Search

Background: The early prediction of cerebral edema changes in patients with spontaneous intracerebral hemorrhage (SICH) may facilitate earlier interventions and result in improved outcomes. This study aimed to develop and validate machine learning models to predict cerebral edema changes within 72 h, using readily available clinical parameters, and to identify relevant influencing factors. Methods: An observational study was conducted between April 2021 and October 2023 at the Quzhou Affiliated Hospital of Wenzhou Medical University. After preprocessing the data, the study population was randomly divided into training and internal validation cohorts in a 7:3 ratio (training: N = 150; validation: N = 65). The most relevant variables were selected using Support Vector Machine Recursive Feature Elimination (SVM-RFE) and Least Absolute Shrinkage and Selection Operator (LASSO) algorithms. The predictive performance of random forest (RF), GDBT, linear regression (LR), and XGBoost models was evaluated using the area under the receiver operating characteristic curve (AUROC), precision--recall curve (AUPRC), accuracy, F1-score, precision, recall, sensitivity, and specificity. Feature importance was calculated, and the SHapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) methods were employed to explain the top-performing model. Results: A total of 84 (39.1%) patients developed cerebral edema changes. In the validation cohort, GDBT outperformed LR and RF, achieving an AUC of 0.654 (95% CI: 0.611-0.699) compared to LR of 0.578 (95% CI, 0.535-0.623, DeLong: p = 0.197) and RF of 0.624 (95% CI, 0.588-0.687, DeLong: p = 0.236). XGBoost also demonstrated similar performance with an AUC of 0.660 (95% CI, 0.611-0.711, DeLong: p = 0.963). However, in the training set, GDBT still outperformed XGBoost, with an AUC of 0.603 ± 0.100 compared to XGBoost of 0.575 ± 0.096. SHAP analysis revealed that serum sodium, HDL, subarachnoid hemorrhage volume, sex, and left basal ganglia hemorrhage volume were the top five most important features for predicting cerebral edema changes in the GDBT model. Conclusion: The GDBT model demonstrated the best performance in predicting 72-h changes in cerebral edema. It has the potential to assist clinicians in identifying high-risk patients and guiding clinical decision-making. [ABSTRACT FROM AUTHOR]

Objective: Coronavirus disease 2019 (COVID-19) can cause intubation and ventilatory support due to respiratory failure, and extubation failure increases mortality risk. This study, therefore, aimed to explore the feasibility of using specific biochemical and ventilator parameters to predict survival status among COVID-19 patients by using machine learning. Methods: This study included COVID-19 patients from Taipei Medical University-affiliated hospitals from May 2021 to May 2022. Sequential data on specific biochemical and ventilator parameters from days 0–2, 3–5, and 6–7 were analyzed to explore differences between the surviving (successfully weaned off the ventilator) and non-surviving groups. These data were further used to establish separate survival prediction models using random forest (RF). Results: The surviving group exhibited significantly lower mean C-reactive protein (CRP) levels and mean potential of hydrogen ions levels (pH) levels on days 0–2 compared to the non-surviving group (CRP: non-surviving group: 13.16 ± 5.15 ng/mL, surviving group: 10.23 ± 5.15 ng/mL; pH: non-surviving group: 7.32 ± 0.07, survival group: 7.37 ± 0.07). Regarding the survival prediction performanace, the RF model trained solely with data from days 0–2 outperformed models trained with data from days 3–5 and 6–7. Subsequently, CRP, the partial pressure of carbon dioxide in arterial blood (PaCO2), pH, and the arterial oxygen partial pressure to fractional inspired oxygen (P/F) ratio served as primary indicators in survival prediction in the day 0–2 model. Conclusions: The present developed models confirmed that early biochemical and ventilatory parameters—specifically, CRP levels, pH, PaCO2, and P/F ratio—were key predictors of survival for COVID-19 patients. Assessed during the initial two days, these indicators effectively predicted the likelihood of successful weaning of from ventilators, emphasizing their importance in early management and improved outcomes in COVID-19-related respiratory failure. [ABSTRACT FROM AUTHOR]

Researchers using machine learning methods for classification can face challenges due to class imbalance, where a certain class is underrepresented. Over or under-sampling of minority or majority class observations, or solely relying on model selection for ensemble methods, may prove ineffective when the class imbalance ratio is extremely high. To address this issue, this paper proposes a method called enhance tree ensemble (ETE), based on generating synthetic data for minority class observations in conjunction with tree selection based on their performance on the training data. The proposed method first generates minority class instances to balance the training data and then uses the idea of tree selection by leveraging out-of-bag ( E T E O O B ) and sub-samples ( E T E S S ) observations, respectively. The efficacy of the proposed method is assessed using twenty benchmark problems for binary classification with moderate to extreme class imbalance, comparing it against other well-known methods such as optimal tree ensemble (OTE), SMOTE random forest ( R F S M O T E ), oversampling random forest ( R F O S ), under-sampling random forest ( R F U S ), k-nearest neighbor (k-NN), support vector machine (SVM), tree, and artificial neural network (ANN). Performance metrics such as classification error rate and precision are used for evaluation purposes. The analyses of the study revealed that the proposed method, based on data balancing and model selection, yielded better results than the other methods. [ABSTRACT FROM AUTHOR]

Forest carbon stock is an important indicator reflecting a forest ecosystem's structures and functions. Its spatial distribution is valuable for managing natural resources, protecting ecosystems and biodiversity, and further promoting sustainability, but accurately mapping the forest carbon stock distribution in a large area is a challenging task. This study selected Changting County, Fujian Province, as a case study to explore a method to map the forest carbon stock distribution using the integration of airborne Lidar, Sentinel-2, and ancillary data in 2022. The Bayesian hierarchical modeling approach was used to estimate the local forest carbon stock based on airborne Lidar data and field measurements, and then the random forest approach was used to develop a regional forest carbon stock estimation model based on the Sentinel-2 and ancillary data. The results indicated that the Lidar-based carbon stock distribution effectively provided sample plots with good spatial representativeness for modeling regional carbon stock with a coefficient of determination (R2) of 0.7 and root mean square error (RMSE) of 12.94 t/ha. The average carbon stocks were 48.55 t/ha, 55.51 t/ha, and 57.04 t/ha for Masson pine, Chinese fir, and broadleaf forests, respectively. The carbon stock in non-conservation regions was 15.2–16.1 t/ha higher than that in conservation regions. This study provides a promising method through the use of airborne Lidar data as a linkage between sample plots and Sentinel-2 data to map the regional carbon stock distribution in those subtropical regions where serious soil erosion has led to a relatively sparse forest canopy density. The results are valuable for local government to make scientific decisions for promoting ecosystem restoration due to water and soil erosion. [ABSTRACT FROM AUTHOR]

Numerous challenges are associated with the classification of satellite images of coffee plantations. The spectral similarity with other types of land use, variations in altitude, topography, production system (shaded and sun), and the change in spectral signature throughout the phenological cycle are examples that affect the process. This research investigates the influence of biennial Arabica coffee productivity on the accuracy of Landsat-8 image classification. The Google Earth Engine (GEE) platform and the Random Forest algorithm were used to process the annual and biennial mosaics of the Média Mogiana Region, São Paulo (Brazil), from 2017 to 2023. The parameters evaluated were the general hits of the seven classes of land use and coffee errors of commission and omission. It was found that the seasonality of the plant and its development phases were fundamental in the quality of coffee classification. The use of biennial mosaics, with Landsat-8 images, Brightness, Greenness, Wetness, SRTM data (elevation, aspect, slope), and LST data (Land Surface Temperature) also contributed to improving the process, generating a classification accuracy of 88.8% and reducing coffee omission errors to 22%. [ABSTRACT FROM AUTHOR]

Estimating forest growing stock volume (GSV) is crucial for forest growth and resource management, as it reflects forest productivity. National measurements are laborious and costly; however, integrating satellite data such as optical, Synthetic Aperture Radar (SAR), and airborne laser scanning (ALS) with National Forest Inventory (NFI) data and machine learning (ML) methods has transformed forest management. In this study, random forest (RF), support vector regression (SVR), and Extreme Gradient Boosting (XGBoost) were used to predict GSV using Estonian NFI data, Sentinel-2 imagery, and ALS point cloud data. Four variable combinations were tested: CO1 (vegetation indices and LiDAR), CO2 (vegetation indices and individual band reflectance), CO3 (LiDAR and individual band reflectance), and CO4 (a combination of vegetation indices, individual band reflectance, and LiDAR). Across Estonia's geographical regions, RF consistently delivered the best performance. In the northwest (NW), the RF model achieved the best performance with the CO3 combination, having an R2 of 0.63 and an RMSE of 125.39 m3/plot. In the southwest (SW), the RF model also performed exceptionally well, achieving an R2 of 0.73 and an RMSE of 128.86 m3/plot with the CO4 variable combination. In the northeast (NE), the RF model outperformed other ML models, achieving an R2 of 0.64 and an RMSE of 133.77 m3/plot under the CO4 combination. Finally, in the southeast (SE) region, the best performance was achieved with the CO4 combination, yielding an R2 of 0.70 and an RMSE of 21,120.72 m3/plot. These results underscore RF's precision in predicting GSV across diverse environments, though refining variable selection and improving tree species data could further enhance accuracy. [ABSTRACT FROM AUTHOR]

This study was designed to develop a 30 m resolution land cover dataset to improve the performance of regional weather forecasting models in East China. A 10-class land cover mapping scheme was established, reflecting East China's diverse landscape characteristics and incorporating a new category for plastic greenhouses. Plastic greenhouses are key to understanding surface heterogeneity in agricultural regions, as they can significantly impact local climate conditions, such as heat flux and evapotranspiration, yet they are often not represented in conventional land cover classifications. This is mainly due to the lack of high-resolution datasets capable of detecting these small yet impactful features. For the six-province study area, we selected and processed Landsat 8 imagery from 2015–2018, filtering for cloud cover. Complementary datasets, such as digital elevation models (DEM) and nighttime lighting data, were integrated to enrich the inputs for the Random Forest classification. A comprehensive training dataset was compiled to support Random Forest training and classification accuracy. We developed an automated workflow to manage the data processing, including satellite image selection, preprocessing, classification, and image mosaicking, thereby ensuring the system's practicality and facilitating future updates. We included three Weather Research and Forecasting (WRF) model experiments in this study to highlight the impact of our land cover maps on daytime and nighttime temperature predictions. The resulting regional land cover dataset achieved an overall accuracy of 83.2% and a Kappa coefficient of 0.81. These accuracy statistics are higher than existing national and global datasets. The model results suggest that the newly developed land cover, combined with a mosaic option in the Unified Noah scheme in WRF, provided the best overall performance for both daytime and nighttime temperature predictions. In addition to supporting the WRF model, our land cover map products, with a planned 3–5-year update schedule, could serve as a valuable data source for ecological assessments in the East China region, informing environmental policy and promoting sustainability. [ABSTRACT FROM AUTHOR]

Soil and water conservation is one of the most important parts of the national ecological civilization. The spatial control area of soil and water conservation can be divided to effectively manage the soil and water loss regions. However, it is still lacking in the spatial delineation in the regional division of soil and water conservation. Only a few studies have been focused on the spatial control of soil and water conservation, according to the small watershed. This study aims to explore the spatial control zoning for soil and water conservation, and then implement the differentiated protection and management measures. The universal soil loss equation (USLE) model was used to calculate the potential and actual soil erosion. The main influencing factors of soil erosion were determined by random forests. A self-organizing feature map (SOFM) was used to determine the spatial control zone of soil and water conservation in the Dabie Mountain area on a small catchment scale. The results showed that: 1) The average potential and actual soil erosion were 84415.7 t/(km²·a) and 210.25 t/(km²·a), respectively. The actual soil erosion was distributed mainly in 0-300 t/(km²·a) at the small watershed scale. There was the basically same distribution of spatial patterns under the potential and actual soil erosion. The high-value area was distributed mainly in the central and eastern mountain areas at the high elevation. 2) Vegetation coverage and slope were the main influencing factors of potential and actual soil erosion at the small watershed scale, indicating a significantly positive correlation with the potential soil erosion (P<0.01). The high vegetation cover area was concentrated in the hinterland of Dabie Mountain. The high slope area was extended from the west to the east along the ridgeline of Dabie Mountain. 3) The SOFM results showed that the spatial control zone of soil and water conservation was divided into three areas: key prevention, general prevention, and the rest area at the small watershed scale. Among them, the key prevention area involved 710 small watersheds with an area of 1 5287.4 km². There were 890 small watersheds in the general prevention area, covering an area of 18 874.4 km². Two prevention areas accounted for 61.2% of the study area. There was an outstanding difference between actual and potential soil erosion and slope among regions. The classification index can serve as the spatial control of soil and water conservation. The finding can provide theoretical support and decision-making on the spatial control regionalization for soil and water conservation. [ABSTRACT FROM AUTHOR]

Peer-to-peer accommodation has gained prominence in the sharing economy and e-commerce sectors, with big data playing a crucial role in understanding customer preferences and evaluating homestay satisfaction. This study proposes a novel methodology that integrates Natural Language Processing (NLP) techniques, a Random Forest model, and Geographic Information System (GIS) functionalities to quantify the complex relationship between homestay satisfaction and diverse customer preferences. Notably, this study addresses the positive bias inherent in listing scores by segmenting homestays into three categories (satisfactory, moderate, and dissatisfactory) based on sentiment analysis from online reviews. Furthermore, this study not only identifies eight key determinants of homestay satisfaction but also unveils the nonlinear relationships and interactions between them. More significantly, we identify specific threshold values for geographic determinants, offering actionable recommendations for homestay planning and layout. These findings provide valuable insights that can be leveraged to improve homestay experiences and promote the sustainable development of urban homestays. [ABSTRACT FROM AUTHOR]

ABSTRACT Class imbalance is a critical challenge in big data analytics, often leading to biased predictive models. This imbalance can lead to biased models that perform well on the majority class but poorly on the minority class. Many machine learning models tend to be biased towards the majority class because they aim to minimise overall error, often leading to poor performance on the minority class. This paper presents the balanced ensemble neural network, a novel solution to effectively address class imbalance in big data. Balanced ensemble neural network combines the robust capabilities of neural networks with the power of ensemble learning, incorporating class balancing strategies to ensure fair representation of minority classes. The methodology involves integrating multiple neural networks, each trained on balanced subsets of data using techniques like Synthetic Minority Over‐sampling Technique and Random Undersampling. This integration aims to leverage the strengths of individual networks while reducing their inherent biases. Our extensive experiments across various datasets reveal that BENN achieves an AUC‐ROC score of 0.94, surpassing other models such as random forest (0.88), support vector (0.84) and single neural net (0.80). It was also observed that BENN's performance is better compared to traditional neural network models and standard ensemble methods in key metrics like accuracy, precision, recall, F1‐score and AUC‐ROC. The results specifically highlight BENN's effectiveness in accurately classifying instances of minority classes, a notable challenge in many existing models. These findings underscore BENN's potential as a substantial advancement in handling class imbalance within big data environments, offering a promising direction for future research and application in machine learning. [ABSTRACT FROM AUTHOR]

Since the inception of satellite remote sensing as a technology, characterization of forests has been one of its major applications. Characterization of forests at community level is essential for conservation, restoration and sustainable management of biodiversity. Recent advances in remote sensing offer opportunities to observe not only the reflectance spectra of forests from space, but also their phenology and structure. In this study, Earth Observation (EO) datasets were divided into 3 sets: spectral, structural and phenological. Then, Random Forest (RF) algorithm was applied on these 3 datasets along with field inventory-based tree data to generate community classification map of Purna wildlife sanctuary in Gujarat, India. The classification accuracy achieved from the spectral datasets (79.08–87.23%) was better than the phenological dataset (80.94%); and the latter in turn was better than the structural datasets (74.11–81.49%). An RF model with combination of the best predictors from the three datasets increased the classification accuracy upto 90.29%. In case of spectral dataset, the last image before the start of summer monsoon season gave the best accuracy. Also the new spectral bands which first became available in relatively newer satellites contributed significantly more to the model as compared to relatively older spectral bands which have been available in remote sensing satellites for quite some time. Overall, this study develops an empirical framework for mapping tropical tree communities by improving accuracy across the readily available remote sensing datasets and can be upscaled with sufficient field inventory data to generate a national level forest tree community map in India. [ABSTRACT FROM AUTHOR]

Groundwater nitrate pollution is a serious environmental problem worldwide, which is demonstrated influenced by land use/land cover (LULC) patterns. The current study was carried out to investigate the effects of LULC patterns to the nitrate pollution of shallow groundwater in the north piedmont plain of the Qinling Mountain (NPQM) using an ensemble machine learning method named random forest. Groundwater nitrate data and existing LULC patterns datasets were utilized to conduct the study. LULC patterns were quantified using a curved streamline shaped contributing area stratagem, and subsequently used to create training and test datasets together with the groundwater nitrate data for construction of random forest model. The results of this study indicated arable and urban land were the main LULC types in the NPQM, and urbanization induced the occupation of arable land by urban land from 2015 to 2019. Shallow groundwater in the NPQM was polluted by nitrate in both 2015 and 2019, with area of groundwater nitrate concentration exceeding the standard limitation recommended by the WHO (50 mg/L as NO3—) reduced from 2762.2 km2 in 2015 to 2184.3 km2 in 2019, showing an alleviating trend. Arable and urban land were the main LULC types contributing to groundwater nitrate pollution. Nitrate accumulated in the soil from manure and chemical fertilizer was the main source for groundwater nitrate pollution in arable land, while manure and sewage were the main source in urban land. The study provides scientific insights for sustainable groundwater protection in the NPQM. [ABSTRACT FROM AUTHOR]

In wafer testing, the test probe needs to contact the surface of the semiconductor wafer to measure electrical parameters such as resistance, capacitance, and current. The probe mark damage frequently occurs on the machine during wafer testing. This phenomenon causes inaccuracies in electrical parameter testing, significantly reducing the yield in IC packaging processes and resulting in substantial manufacturing cost losses. Therefore, this study proposed an effective probe mark damage detection and prediction method to prevent significant yield reduction due to inaccurate testing. This study uses importance analysis from random forests and correlation analysis to identify the critical parameters influencing probe mark damage. Introducing these parameters into a Sparse Transformer with hybrid normalization can successfully train an intelligent model for predicting the occurrence of probe mark damage. The model accurately predicts the probability of probe mark damage and promptly adjusts machine parameters to avoid inaccuracies in electrical parameter settings. The proposed approach can outperform other methods, achieving two very high accuracies of 95.1% (at room temperature) and 93.5% (at high temperature) and significantly reducing the occurrence of large-area probe mark damage. [ABSTRACT FROM AUTHOR]

The accurate prediction of the state of health (SOH) for lithium-ion batteries is a key factor for improving the performance of battery management systems (BMS). However, traditional point prediction methods are difficult to effectively eliminate errors due to the uncertainty of variables and application environments. This paper presents a model for predicting the interval of lithium-ion batteries based on health indicators (HIs) during charging, which addresses the limitations of current point prediction in practical applications. First, twelve HIs are extracted from the current and voltage variables of the charging process. Secondly, feature selection is performed by random forest (RF) training, and the selected HIs are dimensioned using partial least squares (PLS). Finally, a long short-term memory network (LSTM) combined with quantile regression (QR) is used to derive the quantile values of the prediction points and each quantile is employed as input information for Gaussian kernel density estimation (KDE) to obtain the SOH probability density distribution. The experimental results based on the NASA PCOE Li-ion battery dataset and CALCE Li-ion battery dataset show that the SOH interval coverage is more than 90% and the average width of the interval is less than 0.294. [ABSTRACT FROM AUTHOR]

Depression detection through social media analysis has emerged as a promising approach for early intervention and mental health support. This study evaluates the performance of various machine learning and transformer models in identifying depressive content from tweets on X. Utilizing the Sentiment140 and the Suicide-Watch dataset, we built several models which include logistic regression, Bernoulli Naive Bayes, Random Forest, and transformer models such as RoBERTa, DeBERTa, DistilBERT, and SqueezeBERT to detect this content. Our findings indicate that transformer models outperform traditional machine learning algorithms, with RoBERTa and DeBERTa, when predicting depression and suicide rates. This performance is attributed to the transformers' ability to capture contextual nuances in language. On the other hand, logistic regression models outperform transformers in another dataset with more accurate information. This is attributed to the traditional model's ability to understand simple patterns especially when the classes are straighforward. We employed a comprehensive cross-validation approach to ensure robustness, with transformers demonstrating higher stability and reliability across splits. Despite limitations like dataset scope and computational constraints, the findings contribute significantly to mental health monitoring and suggest promising directions for future research and real-world applications in early depression detection and mental health screening tools. The various models used performed outstandingly. [ABSTRACT FROM AUTHOR]

Background: Colorectal cancer (CRC) is the third most frequently diagnosed cancer worldwide. Diabetes and CRC share many overlapping lifestyle risk factors such as obesity, heavy alcohol use, and diet. This study aims to develop a risk scoring system for CRC prediction among diabetes patients using routine medical records. Methods: A retrospective cohort study was conducted using electronic health records of Hong Kong. Patients who received diabetes care in public general outpatient clinics between 2010 and 2019 and had no cancer history were identified, and followed up until December 2019. The outcome was diagnosis of CRC during follow-up. For model building, predictors were first selected using random survival forest, and weights were subsequently assigned to selected predictors using Cox regression. Results: Of the 386,325 patients identified, 4,199 patients developed CRC during a median follow-up of 6.2 years. The overall incidence rate of CRC was 1.93 per 1000 person-years. In the final scoring system, age, waist-to-hip ratio, and serum creatinine were included as predictors. The C-index on test set was 0.651 (95%CI: 0.631-0.669). Elevated serum creatinine (=127 µmol/L) could be a potential important predictor of increased CRC risk. Conclusion: While obesity is a well-known risk factor for CRC, renal dysfunction could be potentially linked to an elevated risk of CRC among diabetes patients. Further studies are warranted to explore whether renal function could be a potential parameter to guide screening recommendation for diabetes patients. [ABSTRACT FROM AUTHOR]

This article is focused mainly on verifying the suitability of data from the experimental milling of heat-treated beech wood and on investigating the effects of the technical and technological parameters of milling on the energy consumption of this process. The independent parameters of the machining process are the cutting speed, feed speed, rake angle, and hydrothermal modification of the experimental wood material. Based on analysis of variance, it can be argued that the cutting speed and rake angle of the tool have the greatest statistically significant effect on energy consumption, while the feed speed has the least influence. The measured data on cutting power during milling were used to build a regression model and validate it, and the most suitable type of model, with a correlation of 87%, is the classification and regression tree, followed by a model created using the random forest method. [ABSTRACT FROM AUTHOR]

The acceleration of urbanization has resulted in a heightened awareness of the impacts of urban environments on residents' emotional states. This present study focuses on the Lixia District of Jinan City. By using urban street view big data and deep learning methods, we undertook a detailed analysis of the impacts of urban color features on residents' emotional perceptions. In particular, a substantial corpus of street scene image data was extracted and processed. This was performed using a deep convolutional neural network (DCNN) and semantic segmentation technology (PSPNet), which enabled the simulation and prediction of the subjective perception of the urban environment by humans. Furthermore, the color complexity and coordination in the street scene were quantified and combined with residents' emotional feedback to carry out a multi-dimensional analysis. The findings revealed that color complexity and coordination were significant elements influencing residents' emotional perceptions. A high color complexity is visually appealing, but can lead to fatigue, discomfort, and boredom; a moderate complexity stimulates vitality and pleasure; high levels of regional harmony and aesthetics can increase perceptions of beauty and security; and low levels of coordination can increase feelings of depression. The environmental characteristics of different areas and differences in the daily activities of residents resulted in regional differences regarding the impacts of color features on emotional perception. This study corroborates the assertion that environmental color coordination has the capacity to enhance residents' emotions, thereby providing an important reference point for urban planning. Planning should be based on the functional characteristics of the region, and color complexity and coordination should be reasonably regulated to optimize the emotional experiences of residents. Differentiated color management enhances urban aesthetics, livability, and residents' happiness and promotes sustainable development. In the future, the influences of color and environmental factors on emotions can be explored in depth, with a view to assist in the formulation of fine urban design. [ABSTRACT FROM AUTHOR]

Aflatoxin B1 is a toxic substance in almonds, other nuts, and grains that poses potential serious health risks to humans and animals, particularly in warm, humid climates. Therefore, it is necessary to remove aflatoxin B1 before almonds enter the supply chain to ensure food safety. Hyperspectral imaging (HSI) is a rapid, non-destructive method for detecting aflatoxin B1 by analyzing specific spectral data. However, HSI increases data dimensionality and often includes irrelevant information, complicating the analysis process. These challenges make classification models for detecting aflatoxin B1 complex and less reliable, especially for real-time, in-line applications. This study proposed a novel hybrid spectral band selection algorithm to detect aflatoxin B1 in almonds based on multilayer perceptron (MLP) network weights and spectral refinement (W-SR). In the proposed process, the hyperspectral imaging (HSI) spectral rank was firstly generated based on MLP network weights. The rank was further updated using a spectral confidence matrix. Then, a spectral refinement process identified more important spectra from the lower-ranked ones through iterative processes. An exhaustive search was performed to select an optimal spectral subset, consisting of only the most significant spectral bands, to make the entire process suitable for real-time, in-line aflatoxin B1 detection in industrial environments. The experimental results using the artificially contaminated almonds dataset achieved a cross-validation accuracy of 98.67% with an F1-score of 0.982 for the standard normal variate (SNV) processed data with only four spectral bands. Comparative experiment results showed that the proposed MLPW-SR spectral band selection algorithm outperforms baseline methods. [ABSTRACT FROM AUTHOR]

Search engines are significant tools for finding and retrieving information. Every day, many new web pages in various languages are added. The threats of cyberattacks are expanding rapidly with this massive volume of data. The majority of studies on the detection of malicious websites focus on English-language websites. This necessitates more studies on malicious detection on Arabic-content websites. In this research, we aimed to investigate the security of Arabic-content websites by developing a detection tool that analyzes Arabic content based on artificial intelligence (AI) techniques. We contributed to the field of cybersecurity and AI by building a new dataset of 4048 Arabic-content websites. We created and conducted a comparative performance evaluation for four different machine-learning (ML) models using feature extraction and selection techniques: extreme gradient boosting, support vector machines, decision trees, and random forests. The best-performing model was then integrated into a Chrome plugin, created based on a random forest (RF) model, and utilized the features selected via the chi-square technique. This produced plugin tool attained an accuracy of 92.96% for classifying Arabic-content websites as phishing, suspicious, or benign. To our knowledge, this is the first tool designed specifically for Arabic-content websites. [ABSTRACT FROM AUTHOR]

Landslide hazards affect the security of human life and property. Landslide hazard maps are essential for landslide prevention and mitigation. In this study, the success of machine learning and statistical methods in predicting landslide hazards in and around the district center of Maden, Elazığ province, within the borders of Turkey, was analyzed, and their performances were compared. The Random Forest method correctly predicted 1.398 of the 1.425 landslide points in the training dataset, but was incorrect on 27 points. The same method predicted 1942 of the 2075 landslide-free points in the training dataset, but incorrectly predicted 133 points as landslide-exposed. As a result of the study, it is evident that the Random Forest and M5P Rule Tree methods yield more successful results than the Frequency Ratio method. In the study area, the landslide hazard is concentrated in areas close to the East Anatolian Fault and in areas with steep slopes. Lithology, slope, and seismicity have been identified as important triggering factors for landslides in the region. It is expected that machine learning methods, which operate with high levels of accuracy, will make a significant contribution to the prediction of landslide hazards. [ABSTRACT FROM AUTHOR]

The biggest issue many students face in today's world is choosing the right career. Especially when there are so many options available to them and the counselling options available are very limited or not very efficient, Career counselling is a very essential process that assists individuals in making informed decisions about their career paths. The use of machine learning in career counselling has gained so much attention due to its potential to analyse vast amounts of data and provide personalised guidance. Previously, there had been so much work done in this field with the help of artificial intelligence and machine learning, but there was a lack of a systematic system where students could explore each and every option thoroughly and get to know what the real outcome would be if they chose that stream. In this study, various factors such as the student's interests, hobbies, past academics and performances, and achievements are taken into consideration to predict the right career option. The model is trained using five different machine learning algorithms: decision tree, Random Forest, Support Vector Machine, Nave Bayes, and K-nearest neighbours Classifier. Out of these, Random Forest gave the highest accuracy of 84.17%, and after hypertuning, it gave the highest accuracy of 85.68%. We also gave some manual inputs to the system and found out that the Random Forest gave the highest accuracy of 85.71%. The prediction results of each algorithm are summarised in this study. [ABSTRACT FROM AUTHOR]

The episodes of elevated concentrations of different gaseous pollutants and particulate matter (PM) are of major concern worldwide, especially in city agglomerations. Krakow is an example of an urban–industrial agglomeration with constantly occurring PM10 air limit value exceedances. In recent years, a number of legislative actions have been undertaken to improve air quality in this area. The multitude of factors affecting the emergence of cases of very high air pollutant concentrations makes it difficult to analyze them using simple statistical methods. Machine learning (ML) methods can be an adequate option, especially when proper amounts of credible data are available. The main aim of this paper was to examine the influence of various factors (including main gaseous pollutant concentrations and some meteorological factors) on the effect of high PM10 concentration episodes in the ambient air in Krakow (Poland) using the random forest algorithm. The original methodology based on the PM10 limit and binary classification of cases with and without the occurrence of high concentration episodes was developed. The data used were derived from routine public air quality monitoring and a local meteorological station. A range of random forest classification models with various predictor sets and for different subsets of the observations coupled with variable importance analysis were performed. The performance of the algorithm was assessed using confusion matrices. The variable importance rankings revealed, among other things, the dominant impact of the mixing layer height on elevated PM10 concentration episode formation. This research work showed the usefulness of the random forest algorithm in identifying factors contributing to poor air quality, even in the absence of reliable emission data. [ABSTRACT FROM AUTHOR]

Achieving sustainable agricultural advancements necessitates optimizing crop yields while maintaining environmental stewardship. Our research addresses this critical imperative by introducing an innovative predictive model that refines crop recommendation systems through advanced machine learning techniques, specifically random forest and SHapley Additive exPlanations (SHAP). This study aims to overcome the limitations of traditional advisory approaches by incorporating interpretability tools, clarifying the model's decision‐making process around specific instances. To enhance the model's local interpretability, we incorporated local interpretable model‐agnostic explanations (LIMEs), providing transparent explanations for each crop recommendation, which fosters user trust, particularly when predictions diverge from established expert opinions. We conducted our empirical investigation using a comprehensive dataset that includes various agricultural parameters, historical crop yields, and environmental conditions to evaluate the model's performance. Our findings indicate a significant improvement in predictive accuracy over traditional methods. The application of SHAP values offers a groundbreaking analysis of feature importance, enabling a precise quantification of the contributions of factors such as soil quality, climatic variables, and historical crop performance to the predictive outcomes. This research advances the field of precision agriculture by presenting a model that excels not only in accuracy but also in providing actionable insights through enhanced interpretability. By balancing advanced predictive capabilities with user‐centric explanations, our model represents a substantial step forward in developing data‐driven, transparent, and trustworthy agricultural advisories. [ABSTRACT FROM AUTHOR]

Background: Quality caregiver-infant interaction is crucial for infant growth, health, and development. Traditional methods for evaluating the quality of caregiver-infant interaction have predominantly relied on rating scales or observational techniques. However, rating scales are prone to inaccuracies, while observational techniques are resource-intensive. The utilization of easily collected medical records in conjunction with machine learning techniques offers a promising and viable strategy for accurate and efficient assessment of caregiver-infant interaction quality. Methods: This study was conducted at a follow-up outpatient clinic at two tertiary maternal and infant health centers located in Shanghai, China. 68 caregivers and their 3-15-month-old infants were videotaped for 3–5 min during playing interactions in non-threatening environment. Two trained experts utilized the Infant CARE-Index (ICI) procedure to assess whether the caregivers were sensitive or not in a dyadic context. This served as the gold standard. Predictors were collected through Health Information Systems (HIS) and questionnaires, which included accessible features such as demographic information, parental coping ability, infant neuropsychological development, maternal depression, parent-infant interaction, and infant temperament. Four classification models with fivefold cross-validation and grid search hyperparameter tuning techniques were employed to yield prediction metrics. Interpretable analyses were conducted to explain the results. Results: The score of sensitive caregiver-infant interaction was 6.34 ± 2.62. The Random Forest model gave the best accuracy (83.85%±6.93%). Convergent findings identified infant age, care skills of infants, mother age, infant temperament-regulatory capacity, birth weight, positive coping, health-care-knowledge-of-infants, type of caregiver, MABIS-bonding issues, ASQ-Fine Motor as the strongest predictors of interaction sensitivity between infants and their caregiver. Conclusions: The proposed method presents a promising and efficient approach that synergistically combines rating scales and artificial technology to detect important features of caregiver-infant interactions. This novel approach holds several implications for the development of automatic computational assessment tools in the field of nursing studies. [ABSTRACT FROM AUTHOR]

Brain tumor still proves to be one of the major causes of death in the field of cancer. The chances of a person surviving more than 10 years after getting a brain tumor is quite low with different ranges as per age, country and other factors. Prediction of the tumor is a large topic with various algorithms and techniques being used such as imaging methods, machine learning and deep learning models. The models used in the majority of the work doesn't have much comparison with other models to hold the ground and also has insufficient accuracy and other evaluation parameters. Also, the lack of post-processing of the data makes the resultant data unclear and the existence and location of the tumor unclear. Thus, this leads to the need of a system which compare various major models to get the most accurate model with a verification method being applied to it. Moreover, to post process the resultant data for making the resultant image visible clearly to understand the existence of tumor and its location easily. In this work, we propose a new brain tumor prediction system with the incorporation of newly developed segmentation method and CNN for predicting the tumor effectively. The proposed segmentation method applies K-Means clustering algorithm and Mini-Batch K-Means Clustering algorithm for performing effective segmentation process. Here, it performs the segmentation process on the positive tumor datasets that clarifies the existence and location of the tumor. Moreover, the segmented image undergoes 4 major morphological operations such as erosion, dilation, intensity linear transformation and de-noisification to make the resultant image clearly. In addition, the proposed system uses the improved CNN for predicting the tumor disease. Finally, the proposed system is proved as better than the existing classifiers such as Logistic Regression, Support Vector Classifier, K Nearest Neighbour, Random Forest and Decision Tree by conducting experiments with MRI brain images in terms of prediction accuracy, precision, recall, f1-score and time taken for prediction. [ABSTRACT FROM AUTHOR]

Breast cancer (BC) is known as the most prevalent form of cancer among women. Recent research has demonstrated the potential of Machine Learning (ML) techniques in predicting the five-year BC risk using personal health data. Support Vector Machine (SVM), Random Forest, K-NN (K-Nearest Neighbour), Naive Bayes, Neural Network, Decision Tree (DT), Logistic Regression (LR), Discriminant Analysis, and their variants are commonly employed in ML for BC analysis. This study investigates the factors influencing the performance of ML techniques in the domain of BC prevention, with a focus on dataset size and feature selection. The study's goal is to examine the effect of dataset cardinality, feature selection, and model selection on analytical performance in terms of Accuracy and Area Under the Curve (AUC). To this aim, 3917 papers were automatically selected from Scopus and PubMed, considering all publications from the previous 5 years, and, after inclusion and exclusion criteria, 54 articles were selected for the analysis. Our findings highlight how a good cardinality of the dataset and effective feature selection have a higher impact on the model's performance than the selected model, as corroborated by one of the studies, which gets extremely good results with all of the models employed. [ABSTRACT FROM AUTHOR]

Distributed Denial of Service (DDoS) attacks continue to pose a significant threat to network infrastructures, exploiting vulnerabilities within existing security protocols and disrupting the seamless availability of online services. The intricate interconnections of nodes within computer networks contribute to the dynamic structure of this environment, complicating efforts to establish a secure and productive user experience. Effectively mitigating DDoS attacks in this complex networked setting remains a challenge. While current strategies primarily rely on anomaly detection and signature-based techniques, utilizing statistical analysis and predefined patterns to identify and thwart attacks, none have consistently demonstrated efficacy or reliability. Consequently, there is a compelling need for advancements in security mechanisms to address DDoS threats more effectively. This research introduces an innovative and highly efficient approach that incorporates various classification algorithms, including Random Forest, Decision Tree, Gradient Boosting, Linear SVM, Logistics, K-nearest neighbors (KNN), and AdaBoost, for DDoS attack detection. The performance of these machine learning classifiers is evaluated using key metrics such as accuracy, recall, F1-score, and precision. Remarkably, experimental results reveal outstanding accuracy rates, with Random Forest achieving the highest accuracy in detecting attacks. Additionally, a genetic algorithm is employed to select optimal features from the dataset, further enhancing the performance of the classifiers. This results in a notable 25% increase in accuracy, surpassing AdaBoost and Logistics, with K-nearest neighbors emerging as the top performer in terms of accuracy. [ABSTRACT FROM AUTHOR]

The aim of this study is to evaluate the utility of an artificial neural network (ANN) model in diagnosing idiopathic generalized epilepsy (IGE) and to compare the results of the diagnostic model constructed by combining the expression levels of miR‐146a, miR‐155, and miR‐132 genes using ANN, random forest (RF), and discriminant analysis (DA). qRT‐PCR is employed to determine the expression levels of the three miRNA genes. Forty‐six IGE patients and 51 healthy controls were included in the study. Three genetic biomarkers were employed to assess the discriminative power of the disease, and they were combined using ANN. Additionally, the performance of ANN was compared with RF and DA. Compared to healthy controls, the miR‐132 gene was significantly higher (p < 0.001) and the miR‐155 and miR‐146a genes were significantly lower in IGE patients (p < 0.001). The area under the curve (AUC) for predictions made by the ANN, RF, and DA were 0.96, 0.87, and 0.75, respectively, with accuracy rates of 0.96, 0.88, and 0.76, respectively. We demonstrate that ANN exhibits the highest accuracy, AUC, sensitivity, and specificity values among the three methods. The obtained results indicate that the combination of the three genes used as markers in IGE plays a significant role in the diagnosis of the disease. Instead of assessing biomarkers individually for the disease, combining them using machine learning methods leads to improved model performance. Additionally, not relying on a single genetic biomarker for the disease enables discrimination based on the collective impact of all biomarkers. [ABSTRACT FROM AUTHOR]

Stroke has remained a major cause of mortality and disability in the United States for years, and its recurrence significantly increased the risks. For predicting stroke recurrence, traditional data aggregation methods have limitations in effectively handling the numerous subcategories of stroke risk factors. This pilot study proposed a Segmented Neural Network-Driven Aggregation (SNA) method, and it aimed to improve the prediction model's accuracy. Utilizing the TriNetX diagnosis dataset, we processed various risk factors and demographic information through traditional and our proposed data aggregation techniques. We applied logistic regression and random forest classifiers to predict stroke recurrence. Our findings revealed that using the SNA method significantly outperformed other aggregation methods for both classifiers. Using the SNA method with a random forest classifier achieved higher accuracy (84.2%) and a better balance between sensitivity and specificity (AUC of ROC = 0.928, AUC of PR = 0.940) compared to other combinations. These results showed the potential of machine-learning supervised encoding methods in stroke recurrence predictions, providing implications for clinical practice and future epidemiological research. [ABSTRACT FROM AUTHOR]

AbstractSocial sensing data, including points-of-interest (POI) and mobile position, are important data sources for population mapping. However, existing studies using POI data disregard the heterogeneity in facility size among the same type of POI and urban–rural differences. Moreover, mobile position data face biased issues. This study presents a hybrid model that considers facility-based service capacity (FSC) and land livability to map fine-scale population distributions. Based on extracting the FSC index by integrating POI, mobile phone positioning (MPP), and road network data, the district-level census population was disaggregated into 100-m grids using a random forest model. Subsequently, a regression equation was developed from the land use data to correct the estimated residuals. The results showed that the hybrid model exhibited considerably smaller errors than those of the POI density-based method and direct mapping of MPP (RMSE = 5,393.31, 7,348.91, and 7,824.76, respectively), effectively reducing population misallocation in extreme-density regions. Compared to WorldPop and LandScan datasets and a comparative model, our method reduced the RMSE by 30–60%, proving the effectiveness of integrating various social sensing and remote sensing data for improving population mapping. This study improves on an existing method as a thought-provoking step toward advancement in fine-scale population mapping research. [ABSTRACT FROM AUTHOR]

This study introduces a method for generating tyre tread images that are highly rated in terms of ‘slip resistance’ sensory evaluations. It involves constructing a regression model to predict ‘slip resistance’ sensory evaluations in tyre tread images and subsequently developing an image generation model informed by this regression model. The regression model was developed using a machine learning framework based on ensemble learning and SHapley Additive exPlanations. The model was then used to filter training data for the image generation process, leading to the creation of a tyre tread image generation model aimed at producing images with high ‘slip resistance’ sensory evaluations, using StyleGAN. The proposed method was applied in a case study using real data, and its effectiveness was validated by comparison with a multiple regression analysis approach. The results revealed that more accurate results were achieved with the random forest (RF) regression model than with the multiple regression model due to the RF regression model’s emphasis on non-linear features. Furthermore, a significant difference was observed between the ‘slip resistance’ sensory evaluation values of the images generated by StyleGAN trained with training data screened by the RF regression model and those from the training data. [ABSTRACT FROM AUTHOR]

Forecasting typhoon waves during typhoons is crucial. In this paper, the numerical wave model SWAN was enhanced through integration with two machine learning methods: the Back Propagation Neural Network and Random Forest. This integration facilitated the development of two distinct models, namely SWAN-BP and SWAN-Tree. Through correlation analysis, key input features were identified for the machine learning models. The forecasts from the SWAN model were subsequently utilized as inputs to enhance further wave prediction. These hybrid models were validated using data from Typhoon Doksuri (2023) and Typhoon Nesat (2017). The results indicated significant improvements in predicting typhoon-induced wave heights with both the SWAN-BP and SWAN-Tree models compared to the original SWAN model. Specifically, the SWAN-BP model demonstrated a 33% improvement in accuracy for the Typhoon Doksuri, whereas the SWAN-Tree model exhibited a 24% improvement. For Typhoon Nesat, the accuracy improvements were 23% for the SWAN-BP model and 21% for the SWAN-Tree model. These findings demonstrate that integrating wave numerical models with machine learning techniques can significantly enhance the predictive accuracy of numerical models. This approach offers a cost-effective means to improve the existing wave forecasting database. Traditionally, the direct use of meteorological and oceanographic data for typhoon wave prediction might be compromised by biases inherent in the numerical wave models. However, the SWAN-BP and SWAN-Tree models effectively reduce these biases, thereby providing more accurate and robust predictions. In conclusion, this paper enhances the predictive accuracy of the SWAN model and establishes a crucial foundation for more precise typhoon wave forecasting through the application of machine learning techniques. [ABSTRACT FROM AUTHOR]

To achieve precise energy supply and energy-efficient operation of the steam system in the primary processing workshop of tobacco enterprises, a Random Forest prediction model is introduced in this study to forecast steam consumption. A correlation analysis is conducted to identify the factors that significantly impact steam load. These key features are then used to develop the Random Forest prediction model, with appropriate hyper parameters selected to enhance prediction accuracy. The results indicate that the Random Forest model offers high prediction precision and strong practical applicability, providing a basis for optimizing the scheduling of the steam system. [ABSTRACT FROM AUTHOR]

The application of carbon dioxide (CO2) in enhanced oil recovery (EOR) has increased significantly, in which CO2 solubility in oil is a key parameter in predicting CO2 flooding performance. Hydrocarbons are the major constituents of oil, thus the focus of this work lies in investigating the solubility of CO2 in hydrocarbons. However, current experimental measurements are time-consuming, and equations of state can be computationally complex. To address these challenges, we developed an artificial intelligence-based model to predict the solubility of CO2 in hydrocarbons under varying conditions of temperature, pressure, molecular weight, and density. Using experimental data from previous studies, we trained and predicted the solubility using four machine learning models: support vector regression (SVR), extreme gradient boosting (XGBoost), random forest (RF), and multilayer perceptron (MLP). Among four models, the XGBoost model has the best predictive performance, with an R2 of 0.9838. Additionally, sensitivity analysis and evaluation of the relative impacts of each input parameter indicate that the prediction of CO2 solubility in hydrocarbons is most sensitive to pressure. Furthermore, our trained model was compared with existing models, demonstrating higher accuracy and applicability of our model. The developed machine learning-based model provides a more efficient and accurate approach for predicting CO2 solubility in hydrocarbons, which may contribute to the advancement of CO2-related applications in the petroleum industry. [ABSTRACT FROM AUTHOR]

Crosslinked polyethylene cable is an important equipment in 10 kV distribution system, and its safety is very important. Making scientific judgments on cable repair decisions can help improve the safety and reduce the economic cost. In view of this, a cable risk assessment and repair decision method based on K‑means clustering and random forest(RF) classification model is proposed. The method first defines the risk level and risk degree of the cable based on the insulation status of the cable. Then the K‑means clustering algorithm is used to cluster the multi‑aging index dataset and classify the risk level intervals to build a multi‑aging index risk matrix. Based on the risk matrix of the multi‑aging index, the classification labels corresponding to the multi‑aging index are determined by using the comprehensive weight method. Finally, the classification model of the repair ways of the cables is established and trained based on the RF algorithm, and the selection results of the repair ways are output. The average accuracy of the proposed method reaches 99.70%, achieving rapid and reliable repair decisions for cables. [ABSTRACT FROM AUTHOR]

Email is a medium in business communication that people use everyday and not only holds sensitive information but also identity for the user and organization. The uniqueness of information contained make phishing detection and remedy cannot be applied generally. Existing method applied in common on security software such as blacklisting keyword or email address is not enough to outpace evolving phishing method. Nowadays the implementation of machine learning and deep learning grow rapidly fast and the method used by machine learning and deep learning that can learn pattern of inputs and natural language processing making classification to detect email phishing promising, this study present proposed method of phishing mail classification by comparing the use of machine learning and deep learning model, The algorithm used in this paper are recurrent convolutional neural network and random forest with tf-idf and fasttext word embedding. The Dataset contain phishing and legitimate of an email gathered from a trading company in Indonesia. The dataset will be split into 70% for training and 30% for testing. As a result of this comparison, the random forest model with tf-idf word embedding achieve highest accuracy of 100% for the dataset used and the highest accuracy for recurrent convolutional neural network model with fasttext is 98.21%. [ABSTRACT FROM AUTHOR]

Coronary artery disease (CAD) presents a significant global health burden, with early and accurate diagnostics crucial for effective management and treatment strategies. This study evaluates the efficacy of human evaluators compared to a Random Forest (RF) machine learning model in predicting CAD risk. It investigates the impact of incorporating human clinical judgments into the RF model's predictive capabilities. We recruited 606 patients from the Department of Nuclear Medicine at the University Hospital of Patras, Greece, from 16 February 2018 to 28 February 2022. Clinical data inputs included age, sex, comprehensive cardiovascular history (including prior myocardial infarction and revascularisation), CAD predisposing factors (such as hypertension, dyslipidemia, smoking, diabetes, and peripheral arteriopathy), baseline ECG abnormalities, and symptomatic descriptions ranging from asymptomatic states to angina-like symptoms and dyspnea on exertion. The diagnostic accuracies of human evaluators and the RF model (when trained with datasets inclusive of human judges' assessments) were comparable at 79% and 80.17%, respectively. However, the performance of the RF model notably declined to 73.76% when human clinical judgments were excluded from its training dataset. These results highlight a potential synergistic relationship between human expertise and advanced algorithmic predictions, suggesting a hybrid approach as a promising direction for enhancing CAD diagnostics. [ABSTRACT FROM AUTHOR]

(1) Background: Accurate inversion of crop water content is key to making an intelligent irrigation decision. However, little effort has been devoted to accurately estimating the crop water content of winter wheat in the North China Plain. (2) Method: The crop water content of winter wheat was measured at jointing, flowering and grain-filling stages, respectively. UAV-based multispectral remote sensing images were used to calculate thirteen vegetation indices, including SAVI, EVI, R-M, NDRE, OSAVI, GOSAVI, REOSAVI, GBNDVI, NDVI, RVI, DVI, GNDVI, and TVI. Five machine learning (ML) algorithms (i.e., MLR, RF, PLSR, ElasticNet, and ridge regression) were adopted to estimate the crop water content of winter wheat at the three growth stages. The benchmark datasets, which include CWC as well as vegetation indices calculated based on spectral indices, were adopted to validate the performance of the ML models. (3) Results: The correlation coefficients ranged from 0.64 to 0.82 at different growth stages. The optimal vegetation indices were GNDVI for the jointing stage, NDRE for the flowering and the grain-filling stage, respectively. Among the five machine learning methods, random forest (RF) showed the best performance across the three growth stages, with its coefficient of determination (R2) of 0.80, or an increase by 20.1% than those of other models. In addition, the RMSE and RPD of the RF model at the flowering stage were 3.00% and 2.01, which significantly outperformed other models and growth stages. (4) Conclusion: This study may provide theoretical support and technical guidance for monitoring current water status in wheat crops, which is useful to develop a precise irrigation prescription map for local farmers. (5) Limitation: The main limitation of this study is that the sample size is relatively small and may not fully reflect the characteristics of the target groups. At the same time, subjectivity and bias may exist in the data collection, which may have a certain impact on the accuracy of the results. Future studies could consider expanding sample sizes and improving data collection methods to overcome these limitations. [ABSTRACT FROM AUTHOR]

Accurately classifying road signs is crucial for autonomous driving due to the high stakes involved in ensuring safety and compliance. As Convolutional Neural Networks (CNNs) have largely replaced traditional Machine Learning models in this domain, the demand for substantial training data has increased. This study aims to compare the performance of classical Machine Learning (ML) models and Deep Learning (DL) models under varying amounts of training data, particularly focusing on altered signs to mimic real-world conditions. We evaluated three classical models: Support Vector Machine (SVM), Random Forest, and Linear Discriminant Analysis (LDA), and one Deep Learning model: Convolutional Neural Network (CNN). Using the German Traffic Sign Recognition Benchmark (GTSRB) dataset, which includes approximately 40,000 German traffic signs, we introduced digital alterations to simulate conditions such as environmental wear or vandalism. Additionally, the Histogram of Oriented Gradients (HOG) descriptor was used to assist classical models. Bayesian optimization and k-fold cross-validation were employed for model fine-tuning and performance assessment. Our findings reveal a threshold in training data beyond which accuracy plateaus. Classical models showed a linear performance decrease under increasing alteration, while CNNs, despite being more robust to alterations, did not significantly outperform classical models in overall accuracy. Ultimately, classical Machine Learning models demonstrated performance comparable to CNNs under certain conditions, suggesting that effective road sign classification can be achieved with less computationally intensive approaches. [ABSTRACT FROM AUTHOR]

The onset of the COVID-19 pandemic has compelled a swift transformation in higher-education methodologies, particularly in the domain of course modality. This study highlights the potential for artificial intelligence and machine learning to improve decision-making in advanced engineering education. We focus on the potential for large existing datasets to align institutional decisions with student and faculty preferences in the face of rapid changes in instructional approaches prompted by the COVID-19 pandemic. To ascertain the preferences of students and instructors regarding class modalities across various courses, we utilized the Cognitive Process-Embedded Systems and e-learning conceptual framework. This framework effectively delineates the task execution process within the scope of technology-enhanced learning environments for both students and instructors. This study was conducted in seven Iranian universities and their STEM departments, examining their preferences for different learning styles. After analyzing the variables by different feature selection methods, we used three ML methods—decision trees, support vector machines, and random forest—for comparative analysis. The results demonstrated the high performance of the RF model in predicting curriculum style preferences, making it a powerful decision-making tool in the evolving post-COVID-19 educational landscape. This study not only demonstrates the effectiveness of ML in predicting educational preferences but also contributes to understanding the role of self-regulated learning in educational policy and decision-making in higher education. [ABSTRACT FROM AUTHOR]

The creation and spreading of fake information can be carried out very easily through the internet community. This pervasive escalation of fake news and rumors has an extremely adverse effect on the nation and society. Detecting fake news on the social web is an emerging topic in research today. In this research, the authors review various characteristics of fake news and identify research gaps. In this research, the fake news dataset is modeled and tokenized by applying term frequency and inverse document frequency (TFIDF). Several machine-learning classification approaches are used to compute evaluation metrics. The authors proposed hybridizing SVMs and RF classification algorithms for improved accuracy, precision, recall, and F1-score. The authors also show the comparative analysis of different types of news categories using various machine-learning models and compare the performance of the hybrid RFSVM. Comparative studies of hybrid RFSVM with different algorithms such as Random Forest (RF), naïve Bayes (NB), SVMs, and XGBoost have shown better results of around 8% to 16% in terms of accuracy, precision, recall, and F1-score. [ABSTRACT FROM AUTHOR]

This study contributes to the Sybil node-detecting algorithm in online social networks (OSNs). As major communication platforms, online social networks are significantly guarded from malicious activity. A thorough literature review identified various detection and prevention Sybil attack algorithms. An additional exploration of distinct reputation systems and their practical applications led to this study's discovery of machine learning algorithms, i.e., the KNN, support vector machine, and random forest algorithms, as part of our SybilSocNet. This study details the data-cleansing process for the employed dataset for optimizing the computational demands required to train machine learning algorithms, achieved through dataset partitioning. Such a process led to an explanation and analysis of our conducted experiments and comparing their results. The experiments demonstrated the algorithm's ability to detect Sybil nodes in OSNs (99.9% accuracy in SVM, 99.6% in random forest, and 97% in KNN algorithms), and we propose future research opportunities. [ABSTRACT FROM AUTHOR]

In this study, an early warning system predicting first‐year undergraduate student academic performance is developed for higher education institutions. The significant factors that affect first‐year student success are derived and discussed such that they can be used for policy developments by related bodies. The dataset used in experimental analyses includes 11,698 freshman students' data. The problem is constructed as classification models predicting whether a student will be successful or unsuccessful at the end of the first year. A total of 69 input variables are utilized in the models. Naive Bayes, decision tree and random forest algorithms are compared over model prediction performances. Random forest models outperformed others and reached 90.2% accuracy. Findings show that the models including the fall semester CGPA variable performed dramatically better. Moreover, the student's programme name and university placement exam score are identified as the other most significant variables. A critical discussion based on the findings is provided. The developed model may be used as an early warning system, such that necessary actions can be taken after the second week of the spring semester for students predicted to be unsuccessful to increase their success and prevent attrition. [ABSTRACT FROM AUTHOR]

Word frequency has a long history of being considered the most important predictor of word difficulty and has served as a guideline for several aspects of second language vocabulary teaching, learning, and assessment. However, recent empirical research has challenged the supremacy of frequency as a predictor of word difficulty. Accordingly, applied linguists have questioned the use of frequency as the principal criterion in the development of wordlists and vocabulary tests. Despite being informative, previous studies on the topic have been limited in the way the researchers measured word difficulty and the statistical techniques they employed for exploratory data analysis. In the current study, meaning recall was used as a measure of word difficulty, and random forest was employed to examine the importance of various lexical sophistication metrics in predicting word difficulty. The results showed that frequency was not the most important predictor of word difficulty. Due to the limited scope, research findings are only generalizable to Vietnamese learners of English. [ABSTRACT FROM AUTHOR]

This paper presents deep learning using LSTM, machine learning employing RF and GB algorithms, and the rating curve (RC) that can be used for estimating daily streamflow at the outlet of river basins. The Kone River basin in Vietnam is selected as an example for demonstrating the ability of these approaches. Hydro-meteorological data, including rainfall at Vinh Kim as well as water level and streamflow at Binh Tuong, were collected in the long period from 1/1/1979 to 31/12/2018. Multiple approaches mentioned above are implemented and applied for estimating daily streamflow at Binh Tuong in the Kone River basin. Firstly, coefficients and hyper-parameters in each approach are carefully determined using available hydro-meteorological data from 1/1/1979 to 31/12/2009 and dimensional and dimensionless error indexes. The results revealed that deep learning using LSTM presents the most suitable performance of the observed streamflow, with correlation coefficient r and NSE being close unity, while RMSE and MAE are less than 1.5% of the observed magnitude of streamflow. The RC and machine learning employing RF and GB algorithms procedures acceptably the observed streamflow, with r and NSE varying between 0.77 and 0.98, and RMSE and MAE ranging from 0.4 to 6.0% of the observed magnitude of streamflow. Secondly, multiple approaches are also applied for estimating daily streamflow from 1/1/2010 to 31/12/2018, revealing consistent statistical characteristics of streamflow in the river basin. Finally, the impacts of input data on output streamflow are discussed. [ABSTRACT FROM AUTHOR]

As a special feature type, agricultural greenhouses were relatively easy to recognize on high spatial resolution remote sensing images with high accuracy. However, most high spatial resolution images had to be purchased commercially, and their availability was limited. In order to improve the economy and convenience of precise extraction of county-level agricultural greenhouses, this study used free and easily available non high spatial resolution Sentinel-1 (radar) and Sentinel-2 (optical) remote sensing images for data fusion, combined with spectral index, texture extraction, principal component analysis and other methods, to construct a multidimensional feature set space, and adopted multiple classification and recognition methods (cases) to identify and extract county-level agricultural greenhouses. The results indicated that: 1) The county’s agricultural greenhouses could be extracted with high precision by utilizing just Sentinel-1/2 (10 m resolution) remote sensing images, bolstered by suitable classification techniques (case). 2) The fusion of Sentinel-1 (radar) and Sentinel-2 (optical) remote sensing data could be helpful for improving the recognition accuracy of agricultural greenhouses. Compared to using only Sentinel-2 (optical) remote sensing data, the overall accuracy of Sentinel-1/2 data fusion had improved by an average of 1.72 percentage points, with a maximum improvement of 3.29 percentage points; 3) Of all the classification and recognition techniques (case) applied in the paper, the object-oriented method performed well in areas with high greenhouse density; But in areas with low greenhouse density, the accuracy was mediocre and exhibited a strong dependence on region (or scene). After the fusion of optical and radar information, the pixel based recursive feature elimination random forest (RF-RFE) method can achieve an average accuracy of 96.45%, with high and stable accuracy and strong regional adaptability. It was suitable for accurate and efficient extraction of agricultural greenhouses from non high-resolution image in county-level. The technical solution proposed in this paper based on Sentinel-1/2 images, could be provided technical support for the economic, rapid, and efficient extraction of agricultural greenhouses in most counties. [ABSTRACT FROM AUTHOR]

For precision agriculture, agricultural development planning, and regional food security, accurate crop mapping is essential. The GF-6 satellite offers novel approaches for crop identification. As Jinzhou, Hebei province, China, comprises a mix of grain and fruit trees, this research has selected the area for a case study. First, spectral, index, and texture features are combined in various classification feature schemes. Second, the random forest algorithm is employed to ascertain the significance of these features. Optimal features are then chosen to accurately identify winter wheat, summer maize, and pear orchards. Concurrently, land cover for wheat and maize seasons is mapped. The results demonstrate that: (1) multisource features considerably enhanced the accuracy of crop classification based on GF-6 WFV data. Among these, index features have the greatest contribution, followed by spectral features, while the contribution of texture features, is almost negligible. (2) The classification effect utilizing random forest feature optimization is superior, with an overall accuracy of 94% and a kappa coefficient of 0.91. The extracted land area for each scheme exhibits an accuracy of over 70% when compared to the statistical area. (3) The critical time phases for extracting winter wheat, summer maize, and pear orchards are identified as, in sequence, the leaf-out and jointing stage, the tasselling stage, the leaf-fall stage, fruit development, and the first flowering/full flowering stages. Based on GF-6 WFV data, these research findings will offer a technical reference for the identification of grain crops and the extraction of pear orchards. [ABSTRACT FROM AUTHOR]

Efficiently and accurately identifying the spatial distribution of tea plantations in the subtropical plateau regions of southwest China is of great significance for ecological and environmental protection. However, the lands of those regions are fragmented with complex vegetation types. Moreover, there is much cloudy and rainy weather over those areas, making it very difficult to identify tea plantations using only optical remote sensing data. In order to solve these problems, this paper uses Sentinel-1 (S1) Synthetic Aperture Radar (SAR) data and Sentinel-2 (S2) optical data to design seven classification feature combinations to explore the influence of red edge features, radar features and texture features on the identification accuracy of tea plantations. The feasibility of Jeffreys-Matusita distance (JM) feature selection and Recursive Feature Elimination (RFE) feature selection algorithm to find the optimal feature combination is verified, and the distribution of tea plantations in the study area is acquired by using the object-oriented random forest algorithm. The study shows that (1) the combination of SAR data and optical data can effectively improve the identification accuracy of tea plantations. (2) S2 red edge features and S1 radar features can significantly improve the accuracy of the identification results of tea plantations. (3) After applying the JM distance and RFE feature selection algorithms, the producer's accuracy of tea plantations is improved by 1.39% and 2.38%, and the user's accuracy is improved by 1.02% and 1.3%, respectively, compared with the identification of all features. The overall accuracy of the random forest algorithm combined with RFE is 93.43%. This study proposes the application of feature selection algorithms in identification of tea plantations, which improves accuracy and increases efficiency while minimizing redundant features and provides an effective approach to identify tea plantations in cloudy and rainy areas in the subtropical plateau of southern China. [ABSTRACT FROM AUTHOR]

The meteorological and tobacco leaf weight data from 53 districts (counties) in the tobacco-growing region of central and eastern Guizhou from 2010 to 2021 were selected to analyze the influence of meteorological factors on the leaf weight of flue-cured tobacco. Four linear and non-linear models were established using four meteorological factors on an annual scale (four-factor model) and multiple meteorological factors on a ten-day scale (multi-factor model), respectively. The accuracy and stability of the models were validated to explore the advantages and disadvantages of different models in the central and eastern regions of Guizhou. The results indicated: (1) the average single leaf weight of the lower, cutters, and upper leaves in the tobacco-growing areas of central and eastern Guizhou in the past 12 years were 5.7g, 9.2g, and 10.8g, respectively. The variation range in single leaf weight was smaller in the lower leaves than in the cutters and upper leaves. (2) Based on the modeling with four factors, the BP neural network algorithm model for the single leaf weight of lower leaves showed the best simulation effect, with R² of 0.26 and RMSE of 0.84g. The simulation effect of the random forest algorithm model exhibited the best simulation performance, with R² of 0.33 and RMSE of 1.08g for single leaf weight in the cutters leaf, and R² of 0.16 and RMSE of 1.59g for single leaf weight in the upper leaf. When model with multiple factors, the random forest algorithm model for the single leaf weight of the lower and upper leaves performed best, with R² of 0.22 and RMSE of 0.85g for the lower leaves and R² of 0.16 and RMSE of 1.57g for the upper leaves. For the cutters leaves, the BP neural network algorithm model demonstrated the optimal simulation effect, with R² of 0.18 and RMSE of 1.14g. (3) The accuracy of the multiple-factor stepwise regression algorithm for predicting the single-leaf weight of the lower leaves was 34.86%, while the simulation accuracy of other algorithms for the same leaf position was generally above 70%. The multi-factor BP neural network model predicted the highest accuracy of 86.24% in the lower single leaf weights, while the random forest model predicted the highest accuracy of 89.91% in the cutters single leaf weights under four-factor. The random forest algorithm with multifactor had the highest prediction accuracy of 84.4% for the upper leaf weight. (4) Based on the four-factor algorithm model for predicting the single-leaf weight in the central and eastern regions of Guizhou in 2021, the accuracy of the four algorithmic models for predicting the single-leaf weight of the cutters and lower leaves showed little difference. The accuracy of the random forest algorithm for predicting the single-leaf weight of the upper leaves was the highest, reaching 91%, which was significantly higher than the other three algorithmic models. Under the multi-factor condition, the average simulated accuracy of the various models for the lower, middle, and upper parts was 81%, 74%, and 85% respectively. (5) At the city-level dimension, under the condition of four factors, the average accuracy of the BP neural network model for simulating the single-leaf weight of the lower leaves was 84%. There was no significant difference in the simulation results of the four algorithmic models for the cutters leaves. The upper single leaf weight simulation resulted in the highest average accuracy of 92% with the random forest algorithm. Under the condition of multiple factors, the average accuracy of the linear regression model for simulating the single leaf weight of the lower leaves was 91%, while for the upper leaves, the linear regression model also had the highest average accuracy of 88%. In summary, for the average single-leaf weight of tobacco leaves in the central and eastern regions of Guizhou, using the four-factor BP neural network and random forest algorithm can better simulate the impact of meteorological factors on the single-leaf weight of tobacco leaves, providing scientific basis for the regulation of tobacco leaf production. [ABSTRACT FROM AUTHOR]