolutional Network (TCN), Minimal Gated Unit (MGU), Long Short-Term Memory (LSTM), and Attention mechanisms, the deep learning framework excels in extracting complex features, which the Random Forest Model subsequently analyzes to improve the accuracy of estimating vertical tire forces. Validated with test data, this approach outperforms standard models, achieving an MAE of 0.773 kgf, demonstrating the advantage of the RFADNN method in required vertical force estimation tasks for race tires. This comparison emphasizes the significant benefits of incorporating advanced deep learning with traditional machine learning to provide a comprehensive and interpretable solution for complex estimation challenges in automotive engineering. [ABSTRACT FROM AUTHOR]

Lateral flow assay has been extensively used for at-home testing and point-of-care diagnostics in rural areas. Despite its advantages as convenient and low-cost testing, it suffers from poor quantification capacity where only yes/no or positive/negative diagnostics are achieved. In this study, machine learning and deep learning models were developed to quantify the analyte load from smartphone-captured images of the lateral flow assay test. The comparative analysis identified that random forest and convolutional neural network (CNN) models performed well in classifying the lateral flow assay results compared to other well-established machine learning models. When trained on small-size images, random forest models excelled CNN models in image classification. Contrarily, CNN models outperformed random forest models in classifying noisy images. [ABSTRACT FROM AUTHOR]

Detecting obfuscated malware concentrates on specifying any malicious software (malware) that has been intentionally disguised or concealed to avoid conventional detection systems. Typically, obfuscation methods like runtime obfuscation, packing, and code encryption have been utilized by malware authors to change the appearance of malware while preserving its malicious functionality. As these methods have evolved, conventional detection systems have become less effective, requiring further sophisticated solutions like machine and deep learning approaches. Machine and deep learning approaches work on obfuscated malware detection via auto-learning features and patterns from massive datasets. Additionally, these approaches can specify formerly unseen malware by analyzing various features. In this paper, the Independent Component Analysis (ICA) is first utilized for isolating relevant features from obfuscated data to be prepared for binary classification. For further data analysis, essential feature selection, and a deeper comprehension of the relationships inside the dataset, the Pearson correlation coefficient is applied to the dataset to be prepared for multiclass classification. This dual scheme improves the feature extraction process depending on the classification type, enhancing the system's versatility and performance. Then, a proposed One-Dimensional Convolutional Neural Network (1D-CNN) is leveraged for extracting efficient features from memory traces, and providing an accurate system for classifying obfuscated memory malware. The combination of ICA or Pearson correlation with 1D-CNN in a unified system offers a scalable and inclusive settlement for binary or multiclass classification and contributes to the progress of malware classification systems. Besides the proposed 1DCNN approach, two machine learning approaches are trained and assessed on the CIC-MalMem-2022 dataset, and the attained results depicted that the performance of the proposed 1D-CNN approach was superior, with accuracies of 99% and 88% for binary and multiclass classification, respectively. [ABSTRACT FROM AUTHOR]

Ataxia, a neurological disorder characterized by impaired coordination and unsteady movements, presents significant challenges for accurate diagnosis and classification. traditional machine-learning (ML) and deep-learning (DL) models often struggle to achieve high accuracy in predicting and classifying this complex condition. This study addresses these limitations by introducing a novel hybrid model, XGBoost-multi-layer-perceptron (XGBMLP), specifically designed to enhance the accuracy of ataxia prediction and classification. The objective of this research is to develop a more reliable and precise diagnostic tool that outperforms existing ML and DL approaches. The methodology involved integrating the strengths of XGBoost, known for its powerful gradient boosting, with the multi-layer perceptron (MLP) neural network, creating a robust hybrid model. The proposed XGB-MLP model was rigorously tested against conventional models like random forest (RF), logistic regression (LR), support vector machine (SVM), MLP, and standalone XGBoost. The findings reveal that the XGB-MLP model achieves outstanding accuracy rates of 98.91% for ataxia prediction and 97.91% for classification, significantly surpassing the performance of the traditional models. [ABSTRACT FROM AUTHOR]

Introduction: Pancreaticoduodenectomy (PD) for patients with pancreatic ductal adenocarcinoma (PDAC) is associated with a high risk of postoperative complications (PoCs) and risk prediction of these is therefore critical for optimal treatment planning. We hypothesize that novel deep learning network approaches through transfer learning may be superior to legacy approaches for PoC risk prediction in the PDAC surgical setting. Methods: Data from the US National Surgical Quality Improvement Program (NSQIP) 2002–2018 were used, with a total of 5,881,881 million patients, including 31,728 PD patients. Modelling approaches comprised of a model trained on a general surgery patient cohort and then tested on a PD specific cohort (general model), a transfer learning model trained on the general surgery patients with subsequent transfer and retraining on a PD-specific patient cohort (transfer learning model), a model trained and tested exclusively on the PD-specific patient cohort (direct model), and a benchmark random forest model trained on the PD patient cohort (RF model). The models were subsequently compared against the American College of Surgeons (ACS) surgical risk calculator (SRC) in terms of predicting mortality and morbidity risk. Results: Both the general model and transfer learning model outperformed the RF model in 14 and 16 out of 19 prediction tasks, respectively. Additionally, both models outperformed the direct model on 17 out of the 19 tasks. The transfer learning model also outperformed the general model on 11 out of the 19 prediction tasks. The transfer learning model outperformed the ACS-SRC regarding mortality and all the models outperformed the ACS-SRC regarding the morbidity prediction with the general model achieving the highest Receiver Operator Area Under the Curve (ROC-AUC) of 0.668 compared to the 0.524 of the ACS SRC. Conclusion: DNNs deployed using a transfer learning approach may be of value for PoC risk prediction in the PD setting. [ABSTRACT FROM AUTHOR]

This study addresses the growing importance of hand gesture recognition across diverse fields, such as industry, education, and healthcare, targeting the often-neglected needs of the deaf and dumb community. The primary objective is to improve communication between individuals, thereby enhancing the overall quality of life, particularly in the context of advanced healthcare. This paper presents a novel approach for real-time hand gesture recognition using bio-impedance techniques. The developed device, powered by a Raspberry Pi and connected to electrodes for impedance data acquisition, employs an impedance chip for data collection. To categorize hand gestures, Convolutional Neuron Network (CNN), XGBoost, and Random Forest were used. The model successfully recognized up to nine distinct gestures, achieving an average accuracy of 97.24% across ten subjects using a subject-dependent strategy, showcasing the efficacy of the bioimpedance-based system in hand gesture recognition. The promising results lay a foundation for future developments in nonverbal communication between humans and machines as it contributes to the advancement of technology for the benefit of individuals with hearing impairments, addressing a critical social need. [ABSTRACT FROM AUTHOR]

Classification and regression problems can be challenging when the relevant input features are diluted in noisy datasets, in particular when the sample size is limited. Traditional Feature Selection (FS) methods address this issue by relying on some assumptions such as the linear or additive relationship between features. Recently, a proliferation of Deep Learning (DL) models has emerged to tackle both FS and prediction at the same time, allowing non-linear modeling of the selected features. In this study, we systematically assess the performance of DL-based feature selection methods on synthetic datasets of varying complexity, and benchmark their efficacy in uncovering non-linear relationships between features. We also use the same settings to benchmark the reliability of gradient-based feature attribution techniques for Neural Networks (NNs), such as Saliency Maps (SM). A quantitative evaluation of the reliability of these approaches is currently missing. Our analysis indicates that even simple synthetic datasets can significantly challenge most of the DL-based FS and SM methods, while Random Forests, TreeShap, mRMR and LassoNet are the best performing FS methods. Our conclusion is that when quantifying the relevance of a few non linearly-entangled predictive features diluted in a large number of irrelevant noisy variables, DL-based FS and SM interpretation methods are still far from being reliable. [ABSTRACT FROM AUTHOR]

In recent years, several machine learning (ML) approaches have been proposed to predict gene expression signal and chromatin features from the DNA sequence alone. These models are often used to deduce and, to some extent, assess putative new biological insights about gene regulation, and they have led to very interesting advances in regulatory genomics. This article reviews a selection of these methods, ranging from linear models to random forests, kernel methods, and more advanced deep learning models. Specifically, we detail the different techniques and strategies that can be used to extract new gene-regulation hypotheses from these models. Furthermore, because these putative insights need to be validated with wet-lab experiments, we emphasize that it is important to have a measure of confidence associated with the extracted hypotheses. We review the procedures that have been proposed to measure this confidence for the different types of ML models, and we discuss the fact that they do not provide the same kind of information. [ABSTRACT FROM AUTHOR]

The well log data is represented as raw tabular data with diverse and nonlinear features. This poses a challenge for feature learning by machine learning models. The recent popular decision tree-based algorithms, such as random forest (RF), extreme gradient boosting (XGB) are prominent for learning nonlinear relationships of well log data in comparison with other methods of support vector machines (SVMs) and even deep learning models. In this work, we proposed using Tabnet model for direct learning tabular data of well logs. To our knowledge, this is the first time a state-of-the-art transformer-based model of Tabnet has been utilized for this task. The efficiency of Tabnet-based feature embedding is evaluated in two tasks of rock facies classification and learning feature embedding. We prove the efficiency of Tabnet model by experimental results on two small datasets of public Kansas dataset, which has nine wells for training and two wells for testing, and our own-built dataset, which has four wells for training and one well for testing. Although training on the modest amount of well log data, the proposed Tabnet model still promotes better classification efficiency than tree-based models of RF, XGBoost, LightGBM and deep learning models of MLP, CNN-1D, and ResNet-1D. KEY POINTS: Tabnet efficiency for facies classification and learning feature embedding from well log data. A challenge to learn these raw features directly for separating classes of facies. The superiority of the Tabnet network in comparison with other ruling tree-based methods and deep learning models. Facies classification and learning feature embeddings for categorical variables of well logs. [ABSTRACT FROM AUTHOR]

To improve radar-based quantitative precipitation estimation (QPE) methods, this study investigated the relationship between radar reflectivity (Z) and hourly rainfall intensity (R) using data from 289 precipitation events in Shanghai between September 2020 and March 2024. Two Z-R relationship models were compared in terms of their fitting performance: Z = 270.81 R1.09 (empirically fitted relationship) and Z = 300 R1.4 (standard relationship). The results show that the Z = 270.81 R1.09 model outperforms the Z = 300 R1.4 model in terms of fitting accuracy. Specifically, the Z = 270.81 R1.09 model more effectively captures the nonlinear relationship between radar reflectivity and rainfall intensity, with a higher degree of agreement between the fitted curve and the observed data points. This model demonstrated superior performance across all 289 precipitation events. This study evaluated the performance of four machine learning approaches while incorporating five meteorological features: specific differential phase shift (KDP), echo-top height (ET), vertical liquid water content (VIL), differential reflectivity (ZDR), and correlation coefficient (CC). Nine QPE models were constructed using these inputs. The key findings are as follows: (1) For models with a single-variable input, the KAN deep learning model outperformed Random Forest, Gradient Boosting Decision Trees, Support Vector Machines, and the traditional Z-R relationship. (2) When six features were used as inputs, the accuracy of the machine learning models improved significantly, with the KAN deep learning model outperforming other machine learning methods. Compared to using only radar reflectivity, the KAN deep learning model reduced the MRE by 20.78%, MAE by 4.07%, and RMSE by 12.74%, while increasing the coefficient of determination (R2) by 18.74%. (3) The integration of multiple meteorological features and machine learning optimization significantly enhanced QPE accuracy, with the KAN deep learning model performing best under varying meteorological conditions. This approach offers a promising method for improving radar-based QPE, particularly considering seasonal, weather system, and precipitation stage differentiation. [ABSTRACT FROM AUTHOR]

Over the last decade, there have been a lot of advances in the area of human posture recognition. Among multiple approaches proposed to solve this problem, those based on deep learning have shown promising results. Taking another step in this direction, this paper analyzes the performance of deep learning-based hybrid architecture for fall detection, In this regard, the fusion of the residual network (ResNet-50) deep features with support vector machine (SVM) at the classification layer has been considered. The proposed approach outperforms the existing methods yielding an accuracy of 98.82%, 97.95%, and 99.98% on three datasets i.e. Multi-Camera Fall (MCF) using four postures, UR Fall detection (URFD) using four postures, and UP-Fall detection (UPFD) using four postures respectively. It is important to mention that the existing methods achieve accuracies of 97.9%, 97.33%, and 95.64% on the MCF, URDF and UPFD datasets, respectively. Moreover, we achieved 100% accuracy on the UPFD two-posture task. The URFD and MCF datasets have been utilized to assess the fall detection performance of our method under a realistic environment (e.g. camouflage, occlusion, and variation in lighting conditions due to day/night lighting variation). For comparison purposes, we have also performed experiments using six state-of-the-art deep learning networks, namely; ResNet-50, ResNet-101, VGG-19, InceptionV3, MobileNet, and Xception. The results demonstrate that the proposed approach outperforms other network models both in terms of accuracy and time efficiency. We also compared the performance of SVM with Naive Bayes, Decision Tree, Random Forest, KNN, AdaBoost, and MLP used at the classifier layer and found that SVM outperforms or is on par with other classifiers. [ABSTRACT FROM AUTHOR]

The Internet of Things represents a rapidly evolving networking paradigm that brings numerous benefits through its diverse applications. Advances in embedded system architectures and the compression of IPv6 have facilitated embedding IP stack functionalities within resource-constrained low power and lossy networks (LLNs). However, the widespread deployment of LLN networks within the IoT infrastructure has made them a target for cyber attackers. Machine and Deep Learning methods show promise in detecting anomalies within the Routing Protocol for LLNs. However, they face challenges such as handling imbalanced data, identifying relevant attack features, selecting the most suitable model parameters for detection, and accurately pinpointing attacker nodes across various locations in LLN. This study proposes an approach for detecting Selective Forwarding attacks in the LLN, consisting of four main stages: data preparation, data balancing using an optimized synthetic minority over-sampling technique, feature selection with binary particle swarm optimization algorithm, and attack detection via an optimized Random Forest with GridSearchCV. The proposed approach is evaluated using a grid-based dataset across three scenarios that differ in attacker node positioning relative to the LLN's root node. The findings indicate that the proposed approach achieved the highest accuracy and F-measure score of 99.82% in the second scenario. It also recorded precision scores of 100% in both the first and second scenarios and a recall score of 100% in the third scenario. Ultimately, the proposed approach outperforms existing methods, demonstrating its superiority in detecting SF attacks within LLNs. [ABSTRACT FROM AUTHOR]

This paper investigates the effectiveness of traditional machine learning techniques, namely linear regression and random forest (RF), in enhancing the detection of money laundering (ML) activities within financial systems. As ML schemes evolve in complexity, traditional rule-based methods struggle with high false favorable rates and a lack of adaptability, prompting the need for more sophisticated analytical approaches. In contrast to the complexities of deep learning models, this study explores the potential of these more accessible machine learning methods in identifying and analyzing suspicious transactional patterns. We apply linear regression and RF models to transactional data to detect anomalous activities that could indicate ML. Our research thoroughly compares these models based on key performance metrics such as accuracy, precision, and recall. The findings suggest that while less complex than deep learning frameworks, linear regression, and RF models offer substantial benefits. They provide a more streamlined, interpretable, and efficient alternative to conventional rule-based systems in the context of ML detection. This study contributes to the ongoing discourse on the application of machine learning in financial crime detection, demonstrating the practicality and effectiveness of these methods in a critical area of financial security. [ABSTRACT FROM AUTHOR]

Simple Summary: Papillary thyroid cancer (PTC) is the most common subtype of thyroid cancer (TC), accounting for approximately 80–90% of all TC cases. PTC is generally considered a low-grade malignancy. However, aggressive behaviors, such as cervical lymph node metastasis, can be observed in some cases, with central lymph node metastasis (CLNM) regarded as the primary site of lymph node metastasis. In current clinical practice in China, prophylactic central lymph node dissection is commonly performed regardless of whether CLNM is present. This approach is associated with an increased risk of complications and unnecessary lymph node removal, raising concerns about overtreatment. Therefore, there is an urgent need for an efficient method to predict CLNM in PTC patients, which could guide clinical diagnosis and treatment strategies. In this study, we developed a fusion model based on the attention mechanism-based multimodal classification network (AMMCNet) architecture, integrating MRI images and clinicopathological data to efficiently predict CLNM in PTC patients. Background: Central lymph node metastasis (CLNM) in papillary thyroid cancer (PTC) significantly influences surgical decision-making strategies. Objectives: This study aims to develop a predictive model for CLNM in PTC patients using magnetic resonance imaging (MRI) and clinicopathological data. Methods: By incorporating deep learning (DL) algorithms, the model seeks to address the challenges in diagnosing CLNM and reduce overtreatment. The results were compared with traditional machine learning (ML) models. In this retrospective study, preoperative MRI data from 105 PTC patients were divided into training and testing sets. A radiologist manually outlined the region of interest (ROI) on MRI images. Three classic ML algorithms (support vector machine [SVM], logistic regression [LR], and random forest [RF]) were employed across different data modalities. Additionally, an AMMCNet utilizing convolutional neural networks (CNNs) was proposed to develop DL models for CLNM. Predictive performance was evaluated using receiver operator characteristic (ROC) curve analysis, and clinical utility was assessed through decision curve analysis (DCA). Results: Lesion diameter was identified as an independent risk factor for CLNM. Among ML models, the RF-(T1WI + T2WI, T1WI + T2WI + Clinical) models achieved the highest area under the curve (AUC) at 0.863. The DL fusion model surpassed all ML fusion models with an AUC of 0.891. Conclusions: A fusion model based on the AMMCNet architecture using MRI images and clinicopathological data was developed, effectively predicting CLNM in PTC patients. [ABSTRACT FROM AUTHOR]

There has been a rapid evolution of tree-based ensemble algorithms which have outperformed deep learning in several studies, thus emerging as a competitive solution for many applications. In this study, ten tree-based ensemble algorithms (random forest, bagging meta-estimator, adaptive boosting (AdaBoost), gradient boosting machine (GBM), extreme gradient boosting (XGBoost), light gradient boosting (LightGBM), histogram-based GBM, categorical boosting (CatBoost), natural gradient boosting (NGBoost), and the regularised greedy forest (RGF)) were comparatively evaluated for the enhancement of Copernicus digital elevation model (DEM) in an agricultural landscape. The enhancement methodology combines elevation and terrain parameters alignment, with feature-level fusion into a DEM enhancement workflow. The training dataset is comprised of eight DEM-derived predictor variables, and the target variable (elevation error). In terms of root mean square error (RMSE) reduction, the best enhancements were achieved by GBM, random forest and the regularised greedy forest at the first, second and third implementation sites respectively. The computational time for training LightGBM was nearly five-hundred times faster than NGBoost, and the speed of LightGBM was closely matched by the histogram-based GBM. Our results provide a knowledge base for other researchers to focus their optimisation strategies on the most promising algorithms. [ABSTRACT FROM AUTHOR]

Introduction: Polycystic Ovarian Syndrome (PCOS) is a prevalent endocrine disorder affecting women of reproductive age, characterised by irregular menstrual cycles, hyperandrogenism and polycystic ovaries. Despite its high prevalence, the diagnosis of PCOS remains challenging due to the variability in symptom presentation. Traditional diagnostic methods involve clinical evaluation, biochemical assays and ultrasound imaging. Machine Learning (ML) and Deep Learning (DL) models offer promising avenues for predicting probable cases of PCOS using non invasive qualitative features. Aim: To develop and compare the performance of Random Forest (RF) and Feedforward Neural Network (FFNN) models in predicting PCOS using abundant non invasive qualitative features. Materials and Methods: A retrospective case-control study was conducted with 100 cases and 100 controls, selected based on ultrasound-confirmed PCOS diagnosis in the Obstetrics and Gynaecology, Gayatri Vidya Parishad Institute of Healthcare and Medical Technology (GVP IHC MT), Medical College departments from February 2024 to October 2024. Data were collected using a structured questionnaire capturing demographic and clinical variables. Feature selection was performed using the Chi-square filter method, with 10 features identified as significant. The data were split into training (80%) and testing (20%) sets and stratified 5-fold cross-validation was applied. Model performance was evaluated using accuracy, precision, recall, F1 score and Area Under Curve (AUC). Results: The RF model demonstrated high performance on the training set, with an average accuracy of 0.95, but exhibited variability on the testing set (accuracy of 0.80). The FFNN model showed consistent performance across both training (accuracy of 0.80) and testing datasets (accuracy of 0.82). The RF model identified irregular cycles and hirsutism as key predictors, while the FFNN model highlighted weight gain and abnormal Body Mass Index (BMI) as important features. The RF model required significantly less computational time compared to the FFNN model. Conclusion: The RF model is preferable for tasks requiring computational efficiency, while the FFNN model offers better generalisation. The complementary feature importance rankings suggest that integrating insights from both models could enhance the understanding of PCOS predictors. In epidemiological investigations, these models can be used as preliminary screening tools for identifying probable cases of PCOS using non invasive qualitative features, especially in areas where diagnostic facilities are not available. [ABSTRACT FROM AUTHOR]

Predicting wind speed is a complex undertaking influenced not only by the wind-speed sequence itself but also by various meteorological factors. This paper introduces a novel multivariate deep-learning neural network prediction model that takes into account not only historical wind-speed data but also a series of meteorological features relevant to wind speed. The meteorological features associated with wind speed are initially extracted using the random forest algorithm (RF). Subsequently, Variational Mode Decomposition and Autocorrelation Function analysis are employed for noise reduction in the wind-speed series. Finally, the wind-speed series are predicted using a Gated Recurrent Unit (GRU) deep-learning neural network, and an Improved Sparrow Search Algorithm is proposed to optimize the four parameters of the GRU. To validate the predictive performance of the model, experimental data from three cities in China, Shenyang, Dalian, and Yingkou, are utilized. The experimental results demonstrate that our proposed model outperforms other models, as evidenced by four key performance indicators. [ABSTRACT FROM AUTHOR]

The pursuit of information is an integral aspect of a learner's daily routine, often facilitated by information retrieval systems. In this study, we explore the correlation between text complexity and reading comprehension through the lens of neuroinformatics. Our objective is to ascertain whether it is feasible to infer the difficulty level of a text for a learner based on their brain activity, quantified by an electroencephalogram (EEG), during the information retrieval process. To accomplish this, we administered a section of the Woodcock Reading Mastery Test to our participants, evaluating their average reading fluency. Subsequently, the 18 participants under observation read paragraphs of varying complexity and responded to questions pertaining to the text. Utilizing the amassed data, we trained a deep learning model known as EEGNet to autonomously discern the complexity of the text being read, relying on the EEG signals. Furthermore, we conducted a comparison with previous research that employed a distinct characterization approach and employed a random forest algorithm for classification. The outcomes revealed that the present approach (EEGNet) outperformed the former method, achieving an accuracy rate of 80.83% across all subjects, whereas the random forest algorithm achieved a mere 50.28% accuracy. Our findings suggest the potential to classify the EEG signals in accordance with the level of difficulty a student encounters when comprehending the text. We propose several avenues for extending our work into real-time scenarios. [ABSTRACT FROM AUTHOR]

Above-ground biomass (AGB) is one of the most popular forest attribute used to estimating, monitoring and evaluating global carbon storage. Accurately estimating AGB is one of the most significant steps in decision-making regarding sustainable forest management, climate policy and management efficiency. Thus, developing accurate AGB estimation models using satellite data is essential. In the present study, the capability of Phased array type L-band synthetic aperture radar (ALOS-PALSAR) and SPOT-6 data to model AGB using Deep learning (DL) and Random forest (RF) and Multiple linear regression (MLR) algorithms were evaluated in coniferous planted area, northern Iran. The systematic cluster sampling method was applied to collect field plot data. A total of 180 circular plots were measured to calculate AGB per hectare. The DL, RF and MLR algorithms were used for AGB modeling. The relative root mean squared error (rRMSE) and R2 using ALOS-PALSAR data were 21.99% and 0.21 for the DL, 48.46% and 0.18 for RF and 50.20% and 0.11 for MLR, respectively. Also, the RMSE% and R2 using SPOT-6 data were 18.31% and 0.44 for DL, 39.64% and 0.43 for the RF and 44.08% and 0.38 for MLR, respectively. Compared to modeling AGB using ALOS-PALSAR and SPOT-6 data separately, the combination of ALOS-PALSAR and SPOT-6 improved AGB prediction (1.14–23% decrease in RMSE% and 0.11–0.33 increase in R2).The results showed that using of DL provided an increase in prediction accuracy compared to RF and MLR. Based on the results, we conclude that modeling AGB using a combination of ALOS-PALSAR and SPOT-6 data and DL can be useful for estimating AGB in the coniferous planted forests. [ABSTRACT FROM AUTHOR]

Motivation Post-translational modification (PTM) crosstalk events play critical roles in biological processes. Several machine learning methods have been developed to identify PTM crosstalk within proteins, but the accuracy is still far from satisfactory. Recent breakthroughs in deep learning and protein structure prediction could provide a potential solution to this issue. Results We proposed DeepPCT, a deep learning algorithm to identify PTM crosstalk using AlphaFold2-based structures. In this algorithm, one deep learning classifier was constructed for sequence-based prediction by combining the residue and residue pair embeddings with cross-attention techniques, while the other classifier was established for structure-based prediction by integrating the structural embedding and a graph neural network. Meanwhile, a machine learning classifier was developed using novel structural descriptors and a random forest model to complement the structural deep learning classifier. By integrating the three classifiers, DeepPCT outperformed existing algorithms in different evaluation scenarios and showed better generalizability on new data owing to its less distance dependency. Availability and implementation Datasets, codes, and models of DeepPCT are freely accessible at https://github.com/hzau-liulab/DeepPCT/. [ABSTRACT FROM AUTHOR]

Accurate prediction of the remaining useful life (RUL) of high-speed railroad contact networks can guide the scientific development of maintenance plans, thus reducing economic waste. Based on this, this paper proposes an Attention-LSTM-CNN-based Stacking deep integration method for RUL prediction of contact networks in high-speed railroads. The method first cleans the collected contact network data and analyzes the feature contribution, then selects the stacked base model based on Spearman correlation. Finally, the long short-term convolutional memory neural network (Attention-LSTM-CNN), random forest (RF), and K-nearest neighbor (KNN) are used as the base learners in the Stacking integrated model to predict the RUL of the contact network with the improved attention mechanism; the hyperparameters of each base learner are optimized using genetic algorithm. Using detection data from a line in southeast China as an arithmetic example for comparison experiments, the results show that the RMSE of the prediction performance evaluation index of Attention-LSTM-CNN based on Stacking is 0.068, and the RMSE of the proposed model decreases by 26.9% and 32% compared with the individual models RF and KNN, respectively. Compared with the hybrid model consisting of RF, KNN, the RMSE of the proposed model decreased by 31.3%: the prediction after genetic optimization has an RMSE of 0.045 and MAPE of 4.32%, which is 33.8% reduction in RMSE and 13.8% reduction in MAPE compared to those before optimization. [ABSTRACT FROM AUTHOR]

In the current landscape dominated by Artificial Intelligence, the integration of Machine Learning and Deep Learning within the realm of Cultural Heritage, particularly within architectural contexts, is paramount for the efficient processing and interpretation of point clouds. These advanced methods facilitate automated segmentation and classification, significantly improving both the clarity and practical use of data acquired from laser scanning and photogrammetry. The present study investigates the Palacio de Sástago—a prominent Renaissance palace in Zaragoza, Spain—and introduces a cutting-edge modus operandi for the automated recognition of architectural elements within the palace's inner courtyard. Employing the well-established Random Forest algorithm, implemented in a Python environment, the framework begins with a comprehensive evaluation of the geometric features identified in the LiDAR point cloud. This process employs the Mean Decrease in Impurity metric to evaluate the relevance of each variable. To boost the accuracy and efficiency of the final classifications, the features are refined post-assessment, enhancing both the training phase and the algorithm's later evaluation. The research's findings demonstrate significant potential, supporting advancements in CAD systems and HBIM that will enable more precise, automated modelling of architectural elements, thereby enhancing the accuracy of digital reconstructions and improving conservation planning for heritage sites. [ABSTRACT FROM AUTHOR]

Efficient management of the write buffer in solid-state drives (SSDs) can be achieved by predicting future I/O request patterns using machine learning techniques. However, the computational demands posed by sophisticated approaches like deep learning remain significant, despite the increasing computational power of SSDs. This paper presents a novel approach to write buffer management that addresses these challenges. Our method employs a lightweight yet accurate random forest classifier to predict the forward reuse distances (FRDs) of I/O requests, indicating the likelihood of recurring identical I/O requests. Our key insight is that, rather than aiming for exact FRD predictions for future individual requests, we focus on identifying whether the predicted FRD exceeds the buffer size. With this insight, our method implements efficient buffer management operations, including bypassing the buffer storage when necessary. To achieve this, we introduce a banding method that quantizes FRDs according to the buffer size. This enables predictions at the band level, forming the foundation for a lightweight machine learning model. Subsequently, we assign high caching priority to write requests that are anticipated to have a short FRD band. Through extensive evaluations utilizing a simulator, we demonstrate that our method achieves results comparable to those of the optimal algorithm in terms of hit rate in most scenarios. Moreover, our approach outperforms state-of-the-art algorithms, which depend on past I/O reference patterns, by up to 27%. [ABSTRACT FROM AUTHOR]

The incidence of heart-related illnesses is on the rise worldwide. Heart diseases are primarily caused by a multitude of parameters, including high blood pressure, diabetes, and excessive cholesterol, which are controlled by poor dietary and lifestyle choices. The growth in cardiovascular diseases (CVD) is mostly due to several other behaviors, such as smoking, drinking, and sleeplessness. In the research, machine learning-based prediction methods work on the audio recordings of heartbeats known as phonocardiograms (PCG) to develop an algorithm that differentiates a normal healthy heart from an abnormal heart based on the heart sounds. The data set consists of 831 normal and 260 abnormal data, and the duration of each sample is 5 seconds. Features extracted from the data are up-sampled and applied to the logistic regression and random forest classification models. The developed models record a classification accuracy of 71% for logistic regression and 94% for the random forest model. Further, artificial neural networks (ANN) and Deep learning networks have been trained to improve performance and demonstrated an accuracy of 94.5%. [ABSTRACT FROM AUTHOR]

With the rapid expansion of global e-commerce, effectively managing supply chains and optimizing transportation costs has become a key challenge for businesses. This research proposed a new framework named Intelligent Supply Chain Cost Optimization (ISCCO). ISCCO integrates deep learning with advanced optimization algorithms. It focuses on minimizing transportation costs by accurately predicting customer behavior and dynamically allocating goods. ISCCO significantly enhanced supply chain efficiency by implementing an innovative customer segmentation system. This system combines autoencoders with random forests to categorize customers based on their sensitivity to discounts and likelihood of cancellations. Additionally, ISCCO optimized goods allocation using a genetic algorithm enhanced integer linear programming model. By integrating real-time demand data, ISCCO dynamically adjusts the allocation of resources to minimize transportation inefficiencies. Experimental results show that this framework increased the accuracy of user classification from 50% to 95.73%, and reduced the model loss value from 0.75 to 0.2. Furthermore, the framework significantly reduced order cancellation rates in practical applications by adjusting pre-shipment policies, thereby optimizing profits and customer satisfaction. Specifically, when the pre-shipment ratio was 25%, the optimized profit was approximately 7.5% higher than the actual profit, and the order cancellation rate was reduced from a baseline of 50.79% to 41.39%. These data confirm that the ISCCO framework enhances logistics distribution efficiency. It also improves transparency and responsiveness across the supply chain through precise data-driven decisions. This achieves maximum cost-effectiveness. [ABSTRACT FROM AUTHOR]

The growing threat of URL phishing attacks raises the need for advanced detection systems to protect digital environments. This paper explores the effectiveness of various machine learning models in classifying URLs as phishing or benign, focusing on the random forest model. Using ensemble learning, the random forest demonstrated superior accuracy and reliability compared to traditional methods, achieving consistent performance with accuracy rates between 99.93% and 99.98%. The model's performance was evaluated daily over eight days, highlighting its robustness in handling real-world scenarios. This study utilized GridSearchCV to optimize model hyperparameters, enhancing model robustness and minimizing overfitting. Future research directions include advanced feature engineering, deep learning techniques, and multimodal data integration to further improve phishing detection systems. [ABSTRACT FROM AUTHOR]

The wireless sensor network (WSN) has received significant recognition for its positive impact on environmental monitoring, yet its reliability remains prone to faults. Common factors contributing to faults include connectivity loss from malfunctioning node interfaces, disruptions caused by obstacles, and increased packet loss due to noise or congestion. This research employs a variety of machine learning and deep learning techniques to identify and address these faults, aiming to enhance the overall lifespan and scalability of the WSN. Classification models such as support vector machine (SVM), gradient boosting clasifer (GBC), K-nearest neighbours (KNN), random forest, and decision tree were employed in model training, with the decision tree emerging as the most accurate at 90.23%. Additionally, a deep learning approach, the recurrent neural network (RNN), effectively identified faults in sensor nodes, achieving an accuracy of 93.19%. [ABSTRACT FROM AUTHOR]

BACKGROUND: Rice leafroller is a serious threat to the production of rice. Monitoring the damage caused by rice leafroller is essential for effective pest management. Owing to limitations in collecting decent quality images and high‐performing identification methods to recognize the damage, studies recommending fast and accurate identification of rice leafroller damage are rare. In this study, we employed an ultra‐lightweight unmanned aerial vehicle (UAV) to eliminate the influence of the downwash flow field and obtain very high‐resolution images of the damaged areas of the rice leafroller. We used deep learning technology and the segmentation model, Attention U‐Net, to recognize the damaged area by the rice leafroller. Further, a method is presented to count the damaged patches from the segmented area. RESULTS: The result shows that Attention U‐Net achieves high performance, with an F1 score of 0.908. Further analysis indicates that the deep learning model performs better than the traditional image classification method, Random Forest (RF). The traditional method of RF causes a lot of false alarms around the edge of leaves, and is sensitive to the changes in brightness. Validation based on the ground survey indicates that the UAV and deep learning‐based method achieve a reasonable accuracy in identifying damage patches, with a coefficient of determination of 0.879. The spatial distribution of the damage is uneven, and the UAV‐based image collecting method provides a dense and accurate method to recognize the damaged area. CONCLUSION: Overall, this study presents a vision to monitor the damage caused by the rice leafroller with ultra‐light UAV efficiently. It would also contribute to effectively controlling and managing the hazardous rice leafroller. © 2024 Society of Chemical Industry. [ABSTRACT FROM AUTHOR]

Pedotransfer function (PTFs) is widely developed using machine learning algorithms (MLA) and non-linear regression (NLR). Saturated hydraulic conductivity (Ks) is obviously used to regulate water movement in the soil. The objective of the work is to evaluate the accuracy of MLA to predict Ks. Moreover, is to determine the best class of predictors that can be used to predict Ks with little estimation error. Saturated hydraulic conductivity was measured by direct method using the constant head method. In addition, Ks was predicted using PTFs developed by machine learning algorithms including Random Forest Algorithm (RFA), and Gradient Boosting Algorithm (GBA) models with three categories of soil physical predictors. Furthermore, PTFs developed by RFA, and GBA were compared with PTFs developed by NLR models using the same three categories of soil physical predictors. The three categories of soil physical predictors were the following: category-1 refers to sand, silt, and clay "SSC"; category-2 refers to category-1, in addition to bulk density "SSC+BD"; and category-3 refers to category-2, in addition to organic matter "SSC+BD+OM". The results of error analysis observed that PTFs developed by RFA were more accurate than GBA and NLR models, respectively, for Ks prediction with the three categories of soil physical predictors. Moreover, the category-3 which takes into account sand, silt, clay, bulk density, and organic matter was the best category used to predict Ks using RFA, and NLR models. PTFs developed by machine learning algorithms including RFA, and GBA models can be utilized to predict Ks with little calculation error. However, RFA was greater than GBA for calculating Ks. The efficiency of RFA was associated with the number of soil variables predictors and building classification and regression trees which is more robust than decision trees such as GBA to optimize Ks prediction. [ABSTRACT FROM AUTHOR]

This paper aims to demonstrate how machine deep learning techniques lead to relatively accurate forecasts of quarterly corporate income tax payments. Using quarterly data from Compustat for all U.S. publicly traded corporations from 2000 to 2024, I show that neural nets, the tree method, and random forest models provide robust forecasts despite their encompassing COVID-19 pandemic time periods. The results should be of interest to corporate tax planners, stock analysts, and governments. [ABSTRACT FROM AUTHOR]

A fog-cloud internet of things (IoT) system integrates fog computing with cloud infrastructure to efficiently manage processing data closer to the source, reducing latency and bandwidth usage. Efficient task scheduling in fog-cloud system is crucial for optimizing resource utilization and minimizing energy consumption. Even though many authors proposed energy efficient algorithms, failed to provide efficient method to decide the task placement between fog nodes and cloud nodes. The proposed hybrid approach is used to distinguish the task placement between fog and cloud nodes. The hybrid approach comprises the parametric task categorization algorithm (PTCA) for task categorization and the multi metric forecasting model (MMFM) based on deep deterministic policy gradient (DDPG) recurrent neural networks for scheduling decisions. PTCA classifies tasks based on priority, quality of service (QoS) demands, and computational needs, facilitating informed decisions on task execution locations. MMFM enhances scheduling by optimizing energy efficiency and task completion time. The experimental evaluation outperforms the existing models, including random forest (RF), support vector machine (SVM), and k-nearest neighbors (KNN). The proposed result shows an accuracy rate of 89%, and energy is consumed 50% lesser than the existing models. The proposed research advances energy-efficient task scheduling, enabling intelligent resource management in fog-cloud IoT environments. [ABSTRACT FROM AUTHOR]