Search Results (8,418)

Curling presents a challenging continuous-control problem in which shot outcomes depend on long-horizon interactions between complex physical dynamics, strategic intent, and opponent responses. Despite recent progress in applying reinforcement learning (RL) to games and sports, curling lacks a unified environment that jointly supports stable, rule-consistent simulation, structured state abstraction, and scalable agent training. To address this gap, we introduce a comprehensive learning framework for curling AI, consisting of a full-sized simulation environment, a task-aligned Markov decision process (MDP) formulation, and a two-phase training strategy designed for stable long-horizon optimization. First, we propose a novel MDP formulation that incorporates stone configuration, game context, and dynamic scoring factors, enabling an RL agent to reason simultaneously about physical feasibility and strategic desirability. Second, we present a two-phase curriculum learning procedure that significantly improves sample efficiency: Phase 1 trains the agent to master delivery mechanics by rewarding accurate placement around the tee line, while Phase 2 transitions to strategic learning with score-based rewards that encourage offensive and defensive planning. This staged training stabilizes policy learning and reduces the difficulty of direct exploration in the full curling action space. We integrate this MDP and training procedure into a unified Curling RL Framework, built upon a custom simulator designed for stability, reproducibility, and efficient RL training and a self-play mechanism tailored for strategic decision-making. Agent policies are optimized using Soft Actor–Critic (SAC), an entropy-regularized off-policy algorithm designed for continuous control. As a case study, we compare the learned agent’s shot patterns with elite match records from the men’s division of the Le Gruyère AOP European Curling Championships 2023, using 6512 extracted shot images. Experimental results demonstrate that the proposed framework learns diverse, human-like curling shots and outperforms ablated variants across both learning curves and head-to-head evaluations. Beyond curling, our framework provides a principled template for developing RL agents in physics-driven, strategy-intensive sports environments. Full article

Accurate spatiotemporal forecasting underpins high-stakes decision making in smart urban systems, from traffic control and energy scheduling to environment monitoring. Yet two persistent gaps limit current models: (i) spatial modules are often biased toward low-pass smoothing and struggle to reconcile slow global trends with sharp local dynamics; and (ii) the graph structure required for forecasting is frequently latent, while learned graphs can be unstable when built from temporally derived node features alone. We propose SpeQNet, a query-enhanced spectral graph filtering framework that jointly strengthens node representations and graph construction while enabling frequency-selective spatial reasoning. SpeQNet injects global spatial context into temporal embeddings via lightweight learnable spatiotemporal queries, learns a task-oriented adaptive adjacency matrix, and refines node features with an enhanced ChebNetII-based spectral filtering block equipped with channel-wise recalibration and nonlinear refinement. Across twelve real-world benchmarks spanning traffic, electricity, solar power, and weather, SpeQNet achieves state-of-the-art performance and delivers consistent gains on large-scale graphs. Beyond accuracy, SpeQNet is interpretable and robust: the learned spectral operators exhibit a consistent band-stop-like frequency shaping behavior, and performance remains stable across a wide range of Chebyshev polynomial orders. These results suggest that query-enhanced spatiotemporal representation learning and adaptive spectral filtering form a complementary and effective foundation for effective spatiotemporal forecasting. Full article

With the rapid advancement of smart grids driven by renewable energy integration and the extensive deployment of supervisory control and data acquisition (SCADA) and phasor measurement units (PMUs), addressing the escalating alarm flooding via intelligent analysis of large-scale alarm data is pivotal to safeguarding the safe and stable operation of power grids. To tackle these challenges, this study introduces a pioneering alarm optimization framework based on symmetry-driven crowdsourced active learning and interpretable deep reinforcement learning (DRL). Firstly, an anomaly alarm annotation method integrating differentiated crowdsourcing and active learning is proposed to mitigate the inherent asymmetry in data distribution. Secondly, a symmetrically structured DRL-based hierarchical attention deep Q-network is designed with a dual-path encoder to balance the processing of multi-scale alarm features. Finally, a SHAP-driven interpretability framework is established, providing global and local attribution to enhance decision transparency. Experimental results on a real-world power alarm dataset demonstrate that the proposed method achieves a Fleiss’ Kappa of 0.82 in annotation consistency and an F1-Score of 0.95 in detection performance, significantly outperforming state-of-the-art baselines. Additionally, the false positive rate is reduced to 0.04, verifying the framework’s effectiveness in suppressing alarm flooding while maintaining high recall. Full article

Eutrophication remains a persistent challenge in the Romanian Black Sea coastal zone, driven by excess nutrient inputs from riverine and coastal sources and further intensified by climate change. This study assesses eutrophication dynamics and explores mitigation options using an integrated framework that combines in situ observations, satellite-derived chlorophyll a data, machine learning, and system dynamics modelling. Water samples collected during two field campaigns (2023–2024) were analyzed for nutrient concentrations and linked with chlorophyll a products from the Copernicus Marine Service. Random Forest analysis identified dissolved inorganic nitrogen, phosphate, salinity, and temperature as the most influential predictors of chlorophyll a distribution. A system dynamics model was subsequently used to explore relative ecosystem responses under multiple management scenarios, including nutrient reduction, enhanced zooplankton grazing, and combined interventions. Scenario-based simulations indicate that nutrient reduction alone produces a moderate decrease in chlorophyll a (45% relative to baseline conditions), while restoration of grazing pressure yields a comparable response. The strongest reduction is achieved under the combined scenario, which integrates nutrient reduction with biological control and lowers normalized chlorophyll a levels by approximately two thirds (71%) relative to baseline. In contrast, a bloom-favourable scenario results in a several-fold increase in chlorophyll a of 160%. Spatial analysis highlights persistent eutrophication hotspots near the Danube mouths and urban discharge areas. These results demonstrate that integrated strategies combining nutrient source control with ecological restoration are substantially more effective than single-measure interventions. The proposed framework provides a scenario-based decision-support tool for ecosystem-based management and supports progress toward achieving Good Environmental Status under the Marine Strategy Framework Directive. Full article

Background/Objectives: Parkinsonian gait is characterized by impaired motor adaptability, altered temporal organization, and reduced movement variability. While data augmentation is commonly used to mitigate class imbalance in gait-based machine learning models, conventional strategies often ignore physiological differences between healthy and pathological movements, potentially distorting meaningful motor dynamics. This study explores whether preserving healthy motor variability while selectively augmenting pathological gait signals can improve the robustness and physiological coherence of gait pattern classification models. Methods: Eight patients with Parkinsonian gait patterns and forty-eight healthy participants performed walking tasks on the Motigravity platform under hypogravity conditions. Full-body kinematic data were acquired using wearable inertial sensors. A selective augmentation strategy based on smooth time-warping was applied exclusively to pathological gait segments (×5, σ = 0.2), while healthy gait signals were left unaltered to preserve natural motor variability. Model performance was evaluated using a hybrid convolutional neural network–long short-term memory (CNN–LSTM) architecture across multiple augmentation configurations. Results: Selective augmentation of pathological gait signals achieved the highest classification performance (94.1% accuracy, AUC = 0.97), with balanced sensitivity (93.8%) and specificity (94.3%). Performance decreased when augmentation exceeded an optimal range of variability, suggesting that beneficial augmentation is constrained by physiologically plausible temporal dynamics. Conclusions: These findings demonstrate that physiology-informed, selective data augmentation can improve gait pattern classification under constrained data conditions. Rather than supporting disease-specific diagnosis, this proof-of-concept study highlights the importance of respecting intrinsic differences in motor variability when designing augmentation strategies for clinical gait analysis. Future studies incorporating disease-control cohorts and subject-independent validation are required to assess specificity and clinical generalizability. Full article

Background: Artificial intelligence (AI) in urban and multimodal transport has demonstrated strong potential; however, real-world deployment remains constrained by limited governance-ready design, fragmented data ecosystems, and single-objective optimization practices. The resulting problem is that agencies lack a reproducible, deployment-ready architecture that links data fusion, multi-objective optimization, and electrification constraints into daily multimodal operational decision making. Methods: This study presents a systematic review and synthesis of 145 peer-reviewed studies on network control, green routing, digital twins, and electric-bus scheduling, conducted in accordance with PRISMA 2020 using predefined inclusion and exclusion criteria. Based on these findings, a deployment-oriented operational AI framework is developed. Results: The proposed architecture comprises five interoperable layers—data ingestion, streaming analytics, optimization services, decision evaluation, and governance monitoring—supporting scalability, reproducibility, and transparency. Rather than producing a single optimal solution, the framework provides decision-ready trade-offs across service quality, cost efficiency, and sustainability while accounting for uncertainty, reliability, and electrification constraints. The approach is solver-agnostic, supporting evolutionary and learning-based techniques. Conclusions: A Thessaloniki-based multimodal case study demonstrates how reproducible AI workflows can connect real-time data streams, optimization, and institutional decision making for continuous multimodal transport management under operational constraints. Full article

Wireless Body Area Networks (WBANs) enable real-time data collection for medical monitoring, sports tracking, and environmental sensing, driven by Internet of Things advancements. Their layered architecture supports efficient sensing, aggregation, and analysis, but energy constraints from transmission (over 60% of consumption), idle listening, and dynamic conditions like body motion hinder adoption. Challenges include minimizing energy waste while ensuring data reliability, Quality of Service (QoS), and adaptation to channel variations, alongside algorithm complexity and privacy concerns. This paper reviews recent power control mechanisms in WBANs, encompassing feedback control, dynamic and convex optimization, graph theory-based path optimization, game theory, reinforcement learning, deep reinforcement learning, hybrid frameworks, and emerging architectures such as federated learning and cell-free massive MIMO, adopting a systematic review approach with a focus on healthcare and IoT application scenarios. Achieving energy savings ranging from 6% (simple feedback control) to 50% (hybrid frameworks with emerging architectures), depending on method complexity and application scenario, with prolonged network lifetime and improved reliability while preserving QoS requirements in healthcare and IoT applications. Full article

This paper analyzes different approaches for the mathematical modeling and optimization of process parameters in the hard turning process of 42CrMo4 steel using a hybrid approach combining response surface methodology (RSM), multi-criteria decision making (MCDM), and machine learning through, support vector regression (SVR) with one-factor-at-a-time (OFAT) sensitivity analysis. Controlled process parameters such as cutting speed, depth of cut, feed, and insert radius are applied to conduct the experiments based on a full factorial experimental design. RSM was used to develop models that describe the effect of controlled parameters on surface roughness and cutting forces. Special emphasis was placed on the analysis of standardized residuals to evaluate the predictive capabilities of the RSM-developed model on an unseen data set. For all four outputs considered, analysis of the standardized residuals shows that over 97% of the points lie within ±3 standard deviations. A multi-criteria optimization technique was applied to establish an optimal combination of input parameters. The SVR model had high performance for all outputs, with coefficient of determination values between 89.91% and 99.39%, except for surface roughness on the test set, with a value of 9.92%. While the SVR model achieved high predictive accuracy for cutting forces, its limited generalization capability for surface roughness highlights the higher complexity and stochastic nature of surface formation mechanisms in the turning process. OFAT analysis showed that feed rate and depth of cut have been shown to be the most important input variables for all analyzed outputs. Full article

Despite the ecological sensitivity of the Yangtze River Source Region (YRSR), the current research critically lacks a quantified assessment of the spatial occurrence probability of Retrogressive Thaw Slumps (RTSs) in this specific high-altitude terrain. This study aims to bridge this knowledge gap by establishing a robust susceptibility assessment framework to accurately model the spatial distribution and risk levels of RTSs. The innovations of this research include (i) the construction of a complete and up-to-date 2024 RTS inventory for the entire YRSR based on high-resolution optical remote sensing; (ii) the integration of time-series spectral features (e.g., vegetation and moisture trends) alongside static topographic variables to enhance the physical interpretability of machine learning models; and (iii) the application of advanced ensemble learning algorithms combined with SHAP analysis to establish a comprehensive RTS susceptibility zonation. The results reveal a rapid intensification of instability, evidenced by an 83.5% surge in RTS abundance, with the CatBoost model achieving exceptional accuracy (AUC = 0.994), and identifying that specific static topographic factors (particularly elevations between 4693 and 4812 m and north-to-east aspect) and dynamic spectral anomalies (indicated by declining vegetation vigor and increasing surface wetness) are the dominant drivers controlling RTS distribution. This study provides essential baseline data and spatial guidance for ecological conservation and engineering maintenance in the Asian Water Tower, demonstrating a highly effective paradigm for monitoring permafrost hazards under climate warming. Full article

Collision avoidance between unmanned aerial vehicles (UAVs) and non-cooperative targets (e.g., off-nominal operations or birds) presents significant challenges in urban air mobility (UAM). This difficulty arises due to the highly dynamic and unpredictable flight intentions of these targets. Traditional collision-avoidance methods primarily focus on cooperative targets or non-cooperative ones with fixed behavior, rendering them ineffective when dealing with highly unpredictable flight patterns. To address this, we introduce a deep reinforcement learning-based collision-avoidance approach leveraging global and local intent prediction. Specifically, we propose a Global and Local Perception Prediction Module (GLPPM) that combines a state-space-based global intent association mechanism with a local feature extraction module, enabling accurate prediction of short- and long-term flight intents. Additionally, we propose a Fusion Sector Flight Control Module (FSFCM) that is trained with a Dueling Double Deep Q-Network (D3QN). The module integrates both predicted future and current intents into the state space and employs a specifically designed reward function, thereby ensuring safe UAV operations. Experimental results demonstrate that the proposed method significantly improves mission success rates in high-density environments, with up to 80 non-cooperative targets per square kilometer. In 1000 flight tests, the mission success rate is 15.2 percentage points higher than that of the baseline D3QN. Furthermore, the approach retains an 88.1% success rate even under extreme target densities of 120 targets per square kilometer. Finally, interpretability analysis via Deep SHAP further verifies the decision-making rationality of the algorithm. Full article

Background: Developmental coordination Disorder (DCD) is a prevalent and persistent neurodevelopmental condition characterized by motor learning difficulties that significantly affect daily functioning and participation. Despite growing interest in artificial intelligence (AI) applications within healthcare, the extent and nature of AI use in the evaluation and intervention of DCD remain unclear. Objective: This scoping review aimed to systematically map the existing literature on the use of AI and AI-assisted approaches in the evaluation, screening, monitoring, and intervention of DCD, and to identify current trends, methodological characteristics, and gaps in the evidence base. Methods: A scoping review was conducted in accordance with the PRISMA extension for Scoping Reviews (PRISMA-ScR) guidelines and was registered on the Open Science Framework. Systematic searches were performed in Scopus, PubMed, Web of Science, and IEEE Xplore, supplemented by snowballing. Peer-reviewed studies applying AI methods to DCD-relevant populations were included. Data was extracted and charted to summarize study designs, populations, AI methods, data modalities, clinical purposes, outcomes, and reported limitations. Results: Seven studies published between 2021 and 2025 met the inclusion criteria following a literature search covering the period from January 2010 to 2025. One study listed as 2026 was included based on its early access online publication in 2025. Most studies focused on AI applications for assessment, screening, and classification, using supervised machine learning or deep learning models applied to movement-based data, wearable sensors, video recordings, neurophysiological signals, or electronic health records. Only one randomized controlled trial evaluated an AI-assisted intervention. The evidence base was dominated by early-phase development and validation studies, with limited external validation, heterogeneous diagnostic definitions, and scarce intervention-focused research. Conclusions: Current AI research in DCD is primarily centered on evaluation and early identification, with comparatively limited evidence supporting AI-assisted intervention or rehabilitation. While existing findings suggest that AI has the potential to enhance objectivity and sensitivity in DCD assessment, significant gaps remain in clinical translation, intervention development, and implementation. Future research should prioritize theory-informed, clinician-centered AI applications, including adaptive intervention systems and decision-support tools, to better support occupational therapy and physiotherapy practice in DCD care. Full article

The integration of Software-Defined Networking (SDN), blockchain (BC), and machine learning (ML) has emerged as a promising approach to securing Internet of Things (IoT) and Industrial IoT (IIoT) networks. This paper conducted a comprehensive review of recent studies focusing on multi-layered security across device, control, network, and application layers. The analysis reveals that BC technology ensures decentralised trust, immutability, and secure access validation, while SDN enables programmability, load balancing, and real-time monitoring. In addition, ML/deep learning (DL) techniques, including federated and hybrid learning, strengthen anomaly detection, predictive security, and adaptive mitigation. Reported evaluations show similar gains in detection accuracy, latency, throughput, and energy efficiency, with effective defence against threats, though differing experimental contexts limit direct comparison. It also shows that the solutions’ effectiveness depends on ecosystem factors such as SDN controllers, BC platforms, cryptographic protocols, and ML frameworks. However, most studies rely on simulations or small-scale testbeds, leaving large-scale and heterogeneous deployments unverified. Significant challenges include scalability, computational and energy overhead, dataset dependency, limited adversarial resilience, and the explainability of ML-driven decisions. Based on the findings, future research should focus on lightweight consensus mechanisms for constrained devices, privacy-preserving ML/DL, and cross-layer adversarial-resilient frameworks. Advancing these directions will be important in achieving scalable, interoperable, and trustworthy SDN-IoT/IIoT security solutions. Full article

The widespread adoption of lithium-ion battery-powered electric vehicles has raised increasing concerns regarding battery safety under mechanical abuse conditions. However, mechanical abuse scenarios, such as battery extrusion, are highly diverse, making it impractical to conduct extensive destructive experiments and independent modeling for each specific scenario. In this work, a cross-scenario mechanical safety modeling framework for lithium-ion batteries is proposed based on transfer learning. Three quasi-static mechanical abuse tests, including flat-plate, rigid-rod, and hemispherical compression, are conducted on 18650 lithium-ion batteries. An equivalent mechanical model with a spring–damper parallel structure is employed to characterize the mechanical response and generate simulation data. Based on data from a single mechanical abuse scenario, a backpropagation neural network (BPNN)-based safety model is established to predict the maximum stress in the battery. The learned knowledge is then transferred to other mechanical abuse scenarios using a transfer learning strategy. The results demonstrate that, under limited target-domain data, the transferred models achieve stable prediction performance, with the average relative error controlled within 3.6%, outperforming models trained from scratch under the same conditions. Compared with existing studies that focus on single-scenario modeling, this work explicitly investigates cross-scenario transferability and demonstrates the effectiveness of transfer learning in reducing experimental and modeling effort for battery mechanical safety analysis. Full article

Unmanned aerial vehicles (UAVs) increasingly share low-altitude airspace with birds, making early distinguishing between drones and biological targets critical for safety and security. This work addresses long-range scenarios where objects occupy only a few pixels and appearance-based recognition becomes unreliable. We develop a model-driven simulation pipeline that generates synthetic data with a controlled camera model, atmospheric background and realistic motion of three aerial target types: multicopter, fixed-wing UAV and bird. From these sequences, each track is encoded as a time series of image-plane coordinates and apparent size, and a bidirectional long short-term memory (LSTM) network is trained to classify trajectories as drone-like or bird-like. The model learns characteristic differences in smoothness, turning behavior and velocity fluctuations, and to achieve reliable separation between drone and bird motion patterns on synthetic test data. Motion-trajectory cues alone can support early distinguishing of drones from birds when visual details are scarce, providing a complementary signal to conventional image-based detection. The proposed synthetic data and sequence classification pipeline forms a reproducible testbed that can be extended with real trajectories from radar or video tracking systems and used to prototype and benchmark trajectory-based recognizers for integrated surveillance solutions. The proposed method is designed to generalize naturally to real surveillance systems, as it relies on trajectory-level motion patterns rather than appearance-based features that are sensitive to sensor quality, illumination, or weather conditions. Full article

Cleaning historical silk textiles is a particularly sensitive operation that requires precise control to prevent mechanical or chemical damage. In this study, we investigate using flexible PVA–borax-based gels to remove soot from silk, i.e., polyvinyl alcohol–borax (PVA-B) gels and polyvinyl alcohol–borax–agarose double network gels (PVA-B/AG DN) loaded with different cleaning agents—namely, 30% ethanol and 1% Ecosurf EH-6—in addition to plain gels loaded with water. These gel formulations were tested on simplified model systems (SMS) and were applied using two methods: placing and tamping. The cleaning results were compared with a traditional contact-cleaning approach; micro-vacuuming followed by sponging. Visual inspection, 3D opto-digital microscopy, colorimetry, and machine-learning-assisted (ML) soot counting were exploited for the assessment of cleaning efficacy. Rheological characterization provided information about the flexibility and handling properties of the different gel formulations. Among the tested systems, the DN gel containing only water, applied by tamping, was easy to handle and demonstrated the highest soot-removal effectiveness without leaving residues, as confirmed by micro-Fourier Transform Infrared (micro-FTIR) analysis. Scanning electron microscope (SEM) micrographs proved the structural integrity of the treated silk fibers. Overall, this work allows us to conclude that PVA–borax-based gels offer an effective, adaptable, and low-risk cleaning strategy for historical silk fabrics. Full article