Search Results (3,073)

Imbalanced learning occurs when rare but critical events are missed because classifiers are trained primarily on majority-class samples. This paper introduces WISEST, a locality-aware weighted-interpolation algorithm that generates synthetic minority samples within a controlled threshold near class boundaries. Benchmarked on more than a hundred real-world imbalanced datasets, such as KEEL, with different imbalance ratios, noise levels, geometries, and other security and IoT sets (IoT-23 and BoT–IoT), WISEST consistently improved minority detection in at least one of the metrics on about half of those datasets, achieving up to a 25% relative recall increase and up to an 18% increase in F1 compared to the original training and other approaches. However, in most cases, WISEST’s trade-off gains are in accuracy and precision depending on the dataset and classifier. These results indicate that WISEST is a practical and robust option when minority support and borderline structure permit safe synthesis, although no single sampler uniformly outperforms others across all datasets. Full article

Accurate segmentation and classification of kidney pathologies from medical images remain a major challenge in computer-aided diagnosis due to complex morphological variations, small lesion sizes, and severe class imbalance. This study introduces DiagNeXt, a novel two-stage deep learning framework designed to overcome these challenges through an integrated use of attention-enhanced ConvNeXt architectures for both segmentation and classification. In the first stage, DiagNeXt-Seg employs a U-Net-based design incorporating Enhanced Convolutional Blocks (ECBs) with spatial attention gates and Atrous Spatial Pyramid Pooling (ASPP) to achieve precise multi-class kidney segmentation. In the second stage, DiagNeXt-Cls utilizes the segmented regions of interest (ROIs) for pathology classification through a hierarchical multi-resolution strategy enhanced by Context-Aware Feature Fusion (CAFF) and Evidential Deep Learning (EDL) for uncertainty estimation. The main contributions of this work include: (1) enhanced ConvNeXt blocks with large-kernel depthwise convolutions optimized for 3D medical imaging, (2) a boundary-aware compound loss combining Dice, cross-entropy, focal, and distance transform terms to improve segmentation precision, (3) attention-guided skip connections preserving fine-grained spatial details, (4) hierarchical multi-scale feature modeling for robust pathology recognition, and (5) a confidence-modulated classification approach integrating segmentation quality metrics for reliable decision-making. Extensive experiments on a large kidney CT dataset comprising 3847 patients demonstrate that DiagNeXt achieves 98.9% classification accuracy, outperforming state-of-the-art approaches by 6.8%. The framework attains near-perfect AUC scores across all pathology classes (Normal: 1.000, Tumor: 1.000, Cyst: 0.999, Stone: 0.994) while offering clinically interpretable uncertainty maps and attention visualizations. The superior diagnostic accuracy, computational efficiency (6.2× faster inference), and interpretability of DiagNeXt make it a strong candidate for real-world integration into clinical kidney disease diagnosis and treatment planning systems. Full article

This study explores the impact of synthetic data, both physically based and generatively created, on deep learning analytics for earth observation (EO), focusing on the detection of photovoltaic panels. A YOLOv8 object detection model was trained using a publicly available, multi-resolution very high resolution (VHR) EO dataset (0.8 m, 0.3 m, and 0.1 m), comprising 3716 images from various locations in Jiangsu Province, China. Three benchmarks were established using only real EO data. Subsequent experiments evaluated how the inclusion of synthetic data, in varying types and quantities, influenced the model’s ability to detect photovoltaic panels in VHR imagery. Physically based synthetic images were generated using the Unity engine, which allowed the generation of a wide range of realistic scenes by varying scene parameters automatically. This approach produced not only realistic RGB images but also semantic segmentation maps and pixel-accurate masks identifying photovoltaic panel locations. Generative synthetic data were created using diffusion-based models (DALL·E 3 and Stable Diffusion XL), guided by prompts to simulate satellite-like imagery containing solar panels. All synthetic images were manually reviewed, and corresponding annotations were ensured to be consistent with the real dataset. Integrating synthetic with real data generally improved model performance, with the best results achieved when both data types were combined. Performance gains were dependent on data distribution and volume, with the most significant improvements observed when synthetic data were used to meet the YOLOv8-recommended minimum of 1500 images per class. In this setting, combining real data with both physically based and generative synthetic data yielded improvements of 1.7% in precision, 3.9% in recall, 2.3% in mAP@50, and 3.3% in mAP@95 compared to training with real data alone. The study also emphasizes the importance of carefully managing the inclusion of synthetic data in training and validation phases to avoid overfitting to synthetic features, with the goal of enhancing generalization to real-world data. Additionally, a pre-training experiment using only synthetic data, followed by fine-tuning with real images, demonstrated improved early-stage training performance, particularly during the first five epochs, highlighting potential benefits in computationally constrained environments. Full article

The abstract has been submitted for presentation to the AANS 2026 meeting being held in San Antonio, TX, USA. Introduction: Oligodendrogliomas are an uncommon subset of gliomas that are molecularly defined by 1p/19q codeletion in the setting of an isocitrate dehydrogenase (IDH) 1/2 mutation. Standard-of-care management involves maximal safe resection followed by adjuvant chemoradiation with procarbazine, lomustine, and vincristine (PCV). Although PCV confers a durable survival advantage, treatment-limiting toxicity is common and often necessitates discontinuation. IDH inhibitors such as vorasidenib have demonstrated promising efficacy and more favorable tolerability profiles, but a paucity of comparative data across therapeutic classes limits optimal treatment decision-making. Methods: A systematic search was conducted through to 7 March 2025 in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR). Eligible studies included adult patients (≥18 years) with IDH-mutant, 1p/19q-codeleted oligodendrogliomas treated with PCV chemotherapy or IDH inhibitors and with a minimum follow-up of 12 months. Outcomes of interest included overall survival (OS), progression-free survival (PFS), and grade ≥ 3 adverse events (AEs) that led to treatment discontinuation. Results: Twenty-eight studies met the inclusion criteria, with a total of 406 patients. All 406 patients carried a confirmed diagnosis of oligodendroglioma. For mixed-histology cohorts, only oligodendroglioma-specific data were extracted and analyzed. Among PCV cohorts, median PFS ranged from 24.3 months to 8.4 years and median OS was reported up to 14.7 years in long-term follow-up from RTOG 9402 and EORTC 26951. Grade ≥ 3 AEs resulted in treatment discontinuation in 65–70% of patients, primarily due to hematologic or neurologic events. In comparison, vorasidenib achieved a median PFS of 27.7 months in the phase III INDIGO trial (HR 0.39; 95% CI 0.27–0.56; p < 0.001), with median OS not yet reached at 14.2 months of follow-up. Grade ≥ 3 AEs occurred in 22.8% of patients and led to treatment discontinuation in only 1–3%, primarily due to asymptomatic transaminitis. Early real-world data from expanded-access programs similarly support these tolerability findings. Conclusions: While PCV chemotherapy remains the standard-of-care systemic therapy for oligodendroglioma supported by mature survival data, IDH inhibitors represent a mechanistically targeted alternative with encouraging early-phase outcomes and a significantly improved safety profile. Direct comparison across these regimens is constrained by differences in study design and limited long-term OS data for IDH inhibitors. Prospective head-to-head trials are essential for defining the optimal therapeutic sequence in this evolving treatment landscape. In the interim, we provide a recommend approach for current use. Full article

Automated pollen recognition is a foundational tool for diverse scientific domains, including paleoclimatology, biodiversity monitoring, and agricultural science. However, conventional methods create a critical data bottleneck, limiting the temporal and spatial resolution of ecological analysis. Existing deep learning models often fail to achieve the requisite localization accuracy for microscopic pollen grains, which are characterized by their minute size, indistinct edges, and complex backgrounds. To overcome this, we introduce HieraEdgeNet, a novel object detection framework. The core principle of our architecture is to explicitly extract and hierarchically fuse multi-scale edge information with deep semantic features. This synergistic approach, combined with a computationally efficient large-kernel operator for fine-grained feature refinement, significantly enhances the model’s ability to perceive and precisely delineate object boundaries. On a large-scale dataset comprising 44,471 annotated microscopic images containing 342,706 pollen grains from 120 classes, HieraEdgeNet achieves a mean Average Precision of 0.9501 (mAP@0.5) and 0.8444 (mAP@0.5:0.95), substantially outperforming state-of-the-art models such as YOLOv12n and the Transformer-based RT-DETR family in terms of the accuracy–efficiency trade-off. This work provides a powerful computational tool for generating the high-throughput, high-fidelity data essential for modern ecological research, including tracking phenological shifts, assessing plant biodiversity, and reconstructing paleoenvironments. At the same time, we acknowledge that the current two-dimensional design cannot directly exploit volumetric Z-stack microscopy and that strong domain shifts between training data and real-world deployments may still degrade performance, which we identify as key directions for future work. By also enabling applications in precision agriculture, HieraEdgeNet contributes broadly to advancing ecosystem monitoring and sustainable food security. Full article

As technology advances, developers continually create innovative solutions to enhance smartphone security. However, the rapid spread of Android malware poses significant threats to devices and sensitive data. The Android Operating System (OS)’s open-source nature and Software Development Kit (SDK) availability mainly contribute to this alarming growth. Conventional malware detection methods, such as signature-based, static, and dynamic analysis, face challenges in detecting obfuscated techniques, including encryption, packing, and compression, in malware. Although developers have created several visualization techniques for malware detection using deep learning (DL), they often fail to accurately identify the critical malicious features of malware. This research introduces MalVis, a unified visualization framework that integrates entropy and N-gram analysis to emphasize meaningful structural and anomalous operational patterns within the malware bytecode. By addressing significant limitations of existing visualization methods, such as insufficient feature representation, limited interpretability, small dataset sizes, and restricted data access, MalVis delivers enhanced detection capabilities, particularly for obfuscated and previously unseen (zero-day) malware. The framework leverages the MalVis dataset introduced in this work, a publicly available large-scale dataset comprising more than 1.3 million visual representations in nine malware classes and one benign class. A comprehensive comparative evaluation was performed against existing state-of-the-art visualization techniques using leading convolutional neural network (CNN) architectures, MobileNet-V2, DenseNet201, ResNet50, VGG16, and Inception-V3. To further boost classification performance and mitigate overfitting, the outputs of these models were combined using eight distinct ensemble strategies. To address the issue of imbalanced class distribution in the multiclass dataset, we employed an undersampling technique to ensure balanced learning across all types of malware. MalVis achieved superior results, with 95% accuracy, 90% F1-score, 92% precision, 89% recall, 87% Matthews Correlation Coefficient (MCC), and 98% Receiver Operating Characteristic Area Under Curve (ROC-AUC). These findings highlight the effectiveness of MalVis in providing interpretable and accurate representation features for malware detection and classification, making it valuable for research and real-world security applications. Full article

Artificial Intelligence (AI) plays an important role in driving innovation in smart healthcare by providing accurate, scalable, and privacy-preserving diagnostic options. Pneumonia is still a major global health issue, and early detection is key to improving patient outcomes. This study proposes a federated deep learning (FL) approach for automatic pneumonia detection using chest X-ray images, considering both diagnostic efficacy and data privacy. Two models were developed and tested: a custom-developed convolutional neural network and a VGG16 transfer learning architecture. The framework evaluates diagnostic efficacy in both centralized and federated scenarios, taking into account heterogeneous client distributions and class imbalance. F1-score and accuracy values for the federated models indicate competitive levels, with F1-scores greater than 0.90 for pneumonia, being robust even when the data is not independent and identically distributed. Results confirm that FL could be tested as a privacy-preserving way to manage medical imaging and intelligence across distributed healthcare. This study provides a potential proof of concept of how to incorporate federated AI into smart healthcare and gives direction toward clinically tested and real-world applications. Full article

The rapid growth of artificial intelligence (AI) has enabled efficient crop disease detection even in data-scarce agricultural settings. This study proposes AgriFewNet, a few-shot learning framework designed to improve classification accuracy using RGB imagery captured from publicly available datasets. The objective is to enable fast model adaptation to new disease classes using minimal labeled samples while maintaining high reliability in real-world conditions. AgriFewNet employs a hierarchical attention-enhanced ResNet-18 backbone incorporating dual spatial and channel attention to extract discriminative RGB features. A Model-Agnostic Meta-Learning (MAML) approach facilitates quick generalization to previously unexplored illness categories, while a prototype-based classifier guarantees compact representation learning. Using only RGB images, experiments on the PlantVillage and New PlantVillage datasets produced accuracies of 87.3% (1-shot), 94.8% (5-shot), and 97.1% (10-shot), surpassing leading few-shot baselines by as much as 7.9%. The findings show that AgriFewNet offers a resource-efficient and scalable method for intelligent crop monitoring, enhancing food security and precision agriculture. Full article

Humidity-regulating materials (HRMs) represent a promising class of passive, energy-efficient materials capable of autonomously modulating indoor environmental conditions, particularly in hot and humid regions where conventional HVAC systems account for up to 50% of building energy consumption. While prior reviews have focused on material classification and performance metrics, a systematic synthesis of performance optimization strategies and quantitative application outcomes remains lacking. This review addresses this gap by consolidating advances in HRM enhancement through material compounding, physical modification, and chemical functionalization, resulting in performance improvements such as a 70% increase in moisture absorption with 3% fiber addition, a 1.2-fold enhancement in adsorption capacity via pore optimization, and up to 50% energy savings in building applications. Furthermore, the integration of HRMs into radiant cooling systems elevates the dew point temperature difference by 181%, effectively mitigating condensation risks. Simulation tools—ranging from 1D to 3D multiphysics models—have advanced predictive accuracy for coupled heat and moisture transfer, supporting optimized material design and system integration. By systematically summarizing performance metrics, enhancement mechanisms, and real-world applications, this work provides a quantitative and structured reference for the development and deployment of next-generation HRMs in sustainable building systems. Full article

This paper interrogates two important but obscured admission policy developments at leading American universities in the post-World War II era. First, we critically examine the University of California’s “special admissions,” later formalized as the “Admission by Exception” policy adopted at two flagship campuses (Berkeley and UCLA) to open opportunities for veterans returning from the War under the GI Bill. The scale of this Admission by Exception policy was orders of magnitude larger than any comparable admissions policy in recent decades, including both the eras with and without legally permissible affirmative action. Second, we excavate archival evidence from the immediate aftermath of the 1954 Brown v. Board of Education decision, where leaders at the flagship University of Texas at Austin campus hastily adopted a new standardized exam requirement because their enrollment modeling indicated this was the most efficient way to not face further losses in federal court while excluding the largest number of African Americans (and thereby resisting Brown) and maintaining the same overall size of the freshmen class. These two post-war admission policy changes, one arising in de facto segregated California and the other in de jure segregated Texas, operated as racialized institutional mechanisms analogous to “redlining” racially restrictive housing policies that are a more familiar feature of the post-War era. We draw on historical data about earnings and wealth accumulation of the overwhelmingly white graduates of UC and UT in the 1950s–70s and connect these findings to the theoretical frameworks of Cheryl Harris’s “whiteness as property” and George Lipsitz’s racialized state investment. We show how these admission policies contributed to the intergenerational transfer of advantage. We then turn to the contemporary admissions landscape at highly selective American universities after the Supreme Court’s SFFA v. Harvard ruling. We link current trends at some elite institutions toward a return to standardized testing requirements, maintaining considerations of athletic ability mostly in “country club” sports as manifestations of bias in university admissions, which tend to favor white applicants. The paper connects historical racialization of admissions to ongoing inequities in access and outcomes, showing how both historical and contemporary admissions policies reward inherited forms of cultural capital aligned with whiteness. Full article

Background: Interleukin-1 (IL-1) inhibitors are approved for the treatment of various inflammatory diseases associated with immune system abnormalities. However, large-scale real-world studies to assess their security are still limited. Therefore, a pharmacovigilance study was conducted based on the data from the U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS). Methods: Adverse events (AEs) linked to IL-1 inhibitors were analyzed using the FAERS database from Q1 2004 to Q3 2024. Risk signals were identified through disproportionality analysis algorithms, including reporting odds ratio (ROR), proportional reporting ratio (PRR), Bayesian confidence propagation neural network (BCPNN), and multi-item gamma Poisson shrinker (MGPS). Results: Among 17,670 AE reports where an IL-1 inhibitor was the “primary suspected” drug, 27 significant system organ classes (SOCs) were identified. Notable signals included infections and infestations (ROR: 2.31, 95% CI: 2.25–2.37) and congenital, familial, and genetic disorders (ROR: 2.26, 95% CI: 2.05–2.48). At the preferred term (PT) level, 263 significant AE signals were detected, such as pyrexia (ROR: 5.27, 95% CI: 5.03–5.53), nasopharyngitis (ROR: 2.31, 95% CI: 2.10–2.54), and injection site erythema (ROR: 6.09, 95% CI: 5.67–6.55). Importantly, we also identified less common or previously unreported AEs, including cardiac disorders (e.g., postural orthostatic tachycardia syndrome with anakinra; pulmonary valve incompetence with rilonacept) and endocrine disorders (e.g., secondary adrenocortical insufficiency with canakinumab). Furthermore, 36.33% of cases emerged after more than 360 days of treatment with IL-1 inhibitors. Conclusions: This study revealed real-world safety data on IL-1 inhibitors, providing important insights to enhance the clinical use of IL-1 inhibitors and minimize potential AEs. Full article

Time-series anomaly detection is imperative for ensuring reliability and safety in intelligent manufacturing systems. However, real-world environments typically provide only normal operating data and exhibit significant periodicity, noise, imbalance, and domain variability. The present study proposes CL-OCC, a contrastive learning-based one-class framework that integrates seasonal-trend decomposition using loess (STL) for structure-preserving temporal augmentation, a cosine-regularized soft boundary for compact normal-region formation, and variance-preserving regularization to prevent latent collapse. A convolutional recurrent encoder is first pretrained via an autoencoder objective and subsequently optimized through a unified loss that balances contrastive invariance, soft-boundary constraint, and variance dispersion. Experiments on semiconductor equipment data and three public benchmarks demonstrate that CL-OCC provides competitive or superior performance relative to reconstruction-, prediction-, and contrastive-based baselines. CL-OCC exhibits smoother anomaly trajectories, earlier detection of gradual drifts, and strong robustness to noise, window-length variation, and extreme class imbalance. A study of the effects of ablation and interaction on the stability of representations indicates that STL-based augmentation, boundary shaping, and variance regularization contribute complementary benefits to this stability. While the qualitative results indicate limited sensitivity to extremely short impulsive disturbances, the proposed framework delivers a generalizable and stable solution for unsupervised industrial monitoring, with promising potential for extension to multi-resolution analysis and online prognostics and health management (PHM) applications. Full article

Food safety and quality are paramount global concerns, with the complexities of the modern supply chain demanding advanced technologies for monitoring, preservation, and decontamination. Conventional methods often fall short due to limitations in speed, sensitivity, cost, and functionality. Metal–organic frameworks (MOFs), a class of crystalline porous materials, have emerged as a highly universal platform to address these challenges, owing to their unprecedented structural tunability, ultrahigh surface areas, and tailorable chemical functionalities. This comprehensive review details the state-of-the-art applications of multifunctional MOFs across the entire spectrum of food safety and quality enhancement. First, the review details the application of MOFs in advanced food analysis, covering their transformative roles as sorbents in sample preparation (e.g., solid-phase extraction and microextraction), as novel stationary phases in chromatography, and as the core components of highly sensitive sensing platforms, including luminescent, colorimetric, electrochemical, and SERS-based sensors for contaminant detection. Subsequently, the role of MOFs in food preservation and packaging is explored, highlighting their use in active packaging systems for ethylene scavenging and controlled antimicrobial release, in intelligent packaging for visual spoilage indication, and as functional fillers for enhancing the barrier properties of packaging materials. Furthermore, the review examines the direct application of MOFs in food processing for the selective adsorptive removal of contaminants from complex food matrices (such as oils and beverages) and as robust, recyclable heterogeneous catalysts. Finally, a critical discussion is presented on the significant challenges that impede widespread adoption. These include concerns regarding biocompatibility and toxicology, issues of long-term stability in complex food matrices, and the hurdles of achieving cost-effective, scalable synthesis. This review not only summarizes recent progress but also provides a forward-looking perspective on the interdisciplinary efforts required to translate these promising nanomaterials from laboratory research into practical, real-world solutions for a safer and higher-quality global food supply. Full article

As cyber threats such as denial-of-service (DoS) attacks continue to rise, network intrusion detection systems (NIDS) have become essential components of cybersecurity defense. Although machine learning is widely applied to network intrusion detection, its performance often deteriorates due to the extreme class imbalance present in real-world data. This imbalance causes models to become biased and unable to detect critical attack instances. To address this issue, we propose MCH-Ensemble (Minority Class Highlighting Ensemble), an ensemble framework designed to improve the detection of minority attack classes. The method constructs multiple balanced subsets through random under-sampling and trains base learners, including decision tree, XGBoost, and LightGBM models. Features of correctly predicted attack samples are then amplified by adding a constant value, producing a boosting-like effect that enhances minority class representation. The highlighted subsets are subsequently combined to train a random forest meta-model, which leverages bagging to capture diverse and fine-grained decision boundaries. Experimental evaluations on the UNSW-NB15, CIC-IDS2017, and WSN-DS datasets demonstrate that MCH-Ensemble effectively mitigates class imbalance and achieves superior recognition of DoS attacks. The proposed method achieves enhanced performance compared with those reported previously. On the UNSW-NB15 and CIC-IDS2017 datasets, it achieves improvements in accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC-ROC) by ~1.2% and ~0.61%, ~9.8% and 0.77%, ~0.7% and ~0.56%, ~5.3% and 0.66%, and ~0.1% and ~0.06%, respectively. In addition, it achieves these improvements by ~0.17%, ~1.66%, ~0.11%, ~0.88%, and ~0.06%, respectively, on the WSN-DS dataset. These findings indicate that the proposed framework offers a robust and accurate approach to intrusion detection, contributing to the development of reliable cybersecurity systems in highly imbalanced network environments. Full article

Severely understudied and poorly known ultra-abyssal (hadal) brittle-stars of the genus Abyssura were collected during a recent expedition to the Japan Trench at depths between 6183 and 6539 m and were examined for the first time for both their molecular and detailed morphological data. To date, family-level assignment of the genus Abyssura remains a complete enigma, despite a recent major reorganization of ophiuroid classification. In this study, we infer an all-family level phylogeny of the class Ophiuroidea and find phylogenetic placement for Abyssura, which turns out to be a sister taxon of another little-known ophiuroid genus, Ophiambix, found in hot-vent and cold-seep environments in association with sunken wood at depths between 146 and 5315 m. The sister relationship between the hadal genus Abyssura and the shallow-water-to-abyssal genus Ophiambix is robustly supported by our molecular data, and both external and micromorphological data for these genera are highly consistent. No similar taxa have been found in any of the currently recognized 34 ophiuroid families. Therefore, the genera Abyssura and Ophiambix are assigned to the new family, Abyssuridae fam. nov. This new family shows features of paedomorphic reduction and elucidates the linkage between fauna from both the shallower and the deepest parts of the world’s oceans and provides new insights into the global bathymetric, biogeographic, and diversity patterns of organisms. Full article