The research focused on the thermal decomposition characteristics and the kinetic and thermodynamic parameters of hexane extracts from Coccoloba uvifera leaf (CU), Byrsonima crassifolia bark (BYR), and Bursera copallifera resin (BUR). Additionally, it explored the relationship between these thermal properties and radical-scavenging activity (RSA). The preexponential factor (A), activation energy (Ea), enthalpy (ΔH), entropy (ΔS), and Gibbs free energy (ΔG) were obtained via the Coats–Redfern method. The CU extract showed higher RSA than the BYR and BUR extracts. In thermal analysis, the BYR and BUR extracts had two mass-loss events, whereas the CU extract had three. BUR had the lowest Ea, A, ΔH, and ΔS values, coinciding with its low RSA. Meanwhile, in CU, these parameters had the highest values. The thermal decomposition of the extracts was an endothermic process. Variations in the thermal profile were associated with composition and RSA.
Zinc production faces a significant challenge from cadmium (Cd2+). Therefore, this study investigated the cementation of cadmium from sulfate solutions with zinc powder. This work examined how pH of the solution, reaction time, and temperature affect cadmium cementation. The primary focus of the study was the impact of varying initial pH (1–5) on Cd2+ removal efficiency (%) at 20 minutes. Results showed maximum cadmium cementation efficiency at a pH of 4. The cadmium cementation process was also affected by reaction times (20, 40, 60, and 80 min). Zinc powder was used to study cadmium cementation reaction kinetics in sulfate solutions at various temperatures (50, 60, 70, and 80 ºC). We found that the reaction rate constant (K) rose with temperature, peaking at 80 ºC. This study found the best pH, reaction time, and temperature to be 4, 80 minutes, and 80°C, respectively. When various theories were used to calculate reaction kinetics, activation energies of 13.75297 and 12.18379 were revealed. These results indicate that the Arrhenius theory best explains the reaction mechanism in this study. SEM and EDS analyses revealed a cadmium deposit. These results offer more insight into cadmium cementation.
Geothermal energy is an efficient alternative for convective drying of food. This research aimed to determine the efficiency of a low-temperature geothermal dryer (LTGE) system in dehydrating high-moisture foods and its effects on the kinetic parameters and properties of the dehydrated foods. This study used a prototype laboratory-scale geothermal dehydrator built at the Geothermal Laboratory of the Engineering Institute of UNAM. The influence of the geothermal dehydrator operating parameters- average temperature (60, 65, 70 °C) and airflow velocity (2.5, 3.0, 3.3 m/s) on moisture content kinetics, drying rate, moisture-effective diffusion coefficient (D(eff)), physicochemical properties, and drying efficiency was determined. The results indicated that mean temperature and airflow rate influenced the drying kinetics and properties of dried apples and mangoes. With increasing temperature and airflow velocity, the drying rate and effective diffusion coefficient increased. The drying curves were dominated by the period of decreasing rate, which is typical of the drying kinetics of biological materials under constant-temperature conditions. The increase in temperature and airflow velocity increased the effective diffusion coefficient and drying efficiency. The Deff ranged from 0.43–8.59 x 10(–9) m(2)/s for the dried apple samples and from 5.54–8.92 x 10(–9) m(2)/s for the dehydrated mango samples. Moisture content curves showed moisture contents below 10% at drying times similar to those achieved with a conventional convective dryer, demonstrating that geothermal energy can be used for fruit dehydration, producing high-quality dehydrated products and enabling efficient drying processes.
Pili kernels are prone to oxidative deterioration and rancidity, impacting their shelf life and consumer appeal. Coatings play a pivotal role in preserving quality and extending shelf life, which improves the characteristics of food products. In this study, we coated pili kernels and adapted the layer-by-layer dip coating technique. The coating consists of shellac as the formable matrix embedded with rosemary extract as the anti-rancidity agent. Pili kernels coated with a larger number of coating layers are shown to have lower peroxide and free fatty acid values. These values, which are indicative of levels of rancidity, suggest favourable results in the use of shellac-rosemary extract blend as a coating for pili kernels.
Security systems traditionally rely on CCTV for monitoring spaces accessible only to authorized personnel, yet they struggle with face detection and recognition at distances beyond a few meters. This limitation hampers their effectiveness in enhancing room security. This study addresses this challenge by developing a remote facial recognition system utilizing CCTV cameras to identify faces from 1-3 meters away. We employed YOLOv5 and YOLOv8 algorithms, testing pre-trained models of varying sizes (M and X) to improve detection accuracy. The training phase involved 200 epochs with a batch size of 32, yielding mean Average Precision (mAP) scores of 82.7%, 83%, 85%, and 85.2% for YOLOv5m, YOLOv5x, YOLOv8m, and YOLOv8x, respectively. Offline evaluations demonstrated average accuracy rates of 94%, 95%, 90%, and 91%. Online testing, conducted under varying conditions with 1-3 faces visible, showed YOLOv5x achieving an accuracy of 87.8%, compared to 80.9% for YOLOv8x. The results indicate that while single-face recognition is quick and accurate, performance declines with multiple faces in view. This research offers a promising solution to enhance room security through effective facial recognition at a distance, highlighting the potential of improved surveillance technology in secure environments.
This paper presents an adaptive sensor fusion algorithm designed to enhance the performance of autonomous systems operating under adverse weather conditions. Traditional sensor fusion methods struggle with data inconsistencies caused by environmental factors such as fog, rain, and snow, which compromise the reliability of LiDAR and radar inputs. To address this challenge, we propose a novel fusion framework integrating machine learning-based adaptive weighting to dynamically adjust sensor contributions based on weather conditions. The proposed algorithm is validated using simulated and real-world datasets, demonstrating superior robustness, accuracy, and computational efficiency compared to state-of-the-art methods. Experimental results show an 18% improvement in obstacle detection accuracy and a 23% reduction in false positives under adverse conditions. These findings suggest significant potential for improving the safety and reliability of autonomous systems in real-world scenarios.
Tuberculosis (TB) remains a major global health challenge, causing approximately 1.5 million deaths annually and affecting over 10.6 million people worldwide as of 2021. In countries like South Africa, TB remains a leading cause of mortality. Caused by Mycobacterium tuberculosis, the disease primarily affects the lungs but can spread to other organs. Early and accurate diagnosis is crucial to reduce transmission and mortality rates. This review focuses on the role of optimization algorithms in enhancing machine learning (ML) and deep learning (DL) models for TB detection in medical imaging. It explores chest X-rays (CXR) images as the main diagnostic imaging data, while emphasizing the use of these optimization algorithms for image segmentation, feature selection, and hyperparameter tuning. This study evaluates the performance of seven optimization algorithms in improving TB detection accuracy: genetic algorithm (GA), surrogate algorithm, particle swarm optimization (PSO), pattern search (PS), particle swarm optimization with pattern search (PSOPS), genetic algorithm with pattern search (GAPS), and firefly algorithm. The algorithms were implemented using data from chest X-ray images. The results indicate that the top three performing algorithms are PSOPS (accuracy 78%, recall 80%, and specificity 78%), surrogate (accuracy 72%, recall 86%, and specificity 68%), and GAPS (accuracy 62%, recall 91%, and specificity 55%), based on comparisons with the ground truth image. The experiments and review in this study offer valuable insights for researchers and practitioners while identifying opportunities for future research. These insights can guide practitioners in choosing suitable optimization algorithms for TB detection, improving accuracy, efficiency, and scalability. Such improvements could enhance diagnostics, enabling early detection and intervention and thereby reducing the global TB burden.
The timely detection and pre-emption of cardiovascular diseases (CVDs) remain pivotal challenges in healthcare, necessitating innovative approaches in signal analysis and machine learning. Existing methodologies often fall short in precision, accuracy, and timeliness, underscoring the need for more sophisticated and integrated solutions. This paper presents a novel framework employing advanced deep learning architectures and federated learning techniques for enhanced photoplethysmography (PPG) signal analysis. Our approach integrates transformer networks and capsule networks to effectively capture temporal dependencies and spatial hierarchies in multidimensional PPG data, addressing limitations in current practices by significantly improving the precision and accuracy of the CVD detection process. We incorporate federated averaging algorithms and secure aggregation protocols to train models across multiple devices while ensuring data privacy levels. Further, our methodology leverages interpretable Deep SHAP, providing clarity and transparency in model decisions, a critical factor in clinical settings. The integration of multi-modal data through multiple input convolutional neural networks and recurrent neural networks with (LSTM) and bidirectional gated recurrent unit (BiGRU) networks allows for a comprehensive analysis of varied physiological signals. Additionally, our model employs sophisticated anomaly detection techniques, including autoencoders and Isolation Forest, for early and precise identification of unusual patterns in PPG signals. To cater to individual variances in physiological signals, we implement personalized and adaptive models using model-agnostic meta-learning with few-shot learning, ensuring tailored detection and monitoring processes. The unique blend of machine learning models and rulebased systems through ensemble methods further enhances the efficacy of our framework. Clinical testing across multiple heart diseases has demonstrated the superiority of our approach, showing significant improvements over existing methods in various metrics such as precision, accuracy, recall, aUC, specificity, and response delays. This work not only marks a significant advancement in the detection and pre-emption of CVDs but also sets a new benchmark for the application in medical diagnostics, promising substantial impacts on patient outcomes and healthcare practices.
Dissimilar metal welds between 2507 superduplex stainless steel and API X-52 carbon steel were joined using ER2594 via the gas metal arc welding process. Two joints were configured to be completed by single- and multipass welds to observe the effects of heat input and dilution on solidification across the fusion boundary region. Therefore, a dual etching technique was employed to reveal the microstructure across the interphase between the materials. The optical microscopy analysis shows the microstructural evolution of the weld metal from blocky austenite to Widmanstätten, as dilution promotes redistribution of elements across the fusion boundary region, as observed in the EDS analysis. However, the WRC-1992 diagram indicates that this occurs at dilutions above ~36%. On the other hand, the weld metal/API X-52 interphase comprises an unmixed zone that exhibits morphologies such as beach, bay, and island and is a function of the percentage of base metal in the weld metal, with rapid cooling due to the weld metal’s lower liquidus temperature. The tensile strength of both dissimilar welds is approximately equal and higher than that of the base metal, API X-52. The hardness distribution shows higher values in the weld metal/2507 of the multipass welding, reflecting the effect of heat input; however, it does not influence failure during the tensile test. This work analyzed and described the fusion boundary region in dissimilar welds for different joint designs and welding parameters.
Due to the exponential growth in the number of mobile applications, correctly categorizing app descriptions into genres is essential for enhancing the user experience, discoverability, and personalization. Traditional text classification models struggle to represent the intricate semantic relationships within descriptions and often miss complex multi-genre issues and context- dependent nuances. To address these problems, we introduce a classification framework using Graph Neural Networks (GNNs) with improved semantic representation, supported by deep learning, data science, and reinforcement learning. Our approach begins with a Hybrid Graph Convolutional Network (GCN) paired with a Support Vector Machine (SVM): it captures semantic relationships among words as graph nodes to improve SVM-based classification margins. This hybrid method improves structural learning and yields accurate genre classifications, expected to perform 5-8% better than Graph Convolutional Networks (GCNs) in isolation. In the final integration, we combine a Contextual Graph Attention Network (CGAT) with Bi-directional Encoder Representations from Transformers (BERT) embeddings to capture rich, complex contextual relationships; we expect the model to achieve 92-94% accuracy for multi-genre descriptions. We improve data diversity through topic modeling with Non-negative Matrix Factorization and semantic data augmentation for thematic components to enhance the generalization and explainability of the models. Furthermore, SHapley Additive exPlanations (SHAP) explains model decisions by quantifying the contributions that words make to genre predictions, bringing much-eeded transparency to the models. Curriculum Learning with PPO is also a progressive optimizer for the model, saving 10-12% of time and further improving accuracy. This integrated framework moves toward both increased classification accuracy and interpretability and provides a new baseline for semantic understanding in classification tasks across multiple genres of documents.
The lightweight design of 3D-printed polylactic acid (PLA) components requires optimization strategies that reduce material usage while preserving mechanical performance under loading conditions. In this study, cubic lattice topology optimization was applied to tensile and compression specimens manufactured by 3D printing using PLA. The mechanical properties of the base material were experimentally determined and incorporated into finite element analyses. The boundary conditions were defined to reproduce the experimental stress state under standardized testing, with tensile and compressive loads selected based on the material’s yield strength. The optimized geometries were subsequently fabricated by 3D printing and mechanically tested. A qualitative agreement was observed between the simulated and experimental responses, confirming that the gradient-driven optimization approach implemented in ANSYS provided physically representative and experimentally validated designs. The printed PLA exhibited ductile-like behavior attributed to the fused deposition modeling process. Thus, this work demonstrates the feasibility of integrating topology optimization, finite element analysis, and experimental validation to develop PLA components under realistic loading conditions.
The sudden appearance of high-quality image and video forgeries, such as deepfakes and splicing, has urgently called for more advanced and generalizable detection frameworks. Most existing forgery detection methods suffer from limited robustness and generalization across different forgery techniques and modalities. To address these limitations, we extend a unified multimodal image and video forgery detection framework by using improved feature extraction and fusion techniques. For image forgery detection, our framework combines the strengths of a DenseNet-based deep feature extraction technique with Haralick texture features to capture both spatial and texture-based manipulations. DenseNet is selected because, by using dense connections, it can reuse features in a very effective manner; hence, it provides a strong mechanism for detecting even fine-grained forgeries. It incorporates Haralick features into its architecture so that any texture anomalies arising from manipulations such as copy-move sets can be identified. The combination of these features is achieved via an attention-based mechanism that dynamically balances the contributions of both feature types based on the nature of the forgery. We also use a pre-trained EfficientNet-B3, fine-tuned with GAN-generated adversarial examples, to make our model more robust to sophisticated forgeries. In the video forgery detection framework, 3D ResNet is incorporated for spatiotemporal feature extraction, LSTM for capturing long-term temporal dependencies, and Temporal Convolutional Networks for ensuring short-term temporal consistency. A dual attention mechanism is utilized to emphasize manipulated spatial regions and key temporal intervals, thereby improving the accuracy of video forgery detection. It achieved competitive accuracy-95-97% on images and 92-95% for videos-along with improved adversarial robustness, while at the same time presenting a scalable solution for practical forgery detection across different domains.
Nanofibers possess properties that make them suitable for use in a variety of applications, including high specific surface area and biomimetic potential. This has led to numerous potential applications for electrospinning fibers. Precise control and prediction of nanofiber alignment and diameter are critical for these applications. Several variables affect fiber properties, including the collector’s speed and shape. In this research, the conventional collector used in electrospinning technology was modified by using a polyamide collector in the form of a cone-shaped trunk, and the effect of changing the surface inclination angle on the properties of the resulting fibers was studied. The effect of the conical collector’s rotational speed on fiber morphology, in terms of diameter alignment and density, was also studied. 5 different rotational speeds were applied within the range of (0-6000) rpm, and three inclination angles of 10, 15, and 30 degrees were tested, with the remaining process parameters held constant. The results showed that increasing the inclination angle of the surface of the cone increases the diameter of the produced fibers and reduces their density. Regarding the orientation of the fibers, the study showed that high rotational speeds (4500 rpm and above) produced clear parallelism and alignment between the resulting nanofibers and those taken from the base side of the conical collector, with 82% alignment relative to the diameters of the formed fibers. On the collector, the results showed that a rotational speed of 3000 rpm gave the smallest fiber diameter, reaching 50.07 nm. Also, samples taken from the base side showed identical results, and the smallest diameter measured was 40.1 nm. By comparing the fiber diameters taken from the two bases of the cone, the results showed the presence of a gradient in the alignment, diameters, and density, allowing the formation of a three-dimensional structure for the resulting fiber network. This study demonstrates the importance of calibrating the rotational speed of the collector and its inclination angle in determining the optimum values to obtain three-dimensional fibers at the nanoscale, which can be used in many potential applications.
This study examines the use of recycled materials in sustainable construction, specifically a rice husk-newspaper-PVAc-borax composite made from recycled newspaper cellulose (9%), rice husk (14%), borax (15%), and polyvinyl acetate-PVAc (62%). Tests for water absorption, density, fire resistance, and mold growth were conducted following ASTM and European standards. The composite showed high water absorption but improved moisture resistance due to rice husk and borax. Its intermediate density balances strength and lightness, making it suitable for various applications. Fire tests revealed reduced fire propagation in samples containing borax, enhancing fireproofing properties. Borax also inhibited fungal growth, aligning with previous studies. While these results are promising, further research is needed to evaluate the composite’s commercial viability and performance.
In the contemporary educational environment, where the success of millions of students depends on precise forecasting, it is essential to conduct in-depth research into the many factors that influence academic chievement. Beyond simply analyzing students’ grades, our research aims to provide a holistic picture of student achievement by examining a wide range of student demographics, academic backgrounds, and behavioral factors. We use advanced machine learning techniques, such as regression and classification, to decipher the complex patterns embedded in the data. This enables us to gain nuanced insights into the factors that predict student performance. We hope that by using these approaches, we will not only forecast academic outcomes but also identify the underlying factors that influence overall student success. In addition, our research seeks to determine the primary factors that have the greatest impact on students’ academic performance. Educators receive vital insights that enable them to personalize interventions that target both academic and non-academic aspects that affect student progress. After an in-depth investigation, we concluded that the Artificial Neural Network (ANN) and Decision Tree (DT) models were the most accurate predictors. These models achieved accuracy rates of 81% and 76%, respectively. The results of this study demonstrate that the use of sophisticated machine learning algorithms is an effective method for predicting student performance and guiding interventions specifically designed to support student achievement.
This study aims to analyze the state of the art of open science (OS) related public policies adopted in 14 countries, to propose good practices for Latin America. With a qualitative approach, all pertinent policy documents were collected through a documentary study, and 12 in-depth interviews were conducted with key public and academic actors, all experienced in designing and implementing OS public policies. The main findings include: the importance of having explicit regulatory instruments; commitment to open access to publicly funded scientific production; attention to human capital formation; adoption of FAIR principles; and the incipient promotion of inclusion. The insights gained from this study can provide valuable lessons for developing OS in Latin America.
This study aims to suppress vibrations in hybrid Rayleigh-Van der Pol-Duffing oscillators by employing negative cubic velocity feedback control. The system's behavior is analyzed using the multiple-scales method to derive solutions up to the second-order approximation. The effect of negative cubic velocity feedback is specifically examined under primary resonance conditions. The influence of various system parameters is explored numerically using MATLAB. The findings reveal a strong correlation between the approximate analytical solutions and numerical results, confirming the effectiveness of the proposed approach.
The permanent magnet brushless dc (BLDC) motor turned on and off by adjusting the magnetic fields created by the adjacent stationary coils in the appropriate direction. A BLDC motor’s six terminals—two per coil—extend from the stator and can be used to regulate the motor’s rotation. Feedback systems can be used to precisely control the torque and rotation speed of BLDC motors. There are two main types of three-phase BLDC motors: sensor and sensorless. The sensorless BLDC motor control approach is based on the Back Electromotive Force (BEMF) produced in the stator windings. The suggested CPLD- and BEMF-based Hybrid Commutation Model significantly outperforms traditional microcontroller-based, sensor-dependent BLDC motor control systems. The approach achieves the following goals by using BEMF for sensorless operation and a CPLD for high-speed logic processing: faster commutation transitions, higher efficiency, greater fault tolerance, and reduced torque ripple. Our realization is divided into three main sections. These sections are discussed in the methodology. This method uses hybrid commutation and sensorless feedback via the back electromagnetic field method. Plots of speed and motor current, the microcontroller’s triggering pulses, and the MOSFET-manipulated BLDC motor input voltage have been shown. CPLD + BEMF-based Hybrid offers the best efficiency, seamless switching, and the highest torque ripple reduction (~45%), making it ideal for high-performance applications.
This paper presents and assesses a comprehensive approach to stock price forecasting and portfolio selection that integrates advanced computational intelligence techniques with fundamental and technical analyses. Several outstanding forecasting methods are compared to identify the most accurate model. Subsequently, differential evolution is used to optimize a stock portfolio, leveraging the results of the selected forecasting method along with key technical and fundamental indicators. Experiments show that the proposed method consistently yields higher returns and better risk management than several benchmarks. Statistical validation confirms the model’s superior performance, highlighting its potential as a robust tool for optimizing investment portfolios.
Inverted pendulum systems are widely used in both education and research within control theory. There are different types of inverted pendulums, such as the Furuta pendulum, cart-pole system, reaction wheel pendulum, etc. This study investigates the dynamics of a Two-Wheeled Self-Balancing Robot, a complex electromechanical system characterized by inherentlynonlinear dynamics, unstable equilibrium points, and underactuation (i.e., more degrees of freedom than control inputs), inspired by inverted pendulum systems. The primary objective pursued in this investigation is to guide the robot along a predefined trajectory while maintaining its vertical orientation. To achieve this, we designed and implemented two controllers with an integrator in the feedback loop for a Two-Wheeled Self-Balancing Robot prototype. One controller considers the Linear Quadratic Regulator (LQR) technique, and the second is based on the H∞ approach. Due to the use of a microcontroller platform, we designed the controller in discrete time to implement it. Ramp-like and sinusoidal references were used to set the robot’s displacement targets, aiming to track these references while maintaining its vertical upright position. Simulations and practical tests were conducted using the Simulink® software within the Matlab® environment to evaluate the effectiveness of the proposed control strategies. The performances of the closed-loop system using each controller are compared. The results demonstrate the capability of the designed controllers to achieve the predefined objective, even in the presence.
