Search Results (5,786)

Early and accurate diagnosis of dementia is essential to improving patient outcomes and reducing societal burden. The Mini-Mental State Examination (MMSE) is widely used to assess cognitive function, yet traditional statistical and machine learning approaches often face limitations in capturing nonlinear interactions and subtle decline patterns. This study developed a novel deep learning-based dementia prediction model using MMSE data collected from domestic clinical settings and compared its performance with traditional machine learning models. A notable strength of this work lies in its use of item-level MMSE features combined with explainable AI (SHAP analysis), enabling both high predictive accuracy and clinical interpretability—an advancement over prior approaches that primarily relied on total scores or linear modeling. Data from 164 participants, classified into cognitively normal, mild cognitive impairment (MCI), and dementia groups, were analyzed. Individual MMSE items and total scores were used as input features, and the dataset was divided into training and validation sets (8:2 split). A fully connected neural network with regularization techniques was constructed and evaluated alongside Random Forest and support vector machine (SVM) classifiers. Model performance was assessed using accuracy, F1-score, confusion matrices, and receiver operating characteristic (ROC) curves. The deep learning model achieved the highest performance (accuracy 0.90, F1-score 0.90), surpassing Random Forest (0.86) and SVM (0.82). SHAP analysis identified Q11 (immediate memory), Q12 (calculation), and Q17 (drawing shapes) as the most influential variables, aligning with clinical diagnostic practices. These findings suggest that deep learning not only enhances predictive accuracy but also offers interpretable insights aligned with clinical reasoning, underscoring its potential utility as a reliable tool for early dementia diagnosis. However, the study is limited by the use of data from a single clinical site with a relatively small sample size, which may restrict generalizability. Future research should validate the model using larger, multi-institutional, and multimodal datasets to strengthen clinical applicability and robustness. Full article

Crop canopy height is a key structural indicator that is strongly associated with crop development, biomass accumulation, and crop health. To overcome the limitations of time-consuming and labor-intensive traditional field measurements, Unmanned Aerial Vehicle (UAV)-based Light Detection and Ranging (LiDAR) offers an efficient alternative by capturing three-dimensional point cloud data (PCD). In this study, UAV-LiDAR data were acquired using a DJI Matrice 600 Pro equipped with a 16-channel LiDAR system. Three canopy height estimation methodological approaches were evaluated across three crop types: corn, soybean, and winter wheat. Specifically, this study assessed machine learning regression modeling, ground point classification techniques, percentile-based method and a newly proposed Dual-Range Averaging (DRA) method to identify the most effective method while ensuring practicality and reproducibility. The best-performing method for corn was Support Vector Regression (SVR) with a linear kernel (R² = 0.95, RMSE = 0.137 m). For soybean, the DRA method yielded the highest accuracy (R² = 0.93, RMSE = 0.032 m). For winter wheat, the PointCNN deep learning model demonstrated the best performance (R² = 0.93, RMSE = 0.046 m). These results highlight the effectiveness of integrating UAV-LiDAR data with optimized processing methods for accurate and widely applicable crop height estimation in support of precision agriculture practices. Full article

Financial time series prediction remains a challenging task due to the inherent non-stationarity, noise, and complex temporal dependencies present in market data. Traditional forecasting methods often fail to capture the multifaceted nature of financial markets, where temporal proximity, trend dynamics, and volatility patterns simultaneously influence price movements. To address these limitations, this paper proposes the Enhanced Multi-Aspect Transformer (EMAT), a novel deep learning architecture specifically designed for stock market prediction. EMAT incorporates a Multi-Aspect Attention Mechanism that simultaneously captures temporal decay patterns, trend dynamics, and volatility regimes through specialized attention components. The model employs an encoder–decoder architecture with enhanced feed-forward networks utilizing SwiGLU activation, enabling superior modeling of complex non-linear relationships. Furthermore, we introduce a comprehensive multi-objective loss function that balances point-wise prediction accuracy with volatility consistency. Extensive experiments on multiple stock market datasets demonstrate that EMAT consistently outperforms a wide range of state-of-the-art baseline models, including various recurrent, hybrid, and Transformer architectures. Our ablation studies further validate the design, confirming that each component of the Multi-Aspect Attention Mechanism makes a critical and quantifiable contribution to the model’s predictive power. The proposed architecture’s ability to simultaneously model these distinct financial characteristics makes it a particularly effective and robust tool for financial forecasting, offering significant improvements in accuracy compared to existing approaches. Full article

Background: Timely intervention and effective control of Alzheimer’s disease (AD) have been shown to limit memory loss and preserve cognitive function and the ability to perform simple activities in older adults. In addition, magnetic resonance imaging (MRI) scans are one of the most common and effective methods for early detection of AD. With the rapid development of deep learning (DL) algorithms, AD detection based on deep learning has wide applications. Methods: In this research, we have developed an AD detection method based on three-dimensional (3D) convolutional neural networks (CNNs) for 3D MRI images, which can achieve strong accuracy when compared with traditional 3D CNN models. The proposed model has four main blocks, and the multi-layer fusion functionality of each block was used to improve the efficiency of the proposed model. The performance of the proposed model was compared with three different pre-trained 3D CNN architectures (i.e., 3D ResNet-18, 3D InceptionResNet-v2, and 3D Efficientnet-b2) in both tasks of multi-/binary-class classification of AD. Results: Our model achieved impressive classification results of 91.4% for binary-class as well as 80.6% for multi-class classification on the Open Access Series of Imaging Studies (OASIS) database. Conclusions: Such results serve to demonstrate that multi-stage feature fusion of 3D CNN is an effective solution to improve the accuracy of diagnosis of AD with 3D MRI, thus enabling earlier and more accurate diagnosis. Full article

Multi-modal sensor fusion has become a cornerstone of robust autonomous driving systems, enabling perception models to integrate complementary cues from cameras, LiDARs, radars, and other modalities. This survey provides a structured overview of recent advances in deep learning-based fusion methods, categorizing them by architectural paradigms (e.g., BEV-centric fusion and cross-modal attention), learning strategies, and task adaptations. We highlight two dominant architectural trends: unified BEV representation and token-level cross-modal alignment, analyzing their design trade-offs and integration challenges. Furthermore, we review a wide range of applications, from object detection and semantic segmentation to behavior prediction and planning. Despite considerable progress, real-world deployment is hindered by issues such as spatio-temporal misalignment, domain shifts, and limited interpretability. We discuss how recent developments, such as diffusion models for generative fusion, Mamba-style recurrent architectures, and large vision–language models, may unlock future directions for scalable and trustworthy perception systems. Extensive comparisons, benchmark analyses, and design insights are provided to guide future research in this rapidly evolving field. Full article

Graph neural networks (GNNs) are deep learning models that process structured graph data. By leveraging their graphs/node classification and link prediction capabilities, they have been effectively applied in multiple domains such as community detection, location sharing services, and drug discovery. These powerful applications and the vast availability of graphs in diverse fields have facilitated the adoption of GNNs in privacy-sensitive contexts (e.g., banking systems and healthcare). Unfortunately, GNNs are vulnerable to the leakage of sensitive information through well-defined attacks. Our main focus is on membership inference attacks (MIAs) that allow the attacker to infer whether a given sample belongs to the training dataset. To prevent this, we introduce three LLM-guided defense mechanisms applied at the posterior level: posterior encoding with noise, knowledge distillation, and secure aggregation. Our proposed approaches not only successfully reduce MIA accuracy but also maintain the model’s performance on the node classification task. Our findings, validated through extensive experiments on widely used GNN architectures, offer insights into balancing privacy preservation with predictive performance. Full article

Advancements in electroencephalography (EEG) technology and feature extraction methods have paved the way for wearable, non-invasive systems that enable continuous sleep monitoring outside clinical environments. This study presents the development and evaluation of an EEG-based acquisition system for sleep staging, which can be adapted for wearable applications. The system utilizes a custom experimental setup with the ADS1299EEG-FE-PDK evaluation board to acquire EEG signals from the forehead and in-ear regions under various conditions, including visual and auditory stimuli. Afterward, the acquired signals were processed to extract a wide range of features in time, frequency, and non-linear domains, selected based on their physiological relevance to sleep stages and disorders. The feature set was reduced using the Minimum Redundancy Maximum Relevance (mRMR) algorithm and Principal Component Analysis (PCA), resulting in a compact and informative subset of principal components. Experiments were conducted on the Bitbrain Open Access Sleep (BOAS) dataset to validate the selected features and assess their robustness across subjects. The feature set extracted from a single EEG frontal derivation (F4-F3) was then used to train and test a two-step deep learning model that combines Long Short-Term Memory (LSTM) and dense layers for 5-class sleep stage classification, utilizing attention and augmentation mechanisms to mitigate the natural imbalance of the feature set. The results—overall accuracies of 93.5% and 94.7% using the reduced feature sets (94% and 98% cumulative explained variance, respectively) and 97.9% using the complete feature set—demonstrate the feasibility of obtaining a reliable classification using a single EEG derivation, mainly for unobtrusive, home-based sleep monitoring systems. Full article

Edge intelligence plays an increasingly vital role in ensuring the reliability of distributed microservice-based applications, which are widely used in domains such as e-commerce, industrial IoT, and cloud-edge collaborative platforms. However, anomaly detection in these systems encounters a critical challenge: labeled anomaly data are scarce. This scarcity leads to severe class asymmetry and compromised detection performance, particularly under the resource constraints of edge environments. Recent approaches based on Graph Neural Networks (GNNs)—often integrated with DeepSVDD and regularization techniques—have shown potential, but they rarely address this asymmetry in an adaptive, scenario-specific way. This work proposes Heterogeneous Graph Adaptive Augmentation (HGAA), a framework tailored for edge intelligence scenarios. HGAA dynamically optimizes graph data augmentation by leveraging feedback from online anomaly detection. To enhance detection accuracy while adhering to resource constraints, the framework incorporates a selective bias toward underrepresented anomaly types. It uses knowledge distillation to model dataset-dependent distributions and adaptively adjusts augmentation probabilities, thus avoiding excessive computational overhead in edge environments. Additionally, a dynamic adjustment mechanism evaluates augmentation success rates in real time, refining the selection processes to maintain model robustness. Experiments were conducted on two real-world datasets (TraceLog and FlowGraph) under simulated edge scenarios. Results show that HGAA consistently outperforms competitive baseline methods. Specifically, compared with the best non-adaptive augmentation strategies, HGAA achieves an average improvement of 4.5% in AUC and 4.6% in AP. Even larger gains are observed in challenging cases: for example, when using the HGT model on the TraceLog dataset, AUC improves by 14.6% and AP by 18.1%. Beyond accuracy, HGAA also significantly enhances efficiency: compared with filter-based methods, training time is reduced by up to 71% on TraceLog and 8.6% on FlowGraph, confirming its suitability for resource-constrained edge environments. These results highlight the potential of adaptive, edge-aware augmentation techniques in improving microservice anomaly detection within heterogeneous, resource-limited environments. Full article

Accurate and robust power quality disturbance (PQD) classification is critical for modern electrical grids, particularly in noisy environments. This study presents a comprehensive comparative evaluation of machine learning (ML) and deep learning (DL) models for automatic PQD identification. The models evaluated include Support Vector Machines (SVM), Decision Trees (DT), Random Forest (RF), k-Nearest Neighbors (kNN), Gradient Boosting (GB), and Dense Neural Networks (DNN). For experimentation, a hybrid dataset, comprising both synthetic and real signals, was used to assess model performance. The robustness of the models was evaluated by systematically introducing Gaussian noise across a wide range of Signal-to-Noise Ratios (SNRs). A central objective was to directly benchmark the practical implementation and performance of these models across two widely used platforms: MATLAB R2024a and Python 3.11. Results show that ML models achieve high accuracies, exceeding $95\%$ at an SNR of 10 dB. DL models exhibited remarkable stability, maintaining $97\%$ accuracy for SNRs above 10 dB. However, their performance degraded significantly at lower SNRs, revealing specific confusion patterns. The analysis underscores the importance of multi-domain feature extraction and adaptive preprocessing for achieving resilient PQD classification. This research provides valuable insights and a practical guide for implementing and optimizing robust PQD classification systems in real-world, noisy scenarios. Full article

The continuous growth in global demand for mineral resources and the increasing difficulty of mineral exploration have created bottlenecks for traditional mineral prediction methods in handling complex geological information and large amounts of data. This review aims to explore the latest research progress in machine learning technology in the field of large-scale mineral prediction from 2016 to 2025. By systematically searching the Web of Science core database, we have screened and analyzed 255 high-quality scientific studies. These studies cover key areas such as mineral information extraction, target area selection, mineral regularity modeling, and resource potential evaluation. The applied machine learning technologies include Random Forests, Support Vector Machines, Convolutional Neural Networks, Recurrent Neural Networks, etc., and have been widely used in the exploration and prediction of various mineral deposits such as porphyry copper, sandstone uranium, and tin. The findings indicate a substantial shift within the discipline towards the utilization of deep learning methodologies and the integration of multi-source geological data. There is a notable rise in the deployment of cutting-edge techniques, including automatic feature extraction, transfer learning, and few-shot learning. This review endeavors to synthesize the prevailing state and prospective developmental trajectory of machine learning within the domain of large-scale mineral prediction. It seeks to delineate the field’s progression, spotlight pivotal research dilemmas, and pinpoint innovative breakthroughs. Full article

Stomata on the leaves of wheat serve as important gateways for gas exchange with the external environment. Their morphological characteristics, such as size and density, are closely related to physiological processes like photosynthesis and transpiration. However, due to the limitations of existing analysis methods, the efficiency of analyzing and mining stomatal phenotypes and their associated genes still requires improvement. To enhance the accuracy and efficiency of stomatal phenotype traits analysis and to uncover the related key genes, this study selected 210 wheat varieties. A novel semantic segmentation model based on transformer for wheat stomata, called Wheat Stoma Former (WSF), was proposed. This model enables fully automated and highly efficient stomatal mask extraction and accurately analyzes phenotypic traits such as the length, width, area, and number of stomata on both the adaxial (Ad) and abaxial (Ab) surfaces of wheat leaves based on the mask images. The model evaluation results indicate that coefficients of determination (R²) between the predicted values and the actual measurements for stomatal length, width, area, and number were 0.88, 0.86, 0.81, and 0.93, respectively, demonstrating the model’s high precision and effectiveness in stomatal phenotypic trait analysis. The phenotypic data were combined with sequencing data from the wheat 660 K SNP chip and subjected to a genome-wide association study (GWAS) to analyze the genetic basis of stomatal traits, including length, width, and number, on both adaxial and abaxial surfaces. A total of 36 SNP peak loci significantly associated with stomatal traits were identified. Through candidate gene identification and functional analysis, two genes—TraesCS2B02G178000 (on chromosome 2B, related to stomatal number on the abaxial surface) and TraesCS6A02G290600 (on chromosome 6A, related to stomatal length on the adaxial surface)—were found to be associated with stomatal traits involved in regulating stomatal movement and closure, respectively. In conclusion, our WSF model demonstrates valuable advances in accurate and efficient stomatal phenotyping for locating genes related to stomatal traits in wheat and provides breeders with accurate phenotypic data for the selection and breeding of water-efficient wheat varieties. Full article

Background & Objective: Stroke remains a leading cause of mortality and long-term disability worldwide, demanding rapid and accurate diagnosis to improve patient outcomes. Computed tomography (CT) scans are widely used in emergency settings due to their speed, availability, and cost-effectiveness. This study proposes an artificial intelligence (AI)-based framework for multiclass stroke classification (ischemic, hemorrhagic, and no stroke) using CT scan images from the Ministry of Health of the Republic of Turkey. Methods: We adopted MaxViT, a state-of-the-art Vision Transformer (ViT)-based architecture, as the primary deep learning model for stroke classification. Additional transformer variants, including Vision Transformer (ViT), Transformer-in-Transformer (TNT), and ConvNeXt, were evaluated for comparison. To improve model generalization and handle class imbalance, classical data augmentation techniques were applied. Furthermore, explainable AI (XAI) was integrated using Grad-CAM++ to provide visual insights into model decisions. Results: The MaxViT model with augmentation achieved the highest performance, reaching an accuracy and F1-score of 98.00%, outperforming the baseline Vision Transformer and other evaluated models. Grad-CAM++ visualizations confirmed that the proposed framework effectively identified stroke-related regions, enhancing transparency and clinical trust. Conclusions: This research contributes to the development of a trustworthy AI-assisted diagnostic tool for stroke, facilitating its integration into clinical practice and improving access to timely and optimal stroke diagnosis in emergency departments. Full article

Conventional fundus (CF) and ultrawidefield fundus (UF) imaging are two primary modalities widely used in ophthalmology. Despite the complementary use of both imaging modalities in clinical practice, existing research on fundus image translation has yet to reach clinical viability and often lacks the necessary accuracy and detail required for practical medical use. Additionally, collecting paired UFI-CFI data from the same patients presents significant limitations, and unpaired learning-based generative models frequently suffer from distortion phenomena, such as hallucinations. This study introduces an enhanced modality transformation method to improve the diagnostic support capabilities of deep learning models in ophthalmology. The proposed method translates UF images (UFIs) into CF images (CFIs), potentially replacing the dual-imaging approach commonly used in clinical practice. This replacement can significantly reduce financial and temporal burdens on patients. To achieve this, this study leveraged UFI–CFI image pairs obtained from the same patient on the same day. This approach minimizes information distortion and accurately converts the two modalities. Our model employs scaled feature registration and distorted vessel correction methods to align UFI–CFI pairs effectively. The generated CFIs not only enhance image quality and better represent the retinal area compared to existing methods but also effectively preserve disease-related details from UFIs, aiding in accurate diagnosis. Furthermore, compared with existing methods, our model demonstrated a substantial 18.2% reduction in MSE, a 7.2% increase in PSNR, and a 12.7% improvement in SSIM metrics. Notably, our results show that the generated CFIs are nearly indistinguishable from the real CFIs, as confirmed by ophthalmologists. Full article

A wide range of research has been conducted on path planning and collision avoidance to enhance the operational efficiency of unmanned aerial vehicles (UAVs). The existing works have mainly assumed an environment with static obstacles and global positioning system (GPS) signals. However, practical environments have often been involved with dynamic obstacles, dense areas with numerous obstacles in confined spaces, and blocked GPS signals. In order to consider these issues for practical implementation, a deep reinforcement learning (DRL)-based method is proposed for path planning and collision avoidance in GPS-denied and dense environments. In the proposed method, robust path planning and collision avoidance can be conducted by using the received signal strength (RSS) value with the extended Kalman filter (EKF). Additionally, the attitude of the UAV is adopted as part of the action space to enable the generation of smooth trajectories. Performance was evaluated under single- and multi-target scenarios with numerous dynamic obstacles. Simulation results demonstrated that the proposed method can generate smoother trajectories and shorter path lengths while consistently maintaining a lower collision rate compared to conventional methods. Full article

At present, there is growing interest in automated biometric identification applications. For these, it is crucial to have a system capable of accurately identifying a specific group of people while also detecting individuals who do not belong to that group. In face identification models that use Deep Learning (DL) techniques, this context is referred to as Open-Set Recognition (OSR), which is the focus of this work. This scenario presents a substantial challenge for this type of system, as it involves the need to effectively identify unknown individuals who were not part of the system’s training data. In this context, where the accuracy of this type of system is considered crucial, selecting the model to be used in each scenario becomes key. It is within this context that our work arises. Here, we present the results of a rigorous comparative analysis examining the precision of some of the most widely used models today for face identification, specifically some Convolutional Neural Network (CNN) models compared with a Vision Transformer (ViT) model. All models were pre-trained on the same large dataset and evaluated in an OSR scenario. The results show that ViT achieves the highest precision, outperforming CNN baselines and demonstrating better generalization for unknown identities. These findings support recent evidence that ViT is a promising alternative to CNN for this type of application. Full article