Search Results (1,415)

Reproducing SM4 (GB/T 32907-2016) hardware-accelerator results on open-source RISC-V platforms is difficult, because most published designs depend on proprietary FPGA toolchains. This paper contributes an asymmetric dual-channel SM4 architecture together with a fully reproducible open-source verification framework; physical on-board acceleration is not claimed and is left as future work. The architecture exploits two algorithmic symmetries of SM4—encryption and decryption differ only in round-key order, and the round transform T shares the byte-wise S-box $\tau$ with the key-expansion transform $T^{\prime }$—but maps them onto an asymmetric workload. Bulk encryption is throughput-bound, whereas key expansion runs once per session. Accordingly, a 32-stage fully unrolled encryption pipeline (one 128-bit block per cycle in steady state) is paired with a single round function reused iteratively for the key schedule, and encryption and decryption share one datapath via round-key reversal. Because the TH1520 SoC on LicheePi 4A does not expose the Xuantie C910 RoCC port, we verify the design in three reproducible tiers on the board itself: (T1) RTL co-simulation of an sm4_rocc wrapper passes 1040/1040 vectors for both the standalone datapath and the full system. (T2) A pure-C reference model passes 10/10 GB/T 32907-2016 vectors on the real C910 at a measured 291.9 Mbps. (T3) A Linux illegal-instruction trap-and-emulate prototype confirms ISA and OS-level semantics. Open-source synthesis (Yosys + SkyWater Sky130) gives a measured area of 133 kGE and a switching-dominated post-synthesis power estimate of ≈0.28 W at 100 MHz (≈22 pJ/bit, ≈46 Gbps/W). At 100 MHz the unrolled pipeline reaches an RTL simulation-equivalent steady-state throughput of 12.8 Gbps, about 43.9× the software baseline. Every reported number is reproducible with open-source tools only (Icarus Verilog, GTKWave, GCC, Yosys, Sky130 PDK). Full article

This paper proposes an intelligent time-series early warning framework based on a production LSTM–Transformer network for petrochemical refining processes. A cascaded encoder–decoder architecture is designed, where the LSTM extracts local temporal patterns and medium-term memory from noisy industrial data, while the Transformer models global dependencies and cross-unit interactions via multi-head self-attention. An adaptive feature fusion layer bridges the representational gap between the two networks. A multi-stage preprocessing pipeline tailored for refining MES data handles missing values, outliers, and mixed operating conditions. Using 120 variables from five units of a fluid catalytic cracking unit, the framework predicts the regenerator bed temperature up to 8 h (48 steps) ahead. Comparative experiments show that the production LSTM–Transformer achieves a mean MAE of 0.088, a mean RMSE of 0.113, and the lowest median MAPE of 19.91% among all models, outperforming standalone LSTM (MAE 0.095, MAPE 20.85%) and Transformer (MAE 0.088, MAPE 20.49%). Robustness analysis confirms stable performance under strong noise (down to 5 dB) and missing rates up to 50%, with a median MAE of 0.1027 across tags. This work provides an effective, end-to-end predictive early warning solution that balances accuracy, production importance coverage, and industrial robustness, offering a generalizable data-driven paradigm for process industries. Full article

In bank-dominated financial systems, the accumulation of non-performing loans is a recognised source of systemic vulnerability, as correlated corporate defaults can erode bank capital, impair liquidity, and propagate stress across interconnected portfolios. Firm-level default detection thus constitutes a microprudential foundation of macroprudential stability: the reliable early identification of risky borrowers reduces both individual credit losses and the aggregate exposures that drive system-level fragility. Yet the use of structured data-mining pipelines for this task remains underexplored in Central and Eastern Europe. This study applies the CRISP-DM methodology to predict corporate loan default using data on 302 Slovak corporate borrowers, combining financial ratios from publicly available financial statements with selected company and loan-related information from internal bank records. Seven individual classifiers were developed and compared: decision trees (CART, CHAID, C5.0), logistic regression, discriminant analysis, and neural networks (MLP, RBF), together with a stacked ensemble based on their outputs. Model performance was evaluated using sensitivity, overall classification accuracy, and area under the ROC curve (AUC), with sensitivity treated as the primary criterion because of the asymmetric costs of misclassification in credit risk assessment. The results confirm that historical firm-level information provides a reliable basis for default prediction, with tree-based models consistently outperforming statistical and neural network approaches. The stacked ensemble achieved the strongest overall performance, whereas C5.0 and CHAID showed that interpretable classifiers can also deliver competitive predictive accuracy. A champion–challenger deployment architecture is proposed, in which the ensemble serves as the performance-oriented champion and interpretable models act as challengers; this arrangement contributes to the operational resilience of the credit-risk assessment process and aligns with macroprudential expectations of model governance, auditability, and explainability. The study offers a replicable methodological framework for integrating data-driven decision support into credit evaluation in comparable banking settings. Full article

Deep learning models for power quality (PQ) disturbance classification remain critically vulnerable to adversarial perturbations, with classification performance degrading severely under white-box attacks. Existing defenses address individual models in isolation and provide no mechanism for operators to assess whether the system is under attack or which classifier remains trustworthy. This paper proposes a two-stage framework that combines adversarial training with large language model (LLM) reasoning to improve both robustness and interpretability. In the first stage, four architecturally diverse classifiers, including a proposed Multi-Scale Temporal Attention Network (MSTAN), are evaluated under four adversarial attacks (FGSM, PGD, C&W, and UAP), and their failure patterns are recorded as structured vulnerability fingerprints. The ensemble is then retrained via adversarial training on mixed clean and perturbed signals. In the second stage, an LLM analyzes the ensemble predictions alongside the fingerprint knowledge base to perform attack detection, fingerprint-guided meta-classification, and operator-facing threat report generation. On a 17-class, 255,000-signal synthetic benchmark, adversarial training recovers FGSM and PGD accuracy from below 25% to the 53–78% range, with MSTAN achieving the highest post-training robustness (78.26% under FGSM, 65.41% under PGD). The LLM reasoning layer provides an additional 3.5–6.2 percentage point improvement over majority voting by selecting the most reliable ensemble member based on the inferred attack condition, and detects adversarial attacks with 87.6% overall accuracy. To our knowledge, this is the first integration of LLM-based ensemble reasoning into the PQ adversarial robustness pipeline and the first application of the C&W optimization attack to power quality signals. Full article

Ceramic additive manufacturing offers strong potential for fabricating geometrically complex and application-specific components, yet achieving reliable dimensional fidelity remains challenging because dimensional deviation is governed by highly coupled material, process, thermal, and environmental factors. To address this problem, this study proposes an uncertainty-aware computational framework for dimensional error prediction in ceramic 3D printing under variable material and process conditions. The contribution is positioned as a system-level integration of established learning, uncertainty estimation, calibration, and reliability-interpretation components within a ceramic additive manufacturing dimensional-error prediction workflow, rather than as a fundamental methodological breakthrough. The validation is conducted using the publicly available Ceramic 3D Printing Process Control Dataset, a 1000-sample tabular dataset, and the resulting findings are therefore interpreted as dataset-specific computational evidence rather than direct proof of industrial deployment readiness. The methodology begins with a structured data-driven preprocessing pipeline that transforms the Ceramic 3D Printing Process Control Dataset into a multi-condition feature space through data cleaning, one-hot material encoding, min–max normalization, and engineered descriptors capturing extrusion–speed balance, thermal gradients, cooling intensity, deposition density, and material-conditioned interactions. A multi-branch deep computational architecture is then developed to encode material, process, thermal-environmental, and engineered-feature streams separately, followed by adaptive cross-condition fusion to learn nonlinear dependencies across ceramic printing regimes. To improve reliability beyond deterministic regression, the framework jointly models aleatoric and epistemic uncertainty and incorporates calibration refinement to align predictive confidence with observed error behavior, thereby enabling preliminary reliability-oriented interpretation of stable and high-risk operating conditions. Experimental results demonstrate that the full model achieves the best overall within-dataset performance, with a test MAE of 0.0118, RMSE of 0.0172, $R^2=0.999$, MAPE of 1.74%, calibration error of 0.003, PICP of 0.996, reliability score of 0.992, and a stable prediction rate of 98.7%. Although these values indicate strong predictive behavior under the current structured dataset, the exceptionally high $R^2$ should be interpreted cautiously because external experimental validation, larger measured datasets, and cross-machine ceramic printing trials are still required. These findings show that the proposed framework provides an effective system-level computational strategy for dataset-specific reliability-aware dimensional quality prediction in ceramic additive manufacturing and offers a preliminary data-driven foundation for uncertainty-aware intelligent process optimization. Full article

Suspended ceiling systems are among the most seismically vulnerable non-structural components in buildings, posing significant life-safety risks and economic losses, yet understanding their full-field kinematic behavior under seismic loading remains a major experimental challenge. Conventional contact sensors offer limited spatial coverage and can alter the dynamic properties of lightweight panels due to mass loading. In contrast, non-contact optical alternatives are rarely feasible in shake-table environments due to restricted viewing angles, extensive areal coverage requirements, and the risk of equipment damage from falling panels. This study proposes an end-to-end three-dimensional displacement measurement framework for large-scale shake-table testing of suspended ceiling systems, employing consumer-grade cameras with purpose-built tools that cover the complete experimental workflow, including motion-based video trimming, semi-automated calibration, a robust multi-stage image-tracking pipeline that maintains trajectory continuity under extreme inter-frame displacements, and a ceiling system motion visualization and analysis tool. The framework was validated through a full-scale shake-table experiment continuously tracking 324 spatial nodes across 81 ceiling panels, achieving an RMSE below 3 mm in all spatial directions and exact peak-frequency agreement in 9 out of 10 test cases. A parallel processing architecture reduced total processing time from over 27 h to under 10 min without GPU acceleration, and six-degree-of-freedom rigid-body analysis resolved the complete panel failure sequence from constrained oscillation through multi-axis rotation to gravitational free fall, a level of kinematic detail unattainable with conventional instrumentation. This framework establishes a practical, scalable foundation for full-field seismic performance assessment of non-structural systems where conventional instrumentation is physically or logistically infeasible. Full article

Contemporary buildings and urban spaces are increasingly expected to support psychological well-being—a quality often termed “restorativeness.” Conventional approaches to quantifying restorativeness rely on subjective surveys or coarse green metrics, failing to capture how specific building morphologies and street-level visual configurations shape restorative experiences, particularly for stress-prone groups such as young adults. This study develops a deep-learning-driven framework linking building visual elements to youth-specific perceived restorativeness, using Changsha, China, as a testbed. The framework comprises three AI-powered modules: the TrueSkill algorithm trains a deep learning model to predict six dimensions of youth perception (e.g., beautiful, clean, safe) from pairwise comparisons of street view images; the Mask2Former architecture segments street-level imagery into 18 building and street attributes; and the XGBoost-SHAP pipeline uncovers nonlinear associations and threshold-like patterns between these attributes and the composite Built Environment Restorativeness Index (BERI). Results reveal three key insights: tree coverage shows a sustained positive association without saturation; building density exhibits a weakening association at high levels, suggesting possible saturation; and road proportion follows a bidirectional pattern, shifting from negative to positive beyond a certain range. Spatially, high BERI zones concentrate where ecological assets and diverse building functions co-occur, while youth perception exhibits systematic mismatches (e.g., “beautiful but not clean,” “safe but not lively”), traceable to imbalances in building form, street furniture, and commercial mix. These findings advance AI-assisted evaluation of built environments by shifting from one-dimensional metrics to interpretable, design-relevant diagnostics, offering a replicable evidence base for crafting youth-responsive buildings and streets. Full article

Foggy traffic scenes pose significant challenges for object detection because reduced contrast, blurred object boundaries, and the loss of local details weaken discriminative feature representations. These degradations are particularly detrimental to lightweight detectors used in intelligent transportation and vehicle perception systems, where both accuracy and real-time efficiency are required. To address this problem, this paper proposes YOLO-GCM, a lightweight detector-side feature enhancement framework built upon YOLO11n. Instead of relying on an external image dehazing stage, YOLO-GCM improves the internal feature representation of the detector through three complementary modules: a gated additive feature block (GAFB) for adaptive channel-wise feature selection and noise suppression, a context-aware feature enhancement module (CAFEM) for strengthening high-level semantic context, and a multi-scale adaptive fusion (MSAF) module for enhancing cross-scale feature interaction. By integrating these modules into a unified one-stage detector, the proposed method improves detection robustness under low-visibility traffic conditions while maintaining a compact architecture. Experiments on the FoggyCar dataset show that YOLO-GCM achieved 89.81% mAP@0.5 and 67.99% mAP@0.5:0.95, outperforming standard YOLO baselines and dehazing-assisted detection pipelines under a consistent evaluation protocol. Additional evaluation on Foggy Cityscapes further verified the generalization capability of the proposed method under domain shift. The results demonstrate that detector-side feature enhancement provides an effective and efficient alternative to multi-stage dehazing-plus-detection pipelines for foggy traffic object detection. These findings can provide useful guidance for the development of robust and efficient perception modules in roadside monitoring, intelligent transportation systems, and vehicle-assisted driving applications under adverse weather conditions. Full article

Visually impaired individuals face significant challenges in autonomous mobility and situational awareness. Most existing assistive technologies address isolated tasks, such as object recognition or text reading, while failing to capture broader environmental context. This work addresses this limitation by proposing a scene-sensitive, low-cost assistive system that delivers holistic situational information. We present Vocal-Eyes, an intelligent smart glasses platform that provides periodic audio descriptions of the surrounding environment. The system employs a cloud-based neural processing pipeline in which visual features are extracted using a Transformer-based architecture. Relational context is modeled through scene graph generation, and scene graphs are translated into natural language via a graph-to-text module. A lightweight hardware prototype captures visual data locally, while computationally intensive processing is offloaded to the cloud to reduce power consumption. The experimental results show that relational, scene-based narration produces more coherent and informative descriptions than object-centric approaches while maintaining acceptable periodic latency. Cost analysis further indicates that Vocal-Eyes is significantly more affordable than comparable commercial smart glasses solutions. These results demonstrate that Transformer-based scene understanding with cloud-assisted processing is an effective and practical approach for developing accessible, context-aware assistive technologies for visually impaired users. Full article

Adversarial camouflage has attracted growing research attention owing to its ability to execute multi-view, persistent attacks in real physical environments, outperforming conventional single-view adversarial patches. However, most existing methods are confined to non-targeted attacks, which induce arbitrary incorrect detection results without specifying target categories. This ambiguity weakens attack destructiveness and stealthiness, posing limitations for security evaluation of real-world vision systems. To address this gap, we present TACT, an approach built upon the full-coverage physical camouflage pipeline. By replacing the original category supervision with a predefined target class, TACT redirects the optimization gradient to guide 3D texture toward the target category features. Such a scheme only employs the inherent feature alignment mechanism of off-the-shelf object detectors, without redesigning network modules, defining novel loss functions, or modifying the rendering pipeline. Extensive experiments across digital and physical domains validate its effectiveness: on seven mainstream general-purpose object detectors, TACT-person achieves an average targeted attack success rate of 51.91%, and delivers cross-architecture and cross-version transferability. In physical tests, TACT-bird reduces mAP50-95 by 59.87% on YOLOv8, yet a TCER–TASR gap suggests that the physical pipeline acts as a low-pass filter: coarse-grained target classes transfer robustly while fine-grained ones suffer feature collapse. These results confirm the viability of native supervision redirection and reveal an empirical pattern: coarse-grained target classes transfer more robustly through the physical pipeline than fine-grained ones, suggesting that target class feature granularity consistently influences physical-domain attack effectiveness. Full article

Significant progress in legal natural language processing (NLP) has enabled advancements in tasks such as legal judgment prediction, case retrieval, and question answering. However, the development of analogous technologies for Arabic legal texts remains severely constrained by the scarcity of large-scale, publicly available benchmarks for summarisation and classification. This paper addresses this gap by introducing a novel, comprehensive dataset of 9699 Arabic legal cases sourced from the Saudi Board of Grievances. This corpus is unique in pairing full-length court decisions with expertly human-crafted abstractive summaries and multi-class category labels (Administrative, Commercial, and Criminal), establishing a dedicated benchmark for Arabic legal NLP. The dataset was constructed via a robust, reproducible pipeline that ensures high textual fidelity, incorporating specialised optical character recognition (OCR) via Google Document AI and precise structural segmentation into facts, reasons, and summaries. To establish robust baselines, we conduct an extensive empirical evaluation of seven summarisation models—encompassing four extractive algorithms (TextRank, LexRank, Latent Semantic Analysis, and Luhn) and three transformer-based abstractive architectures (AraT5v2, AraBART, and mBART)—each evaluated in both base and fine-tuned configurations. Results across ROUGE, BERTScore, BLEU metrics and human evaluation demonstrate substantial performance gains achieved through domain-specific fine-tuning, with the fine-tuned AraBART model achieving the strongest performance among all evaluated models. Furthermore, we present a novel analysis of the downstream utility of generated summaries by evaluating their performance on legal category classification using five machine learning models. This investigation reveals a strong positive correlation between summarisation quality and classification accuracy, empirically demonstrating that domain-adapted abstractive summarisation not only enhances intrinsic evaluation scores but also significantly boosts extrinsic task performance. By providing this essential dataset and comprehensive benchmarking, our work contributes a much-needed resource to the field, facilitating future research and innovations in Arabic legal text analysis. Full article

The digitization of existing building stock often depends on legacy 2D raster floor plans (scanned drawings, PDF exports, or photographs) because structured building information models are frequently unavailable for older properties. Manual measurement and visual inspection of such documents are time consuming and error prone. This paper presents an integrated deep learning pipeline that extracts semantic information from unstructured two-dimensional floor plan images of existing structures and supports preliminary compliance screening via locally deployed large language models. The pipeline employs YOLOv8 for the localization and classification of 18 architectural symbols and furniture items, and a U-Net with a ResNet34 encoder for the semantic segmentation of walls and interior room spaces. To translate pixel-level predictions into physical metrics, we implement an area calculation module based on user-defined reference scale calibration. An LLM evaluation module, deployed locally via Ollama with a retrieval-augmented generation pipeline, interprets extracted room metrics and flags potential non-compliance against referenced residential design guidelines; it is intended for the assessment of existing layouts rather than generative co-design. We expand a core dataset of 101 manually annotated source floor plans to 303 augmented instances using label-aligned geometric transformations, while reporting generalization in terms of the 101 unique source plans. On the held-out validation split (10 source plans), YOLOv8 achieves 92.3% mAP50 versus 87.2% for a Faster R-CNN reference model on the same data split (detection baselines differ in training epochs and pretraining; see Experiments); U-Net achieves 95.71% mIoU, surpassing DeepLabv3+ (93.2%) under matched segmentation training settings. The system is deployed as an interactive web application for legacy building survey and preliminary regulatory review when only two-dimensional documentation is available. Full article

The growing complexity of web attacks highlights the need for adaptive, intelligent defense systems that overcome the limitations of traditional rule-based web security. Thus, the architecture proposed in this paper integrates data-driven deep learning with deterministic rule-based logic to enhance real-time detection accuracy and adaptability in dynamic web threat environments. The practical integration of a deep learning-based Gated Recurrent Unit (GRU) model with ModSecurity, an open-source Web Application Firewall (WAF), is employed to improve the detection and classification of malicious HTTP requests. The model, pre-trained on a large labeled up-to-date dataset of web traffic and attack types collected post-2020, is designed to classify requests in real-time, identifying both whether a request is malicious and the corresponding attack category (e.g., SQL Injection, Cross-Site Scripting, Command Injection). We demonstrate how the trained model is incorporated into ModSecurity’s inspection pipeline, allowing it to analyze real-time web traffic alongside traditional rule-based inspection. This hybrid approach aims to significantly reduce false positives and improve adaptability to new attack patterns. Evaluation metrics such as accuracy, receiver operating characteristic (ROC), area under the curve (AUC), Principal Component Analysis (PCA), confusion matrix, and t-Distributed Stochastic Neighbor Embedding (t-SNE) visualization are discussed, along with performance considerations and implementation architecture. The integration presents a robust framework for ML-improved intelligent web security defense. Full article

Malware classification increasingly relies on machine learning models that combine static and dynamic evidence, yet their practical use is often limited by dataset inconsistency, high-dimensional feature spaces, and insufficient transparency. This paper presents an explainable hybrid malware-classification pipeline built on an aligned public dataset in which static and dynamic features are matched at sample level and share the same class space. The framework combines a Random Forest static branch, a calibrated XGBoost dynamic branch, and a weighted late-fusion stage whose branch weights are derived from inner-validation weighted-F1 rather than from test performance. On the corrected no-leak benchmark, static reduction compresses the static space from 771 to 258 features, while sparse-aggressive reduction compresses the dynamic space from 21,918 to 374 features. An early-fusion XGBoost baseline achieves the best multiclass aggregate scores, whereas the validation-weighted calibrated hybrid provides the strongest false-negative-first Benign vs. Malware profile, reaching malware recall 0.9998, benign recall 0.8053, and one false negative on the test set. The study shows that, once leakage is removed and fusion is validation-driven, the preferred hybrid architecture depends on the operational objective rather than on a single aggregate metric. Full article

Unmanned aerial vehicles (UAVs) and deep learning (DL) have introduced a new framework for intelligent building façade defect detection, yet existing studies often focus on isolated technical components and lack a systematic evaluation of the entire pipeline. To address this gap, this paper conducts a systematic literature review of 135 peer-reviewed journal articles retrieved from the Web of Science database over the period 2021–2026. This review investigates four key domains: (1) UAV inspection path planning and data acquisition; (2) multi-modal data fusion; (3) DL-driven defect detection algorithms; and (4) 3D reconstruction and digital twin integration. Our analysis reveals the following main findings. Real-time perception-aware planning is central to UAV path planning, yet most studies lack robustness evaluations under real-world deployment conditions. Multi-modal data fusion improves detection across multiple defect types, yet edge deployment requires balancing lightweight design with recognition stability. Defect recognition algorithms increasingly adopt task-driven architectures, but limited edge-device resources demand joint optimization of efficiency and accuracy. In digital twins, systematic research is still lacking on semantically integrating recognition results into BIM for O&M decision-making, leaving the closed loop from defect detection to maintenance unresolved. This review aims to help researchers and practitioners advance UAV-based inspection from an auxiliary tool to a fully autonomous, reliable intelligent agent for refined management of the urban built environment. Full article