Search Results (7,027)

This study investigates regime-dependent forecasting of the Saudi stock market by combining macro-controlled dependence analysis with nonlinear predictive modeling. Using daily data from September 2010 to August 2025, we analyze the interaction between the Tadawul All Share Index (TASI) returns and crude oil returns while controlling for inflation and interest-rate dynamics. A four-variable VAR with macro controls is estimated separately in pre- and post-COVID regimes to characterize directional predictability and changes in transmission lags. We then evaluate out-of-sample return forecasting performance across econometric benchmarks (ARIMA, ARIMAX, and VAR) and machine learning models (LSTM and XGBoost) under a strictly time-ordered expanding-window design with sequential train/validation/test partitioning. The results indicate that traditional linear benchmarks exhibit limited predictive ability in both regimes, with negative out-of-sample explanatory power. By contrast, XGBoost delivers the strongest overall performance, achieving positive out-of-sample R² in both regimes (0.046 in pre-COVID and 0.010 in post-COVID), together with the lowest forecast errors (RMSE = 0.0081 pre-COVID; 0.0078 post-COVID). Interpretability analysis further reveals a regime-sensitive shift in drivers: short-horizon equity lag dynamics dominate during stable periods, whereas oil-related and macro-financial variables gain importance under turbulent conditions. Economic-value evaluation supports the practical relevance of these gains, showing that XGBoost-based signals yield superior risk-adjusted trading outcomes and remain favorable under downside-risk and drawdown-based assessment. Overall, these findings highlight that forecasting in oil-linked emerging markets is inherently regime-dependent and that nonlinear ensemble learners, particularly XGBoost, provide a more robust and economically meaningful approach under structural change. Full article

Autism Spectrum Disorder (ASD) is a biologically heterogeneous neurodevelopmental condition, presenting a major barrier to the identification of robust and translatable molecular biomarkers. Here, we employ a cross-species proteomic framework to identify conserved protein signatures associated with ASD. Quantitative proteomic profiling of brain and serum from CNTNAP2 knockout mice, integrated with serum proteomes from individuals with ASD, revealed 132 proteins consistently dysregulated across species. Functional pathway analyses implicated coordinated alterations in lipid metabolism, synaptic signaling, and immune regulation. To prioritize diagnostically informative candidates, we applied machine learning-based feature selection and identified a minimal panel of ten proteins (COL1A1, ITIH4, CLU, NID1, C5, MASP1, PON1, PLTP, HSPA5, and FETUB) that robustly discriminated ASD from control samples. Gene ontology and KEGG pathway analyses highlighted enrichment of immune regulatory pathways, synaptic transmission, oxidative stress responses, and lipid metabolic processes, consistent with emerging models linking neuroimmune dysregulation and metabolic imbalance to ASD pathophysiology. An XGBClassifier trained on this biomarker panel achieved strong performance in independent test sets (AUC = 0.75). Together, these findings establish cross-species proteomic integration combined with machine learning as a powerful strategy for uncovering conserved, biologically grounded biomarkers in ASD, providing a framework for future validation and translational development. Full article

Laser metal deposition operates on the principle of layer-by-layer material addition, wherein each layer is formed by overlapping individual single tracks. Consequently, clads formed serve as the fundamental building blocks for this technology. Their quality directly affects the overall build quality, particularly the geometric characteristics, which are also critical to process productivity. In the present work, geometric characteristics of TiC/Ti6Al4V single tracks fabricated via laser metal deposition are optimised. An artificial neural network model was developed to predict the clad width, height, and dilution using processing parameters, laser power, scan speed, and powder feed rate, as model inputs. The Particle Swarm Optimisation algorithm was employed for hyperparameter selection. The hyperparameter-optimised model achieved a mean squared error of 0.00183 and an R² score of 0.979 during training, and a mean squared error of 0.00709 and an R² score of 0.887 during testing. Although the small discrepancy between training and testing metrics suggests slight overfitting, likely due to the size of the dataset, the model achieved a mean absolute percentage error of less than 10% during testing. Subsequently, process plots generated by the model predictions were used to identify suitable parameters, and a processing map was developed to highlight the window that achieves suitable dilution (14–24%), defect-free sound bonding, and thick and dense clads. Full article

Background: Intrinsic capacity (IC) is key to promoting healthy aging, and managing declines in IC is crucial for delaying functional deterioration in older adults. Digital health interventions (DHIs) hold promising potential for addressing IC decline. This scoping review aims to synthesize existing evidence by mapping the types of DHIs employed and examining their effects across the five domains of IC in older adults. Methods: The review was conducted following the five-stage framework of Arksey and O’Malley and the PRISMA-ScR guideline. The search was performed across PubMed, Embase, CINAHL, Cochrane Library, PsycINFO, SinoMed, and CNKI databases for studies published between 1 January 2015 and 31 July 2025. Relevant studies were identified using MeSH terms and free-text terms related to “older adults”, “digital health”, and “intrinsic capacity”. Results: Based on the eligibility criteria, 81 studies were included. The DHIs identified encompassed virtual reality, exergames, computerized cognitive training, mHealth, internet-based interventions, telehealth, digital hearing aids, assistive robotics, and visual biofeedback. Most studies focused on single-domain interventions (74%), with cognition being the most targeted (40.7%), while sensory (4.9%) and vitality (2.5%) domains received the least attention. No digital interventions targeted all five IC domains. Regarding efficacy, many DHIs reported statistically significant improvements in one or more IC domains; however, the magnitude and consistency of these effects varied considerably across studies. Conclusions: Preliminary evidence suggests that DHIs show potential in managing declines in IC among older adults. However, evidence quality varies significantly, often derived from small-scale studies. Future research should focus on establishing clinical effectiveness through adequately powered trials and on integrating DHIs into comprehensive intervention strategies that target all domains of IC, with robust evaluation of their outcomes. Full article

This paper presents new techniques for electric machine diagnostics that combine advanced signal processing and artificial intelligence (AI)-based techniques using magnetic flux measurements acquired under various operating conditions. Developing an effective electric machine diagnostics tool is paramount for increased industrial productivity and extending the service life of the machine. The existing diagnostic tools face issues, including false indication of faults using classical methods, and the proposed data-driven methods based on machine learning lack transferability of model knowledge on an unseen dataset from different motor types or power ratings due to structural differences. To overcome these diagnostic drawbacks of statistical ML classifiers and classical approaches, innovative feature selection methods were employed in this work to preprocess the measured magnetic flux into a spectrogram image, and the transfer learning (TL) technique was applied to fine-tune convolution neural networks (CNNs) ImageNet pretrained models. The experimental results show the trained statistical ML classifiers and traditional CNN performance on unseen BU data and on the external data, and the performance demonstrated a lack of generalization on external datasets of different power ratings or structures. Models with such drawbacks cannot be used for developing effective diagnostic systems. The TL technique was employed on different deep CNN ImageNet pretrained models with spectrogram images as inputs to the deep CN network. This approach demonstrated an advanced and improved electric machine diagnostic system that addresses the drawbacks of the current ML-based diagnostic systems. The generalized model developed using CNN ResNet50 outperformed other deep CNN ImageNet models in correctly diagnosing faults on both the dataset generated from the authors’ lab and on an external dataset of a different machine from another research lab. Full article

In the field of electronic support and radar warning, it is necessary to effectively detect and recognize the modulation types of non-cooperative radar signals, especially for radars with Low Probability of Intercept (LPI) waveforms. Multiple intelligent detection and recognition algorithms based on the Transformer architecture have been proposed, achieving good performance even under low signal-to-noise ratio (SNR). However, Transformer-based radar intelligent detection and recognition algorithms have a huge number of parameters coupled with complex structures, which will result in significant power consumption and computational latency when deployed on general computing platforms. To address the above issues, this paper proposes a lightweight design for Transformer-based radar signal intelligent detection and recognition networks. A Lightweight Joint Detection and Modulation Recognition Networks (JDMR-LNet) is designed. To enhance the feature extraction ability of lightweight networks, this paper designed a hybrid model distillation method. The experimental results demonstrate that, compared with the directly trained JDMR-LNet, the accuracy of automatic modulation type recognition of the JDMR-LNet after distillation is increased by 2.37% at −12 dB, and the signal detection is increased by 2.07% at −10 dB. The number of parameters of the JDMR-LNet has also decreased significantly. Compared with the original model, the JDMR-LNet is compressed by 11.18 times. Furthermore, this paper completed FPGA deployment of the JDMR-LNet model, with simulation verifying its functional correctness. Full article

Static defense mechanisms in blockchain security, while effective against known threats, are inherently vulnerable to intelligent adversaries who can adapt their strategies to evade detection. This paper addresses this critical limitation by proposing a next-generation adaptive security framework powered by deep reinforcement learning (DRL). Building upon the state-of-the-art statistical detection system presented in Part I of this series, we introduce a DRL agent that learns to dynamically adjust security parameters in response to evolving network conditions and adversarial behavior. The agent is trained using a realistic, proxy-based reward function that optimizes for network stability without requiring ground-truth attack labels. We conduct comprehensive evaluation across multiple scenarios, demonstrating that our DRL-enhanced framework consistently renders attacks unprofitable where static models eventually fail. Against adaptive adversaries, the DRL agent drives adversary profit to $-42\pm 13\%$ (deeply unprofitable) compared to $+65\pm 22\%$ (profitable) under the static framework and $+145\pm 18\%$ under baseline detectors. Furthermore, we demonstrate resilience in zero-day scenarios where novel attack variants are suppressed within 24 h, and compare performance against alternative AI methodologies (supervised learning, GANs), achieving a superior F1-score of $0.95\pm 0.02$. This work provides a robust blueprint for creating intelligent, adaptive, and resilient security systems for future decentralized networks. Full article

With the expansion of urban rail transit, increased train operation density, and the large-scale grid integration of renewable energy such as offshore photovoltaic power, traction power supply systems face stricter requirements for operational safety, power supply reliability and energy utilization efficiency. Offshore photovoltaic power, integrated into the traction power supply network via flexible DC transmission technology, promotes renewable energy consumption, but its random and volatile output overlaps with time-varying traction loads, increasing the complexity of DC-side fault characteristics and protection control. Flexible DC technology is a core direction for next-generation traction substations, and three-level converters (key energy conversion units) have advantages over traditional two-level topologies. However, their P-O-N three-terminal DC-side topology introduces new faults (e.g., PO/ON bipolar short circuits, O-point-to-ground faults), making traditional protection strategies ineffective. In addition, wide system current fluctuation (0.5–3 kA) and offshore photovoltaic power fluctuation easily cause fixed-threshold protection maloperation, and the coupling mechanism among modulation strategies, DC bus capacitor voltage dynamics and fault current paths is unclear. To solve these bottlenecks, this paper establishes a simulation model of the system based on the PSCAD/EMTDC(A professional simulation software for electromagnetic transient analysis in power systems V4.5.3) platform, analyzes the transient electrical characteristics of three-level converters under traction and braking conditions for typical faults, clarifies the coupling mechanism, proposes a condition-adaptive fault identification strategy, and designs a reconfigurable fault energy handling system with bypass thyristors and adaptive crowbar circuits. Simulation and hardware-in-the-loop (HIL) experiments show that the proposed scheme completes fault identification and protection within 2–3 ms, suppresses fault peak current by more than 70%, limits DC bus overvoltage within ±10% of the rated voltage, and has good post-fault recovery performance. It provides a reliable and engineering-feasible protection solution for related systems and technical references for similar flexible DC system protection design. Full article

The cement industry is central to sustainable manufacturing due to its high energy demand and associated CO₂ emissions. In cement production, a substantial share of electrical energy is consumed in the clinker grinding circuit, where Blaine fineness (specific surface area, cm²/g), a key quality output, affects both cement performance and specific energy consumption. However, laboratory Blaine measurements are typically available with a 30–60 min delay, which limits timely process interventions and may promote conservative operating practices (e.g., precautionary over-grinding) to secure quality. This study develops machine-learning models to predict the finished-product Blaine fineness (Blaine-F) from routinely recorded industrial quality-control inputs, including XRF-based oxide composition, derived chemical moduli (lime saturation factor, LSF; silica modulus, SM; alumina modulus, AM), laser-diffraction particle-size distribution descriptors (Q10/Q50/Q90 corresponding to D10/D50/D90 percentile diameters; and R3 residual fractions at selected cut sizes), and intermediate in-process fineness (Blaine-P). The models were trained on over 200 finished-product samples obtained from the quality-control laboratory information management system (LIMS) of Seza Cement Factory (SYCS Group, Turkey). Ridge regression, Random Forest, XGBoost, LightGBM, and CatBoost were tuned using RandomizedSearchCV with five-fold cross-validation and evaluated on a held-out test set using MAE, RMSE, and R². The results show that the linear baseline provides limited explanatory power (Ridge: R² ≈ 0.50), consistent with the strongly non-linear behavior of the grinding–separation system, whereas tree-based ensemble methods achieve higher predictive accuracy. XGBoost yields the best overall performance (R² = 0.754; RMSE = 76.9 cm²/g), while Random Forest attains R² = 0.744 with the lowest MAE (61.7 cm²/g). Explainability analyses indicate that Blaine-F is primarily influenced by the fine-tail PSD descriptor Q10 (D10 particle size) and the intermediate fineness Blaine-P, whereas chemistry-related variables (e.g., LSF and SiO₂, and particularly SM) provide secondary yet meaningful contributions. These findings support the use of the proposed model as a virtual sensor to reduce decision latency associated with delayed laboratory Blaine measurements and to enable tighter fineness targeting. Potential energy and CO₂ implications should be quantified using site-specific, plant-calibrated relationships between kWh/t and Blaine fineness, rather than inferred as measured outcomes within the present study. Full article

Background: Sleep is essential for athletic performance, yet the specific effects of sleep deprivation are not well defined. Evidence in resistance-trained populations is limited regarding sex-specific responses and velocity-based performance across different loads. Purpose: This study examined sex differences in the impact of total (0 h) and partial (4 h) sleep deprivation versus normal sleep (8 h) on strength, power, and endurance performance in resistance-trained individuals. Methods: Twenty-four resistance-trained participants (male/female, 12/12; age: 22 ± 3 years) completed a randomized, cross-over, counterbalanced trial including one baseline control night (8 h at home sleep) and three experimental conditions in the laboratory: (a) 8 h sleep (NS), (b) 4 h sleep (ESD), (c) 0 h sleep (SD). Strength was assessed at 25%, 50%, 75%, 90% and 100% 1RM for bench press and back squat (half-squat depth, ~90° knee flexion), in a Smith machine, followed by a muscular endurance test at 65% 1RM (set-to-failure). Isometric strength and vertical jump test were also performed. Results: At 50% 1RM, significant sleep and sleep-by-sex effects were observed for V_mean in both exercise (p < 0.05, η_p² > 0.09), an effect only noted in males, with reduced performance under ESD and SD compared to NS (7–13%, p < 0.05, g > 0.50). In the muscular endurance test, sleep and sleep-by-sex effects were found (p < 0.05, η_p² < 0.22), an effect only found in females during the back squat, showing performance declines in V_mean in ESD and SD compared to NS (7–12%, p < 0.05, g > 0.2). Conclusions: Total and partial sleep deprivation impairs muscular performance differently by sex. Males experienced reduced strength at moderate loads, while females showed declines in muscular endurance. Full article

The hardware power-aware keyword spotting (KWS) implementation requires small memory footprint, low-complex computation, and high accuracy performances. Binary neural networks (BNNs) naturally satisfy these constraints. They quantize both weights and activations to 1-bit. This reduces storage and replaces most multiply–accumulate operations with bitwise operations. However, such extreme quantization incurs substantial information loss and leaves a noticeable accuracy gap relative to full-precision models. Optimization is also more difficult because the sign function is non-differentiable, and surrogate-gradient updates introduce gradient mismatch. To preserve the hardware benefits of BNNs while alleviating the accuracy degeneration induced by 1-bit quantization, this article addresses the problem from two complementary aspects: Firstly, a Stepwise Self-Knowledge Distillation (SSKD) training approach is proposed to effectively improve the student BNN’s accuracy performance. The SSKD training framework achieves effective supervision for student BNNs. A Stepwise Training Strategy is proposed to optimize the training stability and accuracy. Weight Scaling Factor improves the student’s representational capability. Secondly, an extremely lightweight Binary Temporal Convolutional ResNet (BTC-ResNet) is also proposed. The parameters and calculations inside the network are greatly reduced for the inference. Experiments on the GSCD v1 and GSCD v2 benchmarks demonstrate the effectiveness of our methods for low-power keyword spotting. For the 12-class task, BTC-ResNet14 achieves 97.23% accuracy on GSCD v1 and 97.31% on GSCD v2 with 0.75 Mb parameters and 1.35 M FLOPs. For the 35-class task on GSCD v2, it reaches 95.56% accuracy with 0.76 Mb parameters and 1.35 M FLOPs. These results indicate that our method achieves a competitive accuracy–efficiency balance relative to recent distillation-based BNN KWS baselines reported in the comparative experiments. All these studies are helpful and promising for future KWS deployment on low-power hardware devices. Full article

Reliable maximum power point tracking (MPPT) methods are essential in dynamic wind conditions to obtain maximum efficiency in wind energy conversion systems (WECSs). Conventional methods like incremental conductance (INC) and perturb and observe (P&O) are simple and robust but have drawbacks in terms of convergence and oscillations around the maximum power point (MPP) under dynamic conditions. In contrast, intelligent control methods such as artificial neural networks (ANNs) adapt more effectively. This paper presents a comparative analysis of ANN, P&O, and INC methods to obtain MPP for a WECS. A permanent magnet synchronous generator (PMSG) was coupled with a DC–DC boost converter to study the performance of the three MPPT methods under two different wind profiles. The ANN was trained with Bayesian regularization (BR) to estimate wind speed using rotor speed and computed mechanical power as inputs. The INC method achieved MPP using real-time power–voltage curves, while the P&O method perturbs the control variable, observes its results in output power, and adjusts the control variable accordingly. The three MPPT methods were compared in terms of power extraction, voltage stability, robustness, and dynamic response. The ANN achieved faster response, smoother output power and voltage, and reduced oscillation to dynamic conditions with higher output power compared to P&O and INC. On the other hand, the P&O and INC methods are less computationally intensive and do not require offline training. Full article

Background/Objectives: Acute heart failure (AHF) is a common critical condition in emergency departments, and traditional diagnostic methods have limitations, including high subjectivity and limited accuracy. This study aimed to develop an integrated machine learning model based on lung ultrasound (LUS) radiomics and clinical data for diagnosing AHF in patients presenting with acute dyspnea. Methods: A total of 301 patients were included and randomly split into training (n = 210) and testing (n = 91) sets. Using PyRadiomics 3.0, 107 radiomics features were extracted from standardized 6-zone LUS images, combined with 52 clinical features. Three random forest models were developed: clinical-only, radiomics-only, and integrated models. Results: The integrated model achieved optimal performance on the testing set with an AUC of 0.976 (95% CI: 0.950–0.994), accuracy of 90.1%, sensitivity of 91.1%, and specificity of 89.1%, significantly outperforming the radiomics model (AUC 0.940, p = 0.046) and clinical model (AUC 0.931, p = 0.111). Feature importance analysis revealed that radiomics features contributed 75.6% of the model’s predictive power, with gray level run length matrix (GLRLM) features dominating the top-ranked features. Conclusions: As a proof-of-concept study, this research demonstrates the potential value of multimodal data fusion strategies for AHF diagnosis in the emergency department; however, external validation and prospective studies are required to further confirm its clinical applicability. Full article

The style change detection (SCD) task involves identifying the locations of writing style changes in multi-authored documents. This task can be applied to plagiarism detection, security, and commerce applications. Introducing decoder-based Large Language Models (LLMs) marks a pivotal shift in applications. The segment boundaries for SCD models can be represented by concatenating two consecutive segments as pairs. However, LLMs usually restrict their input lengths, where the long-length inputs may exceed the restricted length. This paper seeks to bridge this gap and exploit the power of LLMs by introducing boundary-focused LLM Adaptation for SCD (BF-LLMA-SCD). The proposed solution adapts decoder-based LLMs for SCD using QLoRA. BF-LLMA-SCD truncates long-length input by preserving texts near an examined boundary while removing those at the other sides. BF-LLMA-SCD was trained on three PAN datasets. Comparison results with the top-performing SOTA solutions show that BF-LLMA-SCD achieved the best performance results in terms of F${}_1$ on PAN 2021 and PAN 2022/D1, while obtaining competitive results on PAN 2022/D3. BF-LLMA-SCD was also trained on an Arabic SCD dataset comprising three difficulty levels. It achieved an F${}_1$ score above 0.99 on easy instances. Full article

Employee turnover presents a significant challenge to modern organizations, often resulting in operational disruptions, substantial hiring costs, and a loss of institutional knowledge. While traditional human resource practices have historically been reactive, the emergence of machine learning has introduced a proactive capability to anticipate and mitigate attrition before it occurs. This research utilizes the IBM HR Analytics dataset, which contains 1470 employee records and 35 distinct features, to develop a hybrid machine learning model designed to enhance the accuracy of turnover predictions. To ensure the model’s effectiveness, the researchers employed a comprehensive preprocessing phase that included eliminating non-informative features, applying label encoding to categorical data, and using StandardScaler to normalize quantitative values. A critical component of the study addressed the common issue of class imbalance within HR data. To resolve this, a hybrid sampling strategy was implemented, combining Synthetic Minority Over-sampling Technique (SMOTE) and Adaptive Synthetic Sampling (ADASYN) to create a more balanced learning environment for the algorithms. The core of the predictive engine is a soft voting ensemble that integrates three powerful algorithms: Random Forest, XGBoost, and logistic regression. Evaluated on an 80/20 train–test split, the tuned XGBoost model achieved an impressive 84% accuracy and an Area Under the Curve (AUC) of 0.80. Meanwhile, the logistic regression component contributed the highest F1-score, reinforcing the overall strength and balance of the ensemble approach. These metrics confirm that the hybrid model is both robust and reliable for identifying at-risk employees. Beyond simple prediction, the study prioritized interpretability by using SHapley Additive exPlanations (SHAP) to identify the primary drivers of attrition. The analysis revealed that the most significant variables influencing an employee’s decision to leave include the interaction between job level and experience, frequent overtime, monthly income, current job level, and total years spent at the company. By providing these data-driven insights, the model empowers HR teams to transition from reactive troubleshooting to proactive retention planning, ultimately securing the organization’s talent and stability. Full article