Search Results (24,776)

Knowledge distillation (KD) techniques aim to transfer knowledge from complex teacher neural networks to simpler student networks. In this study, we propose a novel knowledge distillation method called Multiloss Joint Gradient Control Knowledge Distillation (MJKD), which functions by effectively combining feature- and logit-based knowledge distillation methods with gradient control. The proposed knowledge distillation method discretely considers the gradients of the task loss (cross-entropy loss), feature distillation loss, and logit distillation loss. The experimental results suggest that logits may contain more information and should, consequently, be assigned greater weight during the gradient update process in this work. The empirical findings on the CIFAR-100 and Tiny-ImageNet datasets indicate that MJKD generally outperforms traditional knowledge distillation methods, significantly enhancing the generalization ability and classification accuracy of student networks. For instance, MJKD achieves a 63.53% accuracy on Tiny-ImageNet for the ResNet18 MobileNetV2 pair. Furthermore, we present visualizations and analyses to explore its potential working mechanisms. Full article

In recent years, unmanned aerial vehicles (UAVs) vision systems based on deep neural networks (DNNs) have made remarkable advancements, demonstrating impressive performance. However, due to the inherent characteristics of DNNs, these systems have become increasingly vulnerable to adversarial attacks. Traditional black-box attack methods typically require a large number of queries to generate adversarial samples successfully. In this paper, we propose a novel adversarial attack technique designed to achieve efficient black-box attacks with a minimal number of queries. We define a perturbation generator that first decomposes the image into four frequency bands using wavelet decomposition and then searches for adversarial perturbations across these bands by minimizing a weighted loss function on a set of fixed surrogate models. For the target victim model, the perturbation images generated by the perturbation generator are used to query and update the weights in the loss function, as well as the weights for different frequency bands. Experimental results show that, compared to state-of-the-art methods on various image classifiers trained on ImageNet (such as VGG-19, DenseNet-121, and ResNext-50), our method achieves a success rate over 98% for targeted attacks and nearly a 100% success rate for non-targeted attacks with only 1–2 queries per image. Full article

This work investigated the structure, microstructure, and ferroelectric and dielectric behavior of (Pb_0.97Ba_0.03)(Zr_0.98Ti_0.02)_1−xSn_xO₃ (PBZT_xSn) solid solution with variable tin content in the range x = 0.00–0.08. Synthesis was carried out using the powder calcination method, and sintering was carried out using the hot-pressing method. For all the PBZT_xSn samples at room temperature, X-ray diffractograms confirmed the presence of an orthorhombic (OR) crystal structure with space group Pnnm, and the microstructure is characterized by densely packed and properly shaped grains with an average size of 1.36 µm to 1.73 µm. At room temperature, PBZT_xSn materials have low permittivity values ε′ ranging from 265 to 275, whereas, at the ferroelectric–paraelectric phase transition temperature (RE–C), the permittivity is high (from 8923 to 12,141). The increase in the tin dopant in PBZT_xSn lowers permittivity and dielectric loss and changes the scope of occurrence of phase transitions. The occurring dispersion of the dielectric constant and dielectric loss at low frequencies, related to the Maxwell–Wagner behavior, decreases with increasing tin content in the composition of PBZT_xSn. Temperature studies of the dielectric and ferroelectric properties revealed anomalies related to the phase transitions occurring in the PBZT_xSn material. With increasing temperature in PBZT_xSn, phase transitions occur from orthorhombic (OR) to rhombohedral (RE) and cubic (C). The cooling cycle shifts the temperatures of the phase transitions towards lower temperatures. The test results were confirmed by XRD Rietveld analysis at different temperatures. The beneficial dielectric and ferroelectric properties suggest that the PBZT_xSn materials are suitable for micromechatronic applications as pulse capacitors or actuator elements. Full article

Background: The global healthcare system faces challenges in diagnosing and managing lung and colon cancers, which are significant health burdens. Traditional diagnostic methods are inefficient and prone to errors, while data privacy and security concerns persist. Objective: This study aims to develop a secure and transparent framework for remote consultation and classification of lung and colon cancer, leveraging blockchain technology and Microsoft Azure cloud services. Dataset and Features: The framework utilizes the LC25000 dataset, containing 25,000 histopathological images, for training and evaluating advanced machine learning models. Key features include secure data upload, anonymization, encryption, and controlled access via blockchain and Azure services. Methods: The proposed framework integrates Microsoft Azure’s cloud services with a permissioned blockchain network. Patients upload CT scans through a mobile app, which are then preprocessed, anonymized, and stored securely in Azure Blob Storage. Blockchain smart contracts manage data access, ensuring only authorized specialists can retrieve and analyze the scans. Azure Machine Learning is used to train and deploy state-of-the-art machine learning models for cancer classification. Evaluation Metrics: The framework’s performance is evaluated using metrics such as accuracy, precision, recall, and F1-score, demonstrating the effectiveness of the integrated approach in enhancing diagnostic accuracy and data security. Results: The proposed framework achieves an impressive accuracy of 100% for lung and colon cancer classification using DenseNet, ResNet50, and MobileNet models with different split ratios (70–30, 80–20, 90–10). The F1-score and k-fold cross-validation accuracy (5-fold and 10-fold) also demonstrate exceptional performance, with values exceeding 99.9%. Real-time notifications and secure remote consultations enhance the efficiency and transparency of the diagnostic process, contributing to better patient outcomes and streamlined cancer care management. Full article

Background: The goal of this study was to evaluate whether the medical recommendation of Angocin^®Anti-Infekt N (heretofore referenced as Angocin^®) on the day of diagnosis of acute bronchitis is negatively associated with the recurrence of acute bronchitis diagnosis, antibiotic prescriptions, incidence of chronic bronchitis, and duration of sick leave. Methods: This study included patients in general practices in Germany with a first documented diagnosis of acute bronchitis between 2005 and 2022 (index date) and a prescription of Angocin^®, thyme products, essential oils, mucolytics or antibiotics on the index date. The association between Angocin^® prescription and the risks of a relapse of acute bronchitis, development of chronic bronchitis, or subsequent antibiotic prescription were evaluated using Cox regression models. Univariable conditional logistic regression models were used to investigate the association between Angocin^® prescription and duration of sick leave. Results: After a 1:5 propensity score matching, 598 Angocin^® patients and 2990 patients in each of the four comparison cohorts were available for analysis. Angocin^® prescription was associated with significantly lower incidence of a renewed confirmed diagnosis of acute bronchitis as compared to essential oils (Hazard ratio (HR): 0.61; 95% Confidence Interval (CI): 0.46–0.80), thyme products (HR: 0.70; 95% CI: 0.53–0.91), mucolytics (HR: 0.65; 95% CI: 0.49–0.85) or antibiotics (HR: 0.64; 95% CI: 0.49–0.84). Also, there were significantly lower incidences of subsequent re-prescriptions of antibiotics when compared to mucolytics (HR: 0.73; 95% CI: 0.53–0.99) or antibiotics (HR: 0.53; 95% CI: 0.39–0.72) and a significantly lower risk of chronic bronchitis as compared to essential oils (HR: 0.60; 95% CI: 0.46–0.78), thyme products (HR: 0.53; 95% CI: 0.41–0.69), mucolytics (HR: 0.49; 95% CI: 0.38–0.63) or antibiotics (HR: 0.59; 95% CI: 0.45–0.76). Conclusions: Considering the limitations of the study, the results shed light on the sustaining effectiveness of Angocin^® prescription in the management of acute bronchitis and the associated outcomes when compared to several other treatments commonly used for this condition. Full article

This contribution reconstructs the reflection on witches of Johannes Trithemius (1462–1516), a German Benedictine who took up the pen on several occasions to declare against the spread of witchcraft and the need to solve this problem. The method adopted is to understand Trithemius’ thought from the analysis of his own works, specifically the Antipalus maleficiorum (1505–1508), the Liber octo quaestionum (1515) and what can be known of the De daemonibus (1507–1514). What will emerge will be an articulate reflection, which re-proposes the doctrine of the Malleus maleficarum (1486) enriched with original elements often drawn from popular superstitions. Thus, Trithemius proposes artifices to be immune from witches (e.g., apotropaic amulets) and provides specific indications on how to cure evil spells (exorcism), extending the dissertation to broader issues, such as the gender question, the relationship between witches and children (e.g., sacrifices, proselytes, victims) and developments in exorcism practice. Full article

Lung cancer is one of the major cancer types and poses challenges in its treatment, including lack of specificity and harm to healthy cells. Nanoparticle-based drug delivery systems (NDDSs) show promise in overcoming these challenges. While conventional NDDSs have drawbacks, such as immune response and capture by the reticuloendothelial system (RES), extracellular vesicles (EVs) present a potential solution. EVs, which are naturally released from cells, can evade the RES without surface modification and with minimal toxicity to healthy cells. This makes them a promising candidate for developing a lung-cancer-targeting drug delivery system. EVs isolated from vascular endothelial cells, such as human umbilical endothelial-cell-derived EVs (HUVEC-EVs), have shown anti-angiogenic activity in a lung cancer mouse model; therefore, in this study, HUVEC-EVs were chosen as a carrier for drug delivery. To achieve lung-cancer-specific targeting, HUVEC-EVs were engineered to be decorated with GE11 peptides (GE11-HUVEC-EVs) via a postinsertional technique to target the epidermal growth factor receptor (EGFR) that is overexpressed on the surface of lung cancer cells. The GE11-HUVEC-EVs were loaded with vinorelbine (GE11-HUVEC-EVs-Vin), and then characterized and evaluated in in vitro and in vivo lung cancer models. Further, we examined the binding affinity of ABCB1, encoding P-glycoprotein, which plays a crucial role in chemoresistance via the efflux of the drug. Our results indicate that GE11-HUVEC-EVs-Vin effectively showed tumoricidal effects against cell and mouse models of lung cancer. Full article

The reading literacy crisis, among learners, in countries throughout the world and in South Africa seems to be reaching pandemic levels. Hence, the quality of teaching and the preparation that pre-service teachers receive at initial teacher education institutions is under the spotlight. A proactive action research design is used to integrate mixed reality simulation into reading literacy modules. Our data collection methods included professional conversations, WhatsApp voice notes and video calls, reflective journal entries and reflections on observing video recordings of lesson segments in the MRS environment. The data was analyzed using content analysis. The main themes emanating from the data included: lack of focus on high leverage teaching practices, limited use of pedagogies of enactment, add-on to existing content, experimentation, perceptions, planning and preparation, content-method integration, pedagogies of enactment, assessment, resources and feedback. Grounded in a Community of Practice framework, we narrate our experiences of re-imagining mixed reality simulation as a core component of initial teacher education programs. The authors conclude by sharing insights and recommendations for policymakers, faculty leaders, and curriculum designers, contributing to informed decisions regarding integrating and potentially upscaling mixed reality simulation within reading literacy modules in initial teacher education programs. Full article

This paper examines Muslim anarchists in Turkey who developed an Islam-based anarchist theory opposing private property, the state, capitalism, and all forms of authority. By analyzing their online periodical itaatsiz (disobedient), published since 2013, and earlier works by Muslim anarchist writers, this study explores their perspectives on the West, Islam, the Qur’an, and Sufism. Muslim anarchists stand out for their opposition to the hegemony of Enlightenment-based, anti-theist, and positivist thought in anarchist movements in Turkey and for their encouragement to re-examine concepts such as authority, private property, capitalism, and the state within the framework of Islam-based political thought. Studying how Muslim anarchists construct a social movement in today’s Turkey is essential to understanding Islam-based conceptualizations of politics in Turkey and unpacking the relationship between Islam and anarchism. Full article

Autophagosome (AP)–lysosome/vacuole fusion is one of the hallmarks of macroautophagy. Membrane features and changes during the fusion process have mostly been described using two-dimensional (2D) models with one AP and one lysosome/vacuole. The outer membrane (OM) of a closed mature AP has been suggested to fuse with the lysosomal/vacuolar membrane. However, the descriptions in some studies for fusion-related issues are questionable or incomplete. The correct membrane features of APs and lysosomes/vacuoles are the prerequisite for describing the fusion process. We searched the literature for representative membrane features of AP-related structures based on electron microscopy (EM) graphs of both animal and yeast cells and re-evaluated the findings. We also summarized the main 2D models describing the membrane changes during AP–lysosome/vacuole fusion in the literature. We used three-dimensional (3D) models to characterize the known and unknown membrane changes during and after fusion of the most plausible 2D models. The actual situation is more complex, since multiple lysosomes may fuse with the same AP in mammalian cells, multiple APs may fuse with the same vacuole in yeast cells, and in some mutant cells, phagophores (unclosed APs) fuse with lysosomes/vacuoles. This review discusses the membrane features and highly dynamic changes during AP (phagophore)–lysosome/vacuole fusion. The resulting information will improve the understanding of AP–lysosome/vacuole fusion and direct the future research on AP–lysosome/vacuole fusion and regeneration. Full article

Background/Objectives: Proton pump inhibitors (PPIs) are amongst the most commonly prescribed classes of medication. However, inappropriate PPI use can lead to several adverse drug reactions (ADRs). Limited data exist on factors contributing to the risk of ADRs associated with PPI prescribing patterns in the Eastern Region of Saudi Arabia. This retrospective, cross-sectional study aimed to assess the prevalence and the pattern of PPI use and to identify factors contributing to the risk of ADRs. Methods: Data were collected from electronic medical records of patients at Al-Qateef Central Hospital from January 2020 to December 2021. The inclusion criteria included patients aged ≥40 years attending an outpatient medical care clinic. PPI prescribing patterns were categorized based on their dosage intensity into low-dose, medium-dose (MD), and high-dose (HD) categories. Binary and multinominal logistic regression models were used to determine the relationship between PPI prescribing patterns and use, categorized by MD or HD, and patient characteristics, adjusted for significant covariates. Results are presented as adjusted odds ratio (OR) with corresponding 95% confidence intervals (95% CI). Results: The study included 41,084 patients. The prevalence of PPI prescribing was 31%. PPI users were more frequently found to be females than males (52% vs. 50%, p = 0.013); they were also likely to be prescribed more medications (7 vs. 6, p < 0.001), but less likely to have gastritis-related diseases (34% vs. 32%, p < 0.001) compared to non-users. PPI HD users were more likely male (56% vs. 43%, p < 0.001), older (53 vs. 52 years, p < 0.001), and prescribed more medications (11.8 vs. 2.8, p < 0.001) compared to MD users. PPI usage was associated with concurrent use of antiplatelet drugs (OR = 1.08, 95% CI 1.01–1.15). An increasing number of prescribed medications was associated with HD usage (OR = 1.13, 95% CI 1.12–1.14), but negatively associated with MD usage (OR = 0.7 95% CI 0.69–0.71). Female gender was negatively associated with HD usage (OR = 0.85, 95% CI 0.79–0.91). Conclusions: Our findings indicate that 31% of the included cohort were prescribed PPI. Inappropriate PPI prescribing related to the drug’s omission is a concern as PPI non-users presented with valid indications such as gastritis. Male gender and increasing NPM were the common factors contributing to increased risk of PPI ADR. This study points to the importance of re-evaluating PPI use to ensure effective therapy with minimum risks of ADR. Full article

As the main area for photosynthesis in trees, the canopy absorbs a large amount of carbon dioxide and plays an irreplaceable role in regulating the carbon cycle in the atmosphere and mitigating climate change. Therefore, monitoring the growth of the canopy is crucial. However, traditional field investigation methods are often limited by time-consuming and labor-intensive methods, as well as limitations in coverage, which may result in incomplete and inaccurate assessments. In response to the challenges encountered in the application of tree crown segmentation algorithms, such as adhesion between individual tree crowns and insufficient generalization ability of the algorithm, this study proposes an improved algorithm based on Mask R-CNN (Mask Region-based Convolutional Neural Network), which identifies irregular edges of tree crowns in RGB images obtained from drones. Firstly, it optimizes the backbone network by improving it to ResNeXt and embedding the SENet (Squeeze-and-Excitation Networks) module to enhance the model’s feature extraction capability. Secondly, the BiFPN-CBAM module is introduced to enable the model to learn and utilize features more effectively. Finally, it optimizes the mask loss function to the Boundary-Dice loss function to further improve the tree crown segmentation effect. In this study, TCSNet also incorporated the concept of panoptic segmentation, achieving the coherent and consistent segmentation of tree crowns throughout the entire scene through fine tree crown boundary recognition and integration. TCSNet was tested on three datasets with different geographical environments and forest types, namely artificial forests, natural forests, and urban forests, with artificial forests performing the best. Compared with the original algorithm, on the artificial forest dataset, the precision increased by 6.6%, the recall rate increased by 1.8%, and the F₁-score increased by 4.2%, highlighting its potential and robustness in tree detection and segmentation. Full article

The growing integration of renewable energy sources, such as photovoltaic and wind systems, into energy grids has underscored the need for reliable control mechanisms to mitigate the inherent intermittency of these sources. According to the Brazilian grid operator (ONS), there have been cascading disconnections in renewable energy distributed systems (REDs) in recent years, highlighting the need for robust control models. This article addresses this issue by presenting the validation of an active power ramp rate control (PRRC) function for a PV plant coupled with a Battery Energy Storage System (BESS) using WECC generic models. The proposed model underwent rigorous validation over an extended analysis period, demonstrating good accuracy using the Root Mean Squared Error (RMSE) and an R-squared (R²) metrics for the active power injected at the Point of Connection (POI), PV active power, and BESS State of Charge (SOC), providing valuable insights for medium and long-term analyses. The ramp rate control module was implemented within the plant power controller (PPC), leveraging second-generation Renewable Energy Systems (RES) models developed by the Western Electricity Coordination Council (WECC) as a foundational framework. We conducted simulations using the Anatem software, comparing the results with real-world data collected at 100 ms to 1000 ms intervals from a PV plant equipped with a BESS in Brazil. The proposed model underwent rigorous validation over an extended analysis period, with the presented results based on two days of measurements. The positive sequence model used to represent this control demonstrated good accuracy, as confirmed by metrics such as the Root Mean Squared Error (RMSE) and R-squared (R²). Furthermore, the article underscores the critical role of accurately accounting for the power sampling rate when calculating the ramp rate. Full article

The growing interest in data privacy and anonymization presents challenges, as traditional methods such as ordinal discretization often result in information loss by coarsening metric data. Current research suggests that modeling the latent distributions of ordinal classes can reduce the effectiveness of anonymization and increase traceability. In fact, combining probability distributions with a small training sample can effectively infer true metric values from discrete information, depending on the model and data complexity. Our method uses metric values and ordinal classes to model latent normal distributions for each discrete class. This approach, applied with both linear and Bayesian linear regression, aims to enhance supervised learning models. Evaluated with synthetic datasets and real-world datasets from UCI and Kaggle, our method shows improved mean point estimation and narrower prediction intervals compared to the baseline. With 5–10% training data randomly split from each dataset population, it achieves an average 10% reduction in MSE and a ~5–10% increase in R² on out-of-sample test data overall. Full article

As malware samples grow in complexity and employ advanced evasion techniques, traditional detection methods are insufficient for accurately classifying large volumes of sophisticated malware variants. To address this issue, image-based malware classification techniques leveraging machine learning algorithms have been developed as a more optimal solution to this challenge. However, accurately classifying content distribution-based features with unique pixel intensities from grayscale images remains a challenge. This paper proposes an enhanced image-based malware classification system using convolutional neural networks (CNNs) using ResNet-152 and vision transformer (ViT). The two architectures are then compared to determine their classification abilities. A total of 6137 benign files and 9861 malicious executables are converted from text files to unsigned integers and then to images. The ViT examined unsigned integers as pixel values, while ResNet-152 converted the pixel values into floating points for classification. The result of the experiments demonstrates a high-performance accuracy of 99.62% with effective hyperparameters of 10-fold cross-validation. The findings indicate that the proposed model is capable of being implemented in dynamic and complex malware environments, achieving a practical computational efficiency of 47.2 s for the identification and classification of new malware samples. Full article