Search Results (5,613)

Point cloud quality assessment remains a critical challenge due to the high dimensionality and irregular structure of 3D data, as well as the need to align objective predictions with human perception. To solve this, we suggest a novel graph-based learning architecture that integrates perceptual features with advanced graph neural networks. Our method consists of four main stages: First, key perceptual features, including curvature, saliency, and color, are extracted to capture relevant geometric and visual distortions. Second, a graph-based representation of the point cloud is created using these characteristics, where nodes represent perceptual clusters and weighted edges encode their feature similarities, yielding a structured adjacency matrix. Third, a novel Graph Attention Network Transformer Fusion (GATF) module dynamically refines the importance of these features and generates a unified, view-specific representation. Finally, a Graph Convolutional Network (GCN) regresses the fused features into a final quality score. We validate our approach on three benchmark datasets: ICIP2020, WPC, and SJTU-PCQA. Experimental results demonstrate that our method achieves high correlation with human subjective scores, outperforming existing state-of-the-art metrics by effectively modeling the perceptual mechanisms of quality judgment. Full article

Protein–DNA and protein–RNA interactions are central to gene regulation and genetic disease, yet experimental identification remains costly and complex. Machine learning (ML) offers an efficient alternative, though challenges persist in representing protein sequences due to residue variability, dimensionality issues, and the risk of losing biological context. Traditional approaches such as k-mer counting or neural network encodings provide standardized sequence representations but often demand high computational resources and may obscure functional information. To address these limitations, a novel encoding method based on interpolation of physicochemical properties (PCPs) is introduced. Discrete PCPs values are transformed into continuous functions using logarithmic enhancement, highlighting residues that contribute most to nucleic acid interactions while preserving biological relevance across variable sequence lengths. Statistical features extracted from the resulting spectra via Tsfresh are then used for binary classification of DNA- and RNA-binding proteins. Six classifiers were evaluated, and the proposed method achieved up to 99% accuracy, precision, recall, and F1 score when amino acid highlighting was applied, compared with 66% without highlighting. Benchmarking against k-mer and neural network approaches confirmed superior efficiency and reliability, underscoring the potential of this method for protein interaction prediction. Our framework may be extended to multiclass problems and applied to the study of protein variants, offering a scalable tool for broader protein interaction prediction. Full article

The proliferation of sophisticated malware poses a persistent threat to cybersecurity. While visualizing malware as images enables the use of Convolutional Neural Networks, standard architectures are often inefficient and struggle with the high spatial and channel redundancy inherent in these representations. To address this challenge, we propose LR-MalConv, a new detection framework centered on a novel Low-Redundancy Convolution (LR-Conv) module. The LR-Conv module is uniquely designed to synergistically reduce both spatial redundancy, via a gating and reconstruction mechanism, and channel redundancy, through an efficient split–transform–fuse strategy. By integrating LR-Conv into a ResNet backbone, our framework enhances discriminative feature extraction while significantly reducing computational overhead. Extensive experiments on the Malimg benchmark dataset show our method achieves an accuracy of 99.52%, outperforming existing methods. LR-MalConv establishes a new benchmark for visualized malware detection by striking a superior balance between accuracy and computational efficiency, demonstrating the significant potential of redundancy reduction in this domain. Full article

Memory forensics is an essential cybersecurity tool that comprehensively examines volatile memory to detect the malicious activity of fileless malware that can bypass disk analysis. Image-based detection techniques provide a promising solution by visualizing memory data into images to be used and analyzed by image processing tools and machine learning methods. However, the effectiveness of image-based data for detection and classification requires high computational efforts. This paper investigates the efficacy of texture-based methods in detecting and classifying memory-resident or fileless malware using different image resolutions, identifying the best feature descriptors, classifiers, and resolutions that accurately classify malware into specific families and differentiate them from benign software. Moreover, this paper uses both local and global descriptors, where local descriptors include Oriented FAST and Rotated BRIEF (ORB), Scale-Invariant Feature Transform (SIFT), and Histogram of Oriented Gradients (HOG) and global descriptors include Discrete Wavelet Transform (DWT), GIST, and Gray Level Co-occurrence Matrix (GLCM). The results indicate that as image resolution increases, most feature descriptors yield more discriminative features but require higher computational efforts in terms of time and processing resources. To address this challenge, this paper proposes a novel approach that integrates Local Interpretable Model-agnostic Explanations (LIME) with deep learning models to automatically identify and crop the most important regions of memory images. The LIME’s ROI was extracted based on ResNet50 and MobileNet models’ predictions separately, the images were resized to 128 × 128, and the sampling process was performed dynamically to speed up LIME computation. The ROIs of the images are cropped to new images with sizes of (100 × 100) in two stages: the coarse stage and the fine stage. The two generated LIME-based cropped images using ResNet50 and MobileNet are fed to the lightweight neural network to evaluate the effectiveness of the LIME-based identified regions. The results demonstrate that the LIME-based MobileNet model’s prediction improves the efficiency of the model by preserving important features with a classification accuracy of 85% on multi-class classification. Full article

Weeds continue to pose a serious challenge to agriculture, reducing both the productivity and quality of crops. In this paper, we explore how modern deep learning, specifically Vision Transformers (ViTs), can help address this issue through fast and accurate weed classification. We developed a transformer-based model trained on the DeepWeeds dataset, which contains images of nine different weed species collected under various environmental conditions, such as changes in lighting and weather. By leveraging the ViT architecture, the model is able to capture complex patterns and spatial details in high-resolution images, leading to improved prediction accuracy. We also examined the effects of model optimization techniques, including fine-tuning and the use of pre-trained weights, along with different strategies for handling class imbalance. While traditional oversampling actually hurt performance, dropping accuracy to 94%, using class weights alongside strong data augmentation boosted accuracy to 96.9%. Overall, our ViT model outperformed standard Convolutional Neural Networks, achieving 96.9% accuracy on the held-out test set. Attention-based saliency maps were inspected to confirm that predictions were driven by weed regions, and model consistency under location shift and capture perturbations was assessed using the diverse acquisition sites in DeepWeeds. These findings show that with the right combination of model architecture and training strategies, Vision Transformers can offer a powerful solution for smarter weed detection and more efficient farming practices. Full article

This paper focuses on the issue of unmodeled dynamics and large-range parametric uncertainties in air-breathing hypersonic vehicles (AHV), proposing an adaptive dynamic surface control method based on radial basis function (RBF) neural networks. First, the hypersonic longitudinal model is transformed into a strict-feedback control system with model uncertainties. Then, based on backstepping control theory, adaptive dynamic surface controllers incorporating RBF neural networks are designed separately for the velocity and altitude channels. The proposed controller achieves three key functions: (1) preventing “differential explosion” through low-pass filter design; (2) approximating uncertain model components and unmodeled dynamics using RBF neural networks; (3) enabling real-time adjustment of controller parameters via adaptive methods to accomplish online estimation and compensation of system uncertainties. Finally, stability analysis proves that all closed-loop system signals are semi-globally uniformly bounded (SGUB), with tracking errors converging to an arbitrarily small residual set. The simulation results indicate that the proposed control method reduces steady-state error by approximately 20% compared to traditional controllers. Full article

Artificial intelligence (AI) is rapidly transforming rheumatology, particularly in imaging and laboratory diagnostics where data complexity challenges traditional interpretation. This narrative review summarizes current evidence on AI-driven tools across musculoskeletal ultrasound, radiography, MRI, CT, capillaroscopy, and laboratory analytics. A structured literature search (PubMed, Scopus, Web of Science; 2020–2025) identified 90 relevant publications addressing AI applications in diagnostic imaging and biomarker analysis in rheumatic diseases, while twelve supplementary articles were incorporated to provide contextual depth and support conceptual framing. Deep learning models, notably convolutional neural networks and vision transformers, have demonstrated expert-level accuracy in detecting synovitis, bone marrow edema, erosions, and interstitial lung disease, as well as in quantifying microvascular and structural damage. In laboratory diagnostics, AI enhances the integration of traditional biomarkers with high-throughput omics, automates serologic interpretation, and supports molecular and proteomic biomarker discovery. Multi-omics and explainable AI platforms increasingly enable precision diagnostics and personalized risk stratification. Despite promising performance, widespread implementation is constrained by significant domain-specific validation gaps, data heterogeneity, lack of external validation, ethical concerns, and limited workflow integration. Clinically meaningful progress will depend on transparent, validated, and interoperable AI systems supported by robust data governance and clinician education. The transition from concept to clinic is under way—AI will likely serve as an augmenting rather than replacing partner, standardizing interpretation, accelerating decision-making, and ultimately facilitating precision, data-driven rheumatologic care. Full article

Accurate fault detection for Permanent Magnet Synchronous Motors (PMSMs) prevents costly failures and improves overall reliability. This paper presents a hybrid one-dimensional convolutional neural network (1DCNN)–bidirectional gated recurrent unit (BiGRU) deep learning model for PMSM fault detection. Inverter-driven short-circuit, open-circuit, and thermal faults, as well as stator faults, can cause electrical and thermal disturbances that affect PMSMs. Significant harmonic distortions, current and voltage peaks, and transient fluctuations are introduced by these faults. The proposed architecture utilizes handcrafted features, including statistical analysis, fast Fourier transform (FFT), and Discrete Wavelet Transform (DWT), extracted from the raw PMSM signals to efficiently capture these faults. 1DCNN effectively extracts local and high-frequency fault-related patterns that encode the effects of peaks and harmonic distortions, while the BiGRU of this enriched representation models complex temporal dependencies, including global asymmetries across phase currents and long-term fault evolution trends seen in stator faults and thermal faults. The proposed model reveals the highest metrics for inverter-driven and stator winding fault datasets compared to the other approaches, achieving an accuracy of 99.44% and 99.98%, respectively. As a result, the study with realistic and comprehensive datasets guarantees high accuracy and generalizability not only in the laboratory but also in industry. Full article

Dermoscopy, a non-invasive imaging technique, has transformed dermatology by enabling early detection and differentiation of skin conditions. Integrating deep learning with dermoscopic images enhances diagnostic potential but raises computational challenges. This study introduces APNet, an attention-based architecture designed for melanoma detection, offering fewer parameters than conventional convolutional neural networks. Two baseline models are considered: HU-Net, a trimmed U-Net that uses only the encoding path for classification, and Pocket-Net, a lightweight U-Net variant that reduces parameters through fewer feature maps and efficient convolutions. While Pocket-Net is highly resource-efficient, its simplification can reduce performance. APNet extends Pocket-Net by incorporating squeeze-and-excitation (SE) attention blocks into the encoding path. These blocks adaptively highlight the most relevant dermoscopic features, such as subtle melanoma patterns, improving classification accuracy. The study evaluates APNet against Pocket-Net and HU-Net using four large, annotated dermoscopy datasets (ISIC 2017–2020), covering melanoma, benign nevi, and other lesions. Results show that APNet achieves faster processing than HU-Net while overcoming the performance loss observed in Pocket-Net. By reducing parameters without sacrificing accuracy, APNet provides a practical solution for computationally demanding dermoscopy, offering efficient and accurate melanoma detection where medical imaging resources are limited. Full article

Condition monitoring and fault detection in induction motors (IMs) are priorities in the industrial environment to secure safe conditions for the processes and production. Convolutional Neural Networks (CNNs) are gaining interest in these tasks as they allow automatic extraction of features from the inputs, sometimes Time–Frequency Distributions (TFDs) obtained with various transforms, directly into large models for data classification. This work presents a proposal for the application of a widely used texture analysis tool in the medical field, the 2D Discrete Orthonormal Stockwell Transform (2D-DOST), to improve the accuracy of a lightweight CNN when using different TFDs and comparing the results to the use of the TFDs in RGB and grayscale. The results show that the use of the 2D-DOST improves the classification accuracy in a two to five percent range for all motor conditions under study, while having minimal variations to the training times when compared to RGB or grayscale images, opening the possibility for the use of image processing tools on TFDs to improve automatic feature extraction while using small CNN models. Full article

Machine learning-based digital soil mapping often struggles with spatial heterogeneity and long-range dependencies. To address these limitations, this study proposes Multi-Attention Convolutional Neural Networks (MACNN). This deep learning algorithm integrates multiple attention mechanisms to improve mapping accuracy. First, environmental covariates are determined from the soil-landscape model. These are then fed as structured input to the Convolutional Neural Network. Next, by incorporating Transformer self-attention and multi-head attention mechanisms, this study effectively models the long-range dependencies between soil types and features. Concurrently, the Convolutional Block Attention Module (CBAM) is introduced. CBAM features both channel and spatial dual attention, enabling adaptive weighting of crucial feature channels and spatial locations. This significantly enhances the algorithm’s sensitivity to discriminative information. To validate its effectiveness, the proposed MACNN algorithm was used for soil type mapping in Heilongjiang Province. Compared to Random Forest, Decision Tree, and One-Dimensional Convolutional Neural Network algorithms, MACNN demonstrated superior classification performance. It achieved an overall classification accuracy of 81.27%. An ablation study was conducted to investigate the importance of individual modules within the proposed algorithm. The findings indicate that progressively integrating Transformer and CBAM modules into the 1D-CNN baseline significantly enhances algorithm performance through synergistic gains. Therefore, this integrated algorithm offers a feasible solution to improve digital soil mapping accuracy, providing significant reference value for future research and applications. Full article

Ship instance segmentation in remote sensing images is essential for maritime applications such as intelligent surveillance and port management. However, this task remains challenging due to dense target distributions, large variations in ship scales and shapes, and limited high-quality datasets. The existing YOLOv8 framework mainly relies on convolutional neural networks and CIoU loss, which are less effective in modeling global–local interactions and producing accurate mask boundaries. To address these issues, we propose CDSANet, a novel one-stage ship instance segmentation network. CDSANet integrates convolutional operations, Vision Transformers, and attention mechanisms within a unified architecture. The backbone adopts a Convolutional Vision Transformer Attention (CVTA) module to enhance both local feature extraction and global context perception. The neck employs dynamic-weighted DOWConv to adaptively handle multi-scale ship instances, while SIoU loss improves localization accuracy and orientation robustness. Additionally, CBAM enhances the network’s focus on salient regions, and a MixUp-based augmentation strategy is used to improve model generalization. Extensive experiments on the proposed VLRSSD dataset demonstrate that CDSANet achieves state-of-the-art performance with a mask AP (50–95) of 75.9%, surpassing the YOLOv8 baseline by 1.8%. Full article

The article presents a method for detecting low-intensity DDoS attacks, focused on identifying difficult-to-detect “low-and-slow” scenarios that remain undetectable by traditional defence systems. The key feature of the developed method is the statistical criteria’s (χ² and T statistics, energy ratio, reconstruction errors) integration with a combined neural network architecture, including convolutional and transformer blocks coupled with an autoencoder and a calibrated regressor. The developed neural network architecture combines mathematical validity and high sensitivity to weak anomalies with the ability to generate interpretable artefacts that are suitable for subsequent forensic analysis. The developed method implements a multi-layered process, according to which the first level statistically evaluates the flow intensity and interpacket intervals, and the second level processes features using a neural network module, generating an integral blend-score S metric. ROC-AUC and PR-AUC metrics, learning curve analysis, and the estimate of the calibration error (ECE) were used for validation. Experimental results demonstrated the superiority of the proposed method over existing approaches, as the achieved values of ROC-AUC and PR-AUC were 0.80 and 0.866, respectively, with an ECE level of 0.04, indicating a high accuracy of attack detection. The study’s contribution lies in a method combining statistical and neural network analysis development, as well as in ensuring the evidentiary value of the results through the generation of structured incident reports (PCAP slices, time windows, cryptographic hashes). The obtained results expand the toolkit for cyber-attack analysis and open up prospects for the methods’ practical application in monitoring systems and law enforcement agencies. Full article

Introduction: To evaluate the feasibility of multimodal deep learning (DL) for automated staging of familial exudative vitreoretinopathy (FEVR) using color fundus photographs (CFP) and fluorescein angiography (FFA). Methods: We assembled a multimodal dataset across FEVR stages 0–5 and post-laser cases and benchmarked CNNs (Convolutional Neural Networks), Transformers, and multimodal fusion under center-region and multi-image settings. Class imbalance was mitigated via weighted sampling and focal/class-balanced losses. We report accuracy, recall, precision, macro-F1, Cohen’s κ, and class-wise ROC/AUC with 95% Cis. Results: AI system showed balanced performance versus specialists (0.65 vs. Dr. A: 0.48/Dr. B: 0.48) in CFP assessment, maintaining high specificity (0.91–0.92). Among architectures: (1) Transformers outperformed CNNs in single-modal analysis; (2) ResNet showed moderate performance (AUC 0.70–0.85) but limited capability for intermediate grades (AUC < 0.70); (3) CRD-Net achieved peak performance (AUC up to 0.94, severe cases AUC > 0.90). While FFA improved Dr. B’s accuracy to 0.56, it remained below AI levels. Stage-specific accuracy ranged from 0.72 to 0.88 across the FEVR spectrum. Conclusions: Leveraging a novel multimodal database and high-performance AI models, systematic comparisons demonstrated the superiority of Transformer architectures over CNNs in single-modal analysis, while CRD-Net’s multimodal fusion approach achieved optimal performance across all severity grades. Multimodal DL shows feasibility as a decision-support tool for automated FEVR staging within confirmed cohorts. Full article

With ongoing climate transformations, reliable Arctic sea ice forecasts are essential for understanding impacts on shipping, ecosystems, and climate teleconnections. This research examines physics-free neural architectures versus physics-informed statistical models for long-term Arctic projections by implementing Fourier Neural Operator (FNO) and Convolutional Neural Network (CNN) alongside a seasonal SARIMAX time series model incorporating physical predictors including temperature anomalies and ice thickness. We test whether neural models trained on historical ice data can match physics-informed SARIMAX reliability, and whether approaches exhibit systematic biases toward specific emission pathways. Using data from January 1979 to December 2024, we conducted forecasts through 2100, with SARIMAX driven by CMIP6 sea ice thickness under SSP2-4.5 and SSP5-8.5 scenarios. Results decisively reject the first hypothesis: both neural models projected ice free Arctic summer by September 2089 regardless of emission scenario, while SARIMAX maintained physically plausible seasonal coverage throughout the century under both pathways. Neural approaches demonstrated systematic bias toward extreme warming exceeding even high-emission projections, revealing fundamental limitations in physics-free deep learning for climate forecasting where physical constraints are paramount. Full article