Search Results (628)

The unique geographical isolation and climate sensitivity of island ecosystems make them valuable for ecological research. The Normalized Difference Vegetation Index (NDVI) is an important indicator when monitoring and evaluating these systems, and its prediction has become a key research focus. However, island NDVI prediction remains uncertain due to a limited understanding of vegetation growth and insufficient high-quality data. Deterministic models fail to capture or quantify such uncertainty, often leading to overfitting. To address this issue, this study proposes an uncertainty prediction model for the island NDVI within a coding–prediction–decoding framework, referred to as a Convolutional Autoencoder–Regularized Bayesian Neural Network (CAE-RBNN). The model integrates a convolutional autoencoder with feature regularization to extract latent NDVI features, aiming to reconcile spatial scale disparities with environmental data, while a Bayesian Neural Network (BNN) quantifies uncertainty arising from limited samples and an incomplete understanding of the process. Finally, Monte Carlo sampling and SHAP analysis evaluate model performance, quantify predictive uncertainty, and enhance interpretability. Experiments on six islands in the Xisha archipelago demonstrate that CAE-RBNN outperforms the Convolutional Neural Network–Recurrent Neural Network (CNN-RNN), the Convolutional Recurrent Neural Network (ConvRNN), Convolutional Long Short-Term Memory (ConvLSTM), and Random Forest (RF). Among them, CAE-RBNN reduces the MAE and MSE of the single-time-step prediction task by 8.40% and 10.69%, respectively, compared with the suboptimal model and decreases them by 16.31% and 22.57%, respectively, in the continuous prediction task. More importantly, it effectively quantifies the uncertainty of different driving forces, thereby improving the reliability of island NDVI predictions influenced by the environment. Full article

The high peak-to-average power ratio (PAPR) in classical high-speed digital data transmission systems with orthogonal frequency division multiplexing (OFDM) limits energy efficiency and communication range. This paper proposes a method for randomizing OFDM signals via frequency coding using synthesized pseudorandom sequences with improved autocorrelation properties, obtained through machine learning, to minimize PAPR in complex, non-stationary hydroacoustic channels for communicating with underwater robotic systems. A neural network architecture was developed and trained to generate codes of up to 150 elements long based on an analysis of patterns in previously found best short sequences. The obtained class of OFDM signals does not require regular and accurate estimation of channel parameters while remaining resistant to various types of impulse noise, Doppler shifts, and significant multipath interference typical of the underwater environment. The attained spectral efficiency values (up to 0.5 bits/s/Hz) are relatively high for existing hydroacoustic communication systems. It has been shown that the peak power of such multi-frequency information transmission systems can be effectively reduced by an average of 5–10 dB, which allows for an increase in the communication range compared to classical OFDM methods in non-stationary hydrological conditions at acceptable bit error rates (from 10⁻² to 10⁻³ and less). The effectiveness of the proposed methods of randomization with synthesized codes and frequency coding for OFDM signals was confirmed by field experiments at sea on the shelf, over distances of up to 4.2 km, with sea waves of up to 2–3 Beaufort units and mutual movement of the transmitter and receiver. Full article

Grain size is critical for metallic material performance, yet conventional ultrasonic methods rely on strong model assumptions and exhibit limited adaptability. We propose a deep learning architecture that uses multimodal ultrasonic features with spatial coding to predict the grain size distribution of GH4099. A-scan signals from C-scan measurements are converted to time–frequency representations and fed to an encoder–decoder model that combines a dual convolutional compression network with a fully connected decoder. A thickness-encoding branch enables feature decoupling under physical constraints, and an elliptic spatial fusion strategy refines predictions. Experiments show mean and standard deviation MAEs of 1.08 and 0.84 μm, respectively, with a KL divergence of 0.0031, outperforming attenuation- and velocity-based methods. Input-specificity experiments further indicate that transfer learning calibration quickly restores performance under new conditions. These results demonstrate a practical path for integrating deep learning with ultrasonic inspection for accurate, adaptable grain-size characterization. Full article

Endometriosis is a chronic gynecological condition characterized by endometrium-like tissue outside the uterus. In clinical practice, diagnosis and anatomical mapping rely heavily on imaging, yet performance remains operator- and modality-dependent. Artificial intelligence (AI) has been increasingly applied to endometriosis imaging. We conducted a PRISMA-ScR-guided scoping review of primary machine learning and deep learning studies using endometriosis-related imaging. Five databases (MEDLINE, Embase, Scopus, IEEE Xplore, and Google Scholar) were searched from 2015 to 2025. Of 413 records, 32 studies met inclusion and most were single-center, retrospective investigations in reproductive-age cohorts. Ultrasound predominated (50%), followed by laparoscopic imaging (25%) and MRI (22%); ovarian endometrioma and deep infiltrating endometriosis were the most commonly modeled phenotypes. Classification was the dominant AI task (78%), typically using convolutional neural networks (often ResNet-based), whereas segmentation (31%) and object detection (3%) were less explored. Nearly all studies relied on internal validation (97%), most frequently simple hold-out splits with heterogeneous, accuracy-focused performance reporting. The minimal AI-method quality appraisal identified frequent methodological gaps across key domains, including limited reporting of patient-level separation, leakage safeguards, calibration, and data and code availability. Overall, AI-enabled endometriosis imaging is rapidly evolving but remains early-stage; multi-center and prospective validation, standardized reporting, and clinically actionable detection–segmentation pipelines are needed before routine clinical integration. Full article

Cross-site scripting (XSS) attacks are among the threats facing web security, resulting from the diversity and complexity of HTML formats. Research has shown that some text processing-based methods are limited in their ability to detect this type of attack. This article proposes an approach aimed at improving the detection of this type of attack, taking into account the limitations of certain techniques. It combines the effectiveness of deep learning represented by convolutional neural networks (CNN) and the accuracy of classification methods represented by support vector machines (SVM). It takes advantage of the ability of CNNs to effectively detect complex visual patterns in the face of injection variations and the SVM’s powerful classification capability, as XSS attacks often use obfuscation or encryption techniques that are difficult to be detected with textual methods alone. This work relies on a dataset that focuses specifically on XSS attacks, which is available on Kaggle and contains 13,686 sentences in script form, including benign and malicious cases associated with these attacks. Benign data represents 6313 cases, while malicious data represents 7373 cases. The model was trained on 80% of this data, while the remaining 20% was allocated for test. Computer vision techniques were used to analyze the visual patterns in the images and extract distinctive features, moving from a textual representation to a visual one where each character is converted into its ASCII encoding, then into grayscale pixels. In order to visually distinguish the characteristics of normal and malicious code strings and the differences in their visual representation, a CNN model was used in the analysis. The convolution and subsampling (pooling) layers extract significant patterns at different levels of abstraction, while the final output is converted into a feature vector that can be exploited by a classification algorithm such as an Optimized SVM. The experimental results showed excellent performance for the model, with an accuracy of (99.7%), and this model is capable of generalizing effectively without the risk of overfitting or loss of performance. This significantly enhances the security of web applications by providing robust protection against complex XSS threats. Full article

Cardiac arrhythmias, especially premature beats and atrial fibrillation, pose substantial clinical risks and detection hurdles. While deep learning has shown promise for automated arrhythmia diagnosis, single-model architectures often lack sufficient performance in distinguishing these two arrhythmia types. This study seeks to address the limitations of individual deep learning models and boost classification accuracy for premature beats and atrial fibrillation. It proposes an arrhythmia classification model integrating multiscale feature enhancement and bidirectional temporal dependency. First, a four-layer convolutional residual module with skip connections extracts multiscale local electrocardiogram (ECG) features. Then, multi-head self-attention strengthens critical feature global correlations. Next, a bidirectional long-term temporal de-pendency network captures sequence contextual dependencies. Finally, a Dropout-regularized fully connected layer enables six-type arrhythmia classification. Experiments on a fused dataset (MIT-BIH arrhythmia, MIT-BIH atrial fibrillation, and CODE datasets) yield an overall accuracy of 98.55% and F1-score of 0.9531. Notably, the F1-scores for premature beats (0.9916) and atrial fibrillation (0.9888) outperform recent literature by 2.16% and 4.39%, respectively. The model demonstrates robust classification performance with effective identification of the target arrhythmias, highlighting its potential as a supportive tool for automated ECG diagnosis. Full article

With the acceleration of digital transformation, open-source software has become a fundamental component of modern smart grids and other critical infrastructures. However, the complex dependency structures of open-source ecosystems and the continuous emergence of vulnerabilities pose substantial challenges to software supply chain security. In power information networks and cyber–physical control systems, vulnerabilities in open-source components integrated into Supervisory Control and Data Acquisition (SCADA), Energy Management System (EMS), and Distribution Management System (DMS) platforms and distributed energy controllers may propagate along the supply chain, threatening system security and operational stability. In such application scenarios, large language models (LLMs) often suffer from limited semantic accuracy when handling domain-specific security terminology, as well as deployment inefficiencies that hinder their practical adoption in critical infrastructure environments. To address these issues, this paper proposes KD-SecBERT, a domain-specific semantic bidirectional encoder optimized through multi-level knowledge distillation for open-source software supply chain security in smart grid applications. The proposed framework constructs a hierarchical multi-teacher ensemble that integrates general language understanding, cybersecurity-domain knowledge, and code semantic analysis, together with a lightweight student architecture based on depthwise separable convolutions and multi-head self-attention. In addition, a dynamic, multi-dimensional distillation strategy is introduced to jointly perform layer-wise representation alignment, ensemble knowledge fusion, and task-oriented optimization under a progressive curriculum learning scheme. Extensive experiments conducted on a multi-source dataset comprising National Vulnerability Database (NVD) and Common Vulnerabilities and Exposures (CVE) entries, security-related GitHub code, and Open Web Application Security Project (OWASP) test cases show that KD-SecBERT achieves an accuracy of 91.3%, a recall of 90.6%, and an F1-score of 89.2% on vulnerability classification tasks, indicating strong robustness in recognizing both common and low-frequency security semantics. These results demonstrate that KD-SecBERT provides an effective and practical solution for semantic analysis and software supply chain risk assessment in smart grids and other critical-infrastructure environments. Full article

This work investigates the effectiveness of block transform coding (BTC) as a lightweight, training-free quantization strategy for compressing the weights of pretrained deep neural networks. The proposed method applies a rule-based block transform with variance and root mean square error (RMSE)-driven stopping criteria, enabling substantial reductions in bit precision while preserving the statistical structure of convolutional and fully connected layer weights. Unlike uniform 8-bit quantization, BTC dynamically adjusts bit usage across layers and achieves significantly lower distortion for the same compression budget. We evaluate BTC across many pretrained architectures and tabular benchmarks. Experimental results show that BTC consistently reduces storage to 4–7.7 bits per weight while maintaining accuracy within 2–3% of the 32-bit floating point (FP32) baseline. To further assess scalability and baseline strength, BTC is additionally evaluated on large-scale ImageNet models and compared against a calibrated percentile-based uniform post-training quantization method. The results show that BTC achieves a substantially lower effective bit-width while incurring only a modest accuracy reduction relative to calibration-aware 8-bit quantization, highlighting a favorable compression–accuracy trade-off. BTC also exhibits stable behavior across successive post-training quantization (PTQ) configurations, low quantization noise, and smooth RMSE trends, outperforming naïve uniform quantization under aggressive compression. These findings confirm that BTC provides a scalable, architecture-agnostic, and training-free quantization mechanism suitable for deployment in memory- and computing-constrained environments. Full article

Beamforming is highly significant for the physical layer of wireless communication systems, for multi-antenna systems such as multiple input multiple output (MIMO) and massive MIMO, since it improves spectral efficiency and reduces interference. Traditional linear beamforming methods such as zero-forcing beamforming (ZFBF) and minimum mean square error (MMSE) beamforming provide closed-form solutions. Yet, their performance drops when they face non-ideal conditions such as imperfect channel state information (CSI), dynamic propagation environment, or high-dimensional system configurations, primarily due to static assumptions and computational limitations. These limitations have led to the rise of deep learning-based beamforming, where data-driven models derive beamforming solutions directly from CSI. By leveraging the representational capabilities of cutting-edge deep learning architectures, along with the increasing availability of data and computational resources, deep learning presents an adaptive and potentially scalable alternative to traditional methodologies. In this work, we unify and systematically compare our two unsupervised learning architectures for uplink receive beamforming: a simple neural network beamforming (NNBF) model, composed of convolutional and fully connected layers, and a transformer-based NNBF model that integrates grouped convolutions for feature extraction and transformer blocks to capture long-range channel dependencies. They are evaluated in a common multi-user single input multiple output (MU-SIMO) system model to maximize sum-rate across single-antenna user equipments (UEs) under 3GPP-compliant channel models, namely TDL-A and UMa. Furthermore, we present a FLOPs-based asymptotic computational complexity analysis for the NNBF architectures alongside baseline methods, namely ZFBF and MMSE beamforming, explicitly characterizing inference-time scaling behavior. Experiments for the simple NNBF are performed under simplified assumptions such as stationary UEs and perfect CSI across varying antenna configurations in the TDL-A channel. On the other hand, transformer-based NNBF is evaluated in more realistic conditions, including urban macro environments with imperfect CSI, diverse UE mobilities, coding rates, and modulation schemes. Results show that the transformer-based NNBF achieves superior performance under realistic conditions at the cost of increased computational complexity, while the simple NNBF presents comparable or better performance than baseline methods with significantly lower complexity under simplified assumptions. Full article

Due to the critical role of channel coding, convolutional code recognition has attracted growing interest, particularly in non-cooperative communication scenarios such as spectrum surveillance. Deep learning-based approaches have emerged as promising techniques, offering improved classification performance. However, most existing works focus on single-parameter recognition and ignore the inherent correlations between code parameters. To address this, we propose a novel framework named Hybrid Multi-Task Network (HMT-Net), which adopts multi-task learning to simultaneously identify both the code rate and constraint length of convolutional codes. HMT-Net combines dilated convolutions with attention mechanisms and integrates a Transformer backbone to extract robust multi-scale sequence features. It also leverages a Channel-Wise Transformer to capture both local and global information efficiently. Meanwhile, we enhance the dataset by incorporating a comprehensive sequence dataset and further improve the recognition performance by extracting the statistical features of the sequences. Experimental results demonstrate that HMT-Net outperforms single-task models by an average recognition accuracy of 2.89%. Furthermore, HMT-Net exhibits even more remarkable performance, achieving enhancements of 4.57% in code rate recognition and 4.31% in constraint length recognition compared to other notable multi-tasking frameworks such as MAR-Net. These findings underscore the potential of HMT-Net as a robust solution for intelligent signal analysis, offering significant practical value for efficient spectrum management in next-generation communication systems. Full article

Background: There are approximately 5.4 M basal cell (BCC) and squamous cell (SCC) carcinomas diagnosed each year, and the number is increasing. Currently, the gold standard for skin cancer diagnosis is histopathology, which requires the surgical excision of the tumor followed by pathological evaluation of a tissue biopsy. The three-dimensional (3D) nature of human tissue suggests that two-dimensional (2D) cross sections may be insufficient in some cases to represent the complex structure due to sampling bias. There is a need for new techniques that can be used to classify skin lesion types and margins noninvasively. Methods: We use optical coherence tomography volume scan images and AI to noninvasively create 3D images of basal cell and squamous cell carcinomas. Results: Three-dimensional optical coherence tomography images can be broken down into a series of cross sections that can be classified as benign or cancerous using convolutional neural network models developed in this study. These models can identify cancerous regions as well as clear edges. Cancerous regions can also be verified based on visual review of the color-coded images and the loss of the green and blue subchannel pixel intensities. Conclusions: Three-dimensional optical coherence tomography cross sections of cancerous lesions can be collected noninvasively, and AI can be used to classify skin lesions and detect clear lesion edges. These images may provide a means to speed up treatment and promote better patient screening, especially in older patients who will likely develop several lesions as they age. Full article

Coded Aperture Snapshot Spectral Imaging (CASSI) is a rapid hyperspectral imaging technique with broad application prospects. Due to limitations in three-dimensional compressed data acquisition modes and hardware constraints, the compressed measurements output by actual CASSI systems have a finite dynamic range, leading to degraded hyperspectral reconstruction quality. To address this issue, a high-quality hyperspectral reconstruction method based on multi-exposure fusion is proposed. A multi-exposure data acquisition strategy is established to capture low-, medium-, and high-exposure low-dynamic-range (LDR) measurements. A multi-exposure fusion-based high-dynamic-range (HDR) CASSI measurement reconstruction network (HCNet) is designed to reconstruct physically consistent HDR measurement images. Unlike traditional HDR networks for visual enhancement, HCNet employs a multiscale feature fusion architecture and combines local–global convolutional joint attention with residual enhancement mechanisms to efficiently fuse complementary information from multiple exposures. This makes it more suitable for CASSI systems, ensuring high-fidelity reconstruction of hyperspectral data in both spatial and spectral dimensions. A multi-exposure fusion CASSI mathematical model is constructed, and a CASSI experimental system is established. Simulation and real-world experimental results demonstrate that the proposed method significantly improves hyperspectral image reconstruction quality compared to traditional single-exposure strategies, exhibiting high robustness against multi-exposure interval jitters and shot noise in practical systems. Leveraging the higher-dynamic-range target information acquired through multiple exposures, especially in HDR scenes, the method enables reconstruction with enhanced contrast in both bright and dark details and also demonstrates higher spectral correlation, validating the enhancement of CASSI reconstruction and effective measurement capability in HDR scenarios. Full article

Debit and credit card payments have become the preferred method of payment for consumers, replacing paper checks and cash. However, this shift has also led to an increase in concerns regarding identity theft and payment security. To address these challenges, it is crucial to develop an effective, secure, and reliable payment system. This research presents a comprehensive study on payment card fraud detection using deep learning techniques. The introduction highlights the significance of a strong financial system supported by a quick and secure payment system. It emphasizes the need for advanced methods to detect fraudulent activities in card transactions. The proposed methodology focuses on the conversion of a comma-separated values (CSV) dataset into quick response (QR) code images, enabling the application of deep neural networks and transfer learning. This representation enables leveraging pre-trained image-based architectures by encoding numeric transaction attributes into visual patterns suitable for convolutional neural networks. The feature extraction process involves the use of a convolutional neural network, specifically a residual network architecture. The results obtained through the under-sampling dataset balancing method revealed promising performance in terms of precision, accuracy, recall, and F1 score for the traditional models such as K-nearest neighbors (KNN), Decision Tree, Random Forest, AdaBoost, Bagging, and Gaussian Naïve Bayes. Furthermore, the proposed deep neural network model achieved high precision, indicating its effectiveness in detecting card fraud. The model also achieved high accuracy, recall, and F1 score, showcasing its superior performance compared to traditional machine learning models. In summary, this research contributes to the field of payment card fraud detection by leveraging deep learning techniques. The proposed methodology offers a sophisticated approach to detecting fraudulent activities in card payment systems, addressing the growing concerns of identity theft and payment security. By deploying the trained model in an Android application, real-time fraud detection becomes possible, further enhancing the security of card transactions. The findings of this study provide insights and avenues for future advancements in the field of payment card fraud detection. Full article

As precision agriculture places higher demands on real-time field weed detection and recognition accuracy, this paper proposes a multi-scale information-enhanced weed detection algorithm, MIE-YOLO (Multi-scale Information Enhanced), for precision agriculture. Based on the popular YOLO12 (You Only Look Once 12) model, MIE-YOLO combines edge-aware multi-scale fusion with additive gated blocks and two-stage self-distillation to boost small-object and boundary detection while staying lightweight. First, the MS-EIS (Multi-Scale-Edge Information Select) architecture is designed to effectively aggregate and select edge and texture information at different scales to enhance fine-grained feature representation. Next, the Add-CGLU (Additive-Convolutional Gated Linear Unit) pyramid network is proposed, which enhances the representational power and information transfer efficiency of multi-scale features through additive fusion and gating mechanisms. Finally, the DEC (Detail-Enhanced Convolution) detection head is introduced to enhance detail and refine the localization of small objects and fuzzy boundaries. To further improve the model’s detection accuracy and generalization performance, the DS (Double Self-Knowledge Distillation) strategy is defined to perform double self-knowledge distillation within the entire network. Experimental results on the custom Weed dataset, which contains 9257 images of eight weed categories, show that MIE-YOLO improves the F1 score by 1.9% and the mAP by 2.0%. Furthermore, it reduces computational parameters by 29.9%, FLOPs by 6.9%, and model size by 17.0%, achieving a runtime speed of 66.2 FPS. MIE-YOLO improves weed detection performance while maintaining a certain level of inference efficiency, providing an effective technical path and engineering implementation reference for intelligent field inspection and precise weed control in precision agriculture. The source code is available on GitHub. Full article

This study presents a novel framework for digital copyright management that integrates AI-enhanced perceptual hashing, blockchain technology, and digital watermarking to address critical challenges in content protection and verification. Traditional watermarking approaches typically employ content-independent metadata and rely on centralized authorities, introducing risks of tampering and operational inefficiencies. The proposed system utilizes a pre-trained convolutional neural network (CNN) to generate a robust, content-based perceptual hash value, which serves as an unforgeable watermark intrinsically linked to the image content. This hash is embedded as a QR code in the frequency domain and registered on a blockchain, ensuring tamper-proof timestamping and comprehensive traceability. The blockchain infrastructure further enables verification of multiple watermark sequences, thereby clarifying authorship attribution and modification history. Experimental results demonstrate high robustness against common image modifications, strong discriminative capabilities, and effective watermark recovery, supported by decentralized storage via the InterPlanetary File System (IPFS). The framework provides a transparent, secure, and efficient solution for digital rights management, with potential future enhancements including post-quantum cryptography integration. Full article