Search Results (935)

Autonomous task assignment in multi-unmanned aerial vehicle (UAV) systems operating in dynamic and safety-critical airspace environments is highly challenging due to complex spatial interactions and rapidly changing relative geometries. This paper proposes a hierarchical decision-making framework that bridges individual maneuvering behaviors with cooperative task allocation in multi-agent aerial systems. First, a high-fidelity single-agent maneuver model is learned using a physics-consistent simulation environment, where spatial advantage is evaluated based on relative distance and angular relationships within a kinematically feasible interaction zone (KIZ). Subsequently, a Geometry-Aware Graph Attention Network (GA-GAT) is developed to address scalable multi-agent assignment problems. Unlike conventional approaches that rely on flat feature representations, the proposed method explicitly incorporates kinematic feasibility constraints into the attention mechanism via a novel gating module, enabling efficient relational reasoning under dynamic conditions. The proposed framework is applicable to a range of civilian and safety-oriented scenarios, including UAV swarm coordination, emergency response monitoring, infrastructure inspection, and autonomous airspace management. Simulation results demonstrate that the GA-GAT-based approach significantly outperforms heuristic baselines in terms of coordination efficiency and overall system performance in complex multi-agent environments. This study highlights that decoupling maneuver-level control from high-level coordination provides a scalable and computationally efficient solution for real-time multi-UAV decision-making in safety-critical applications. The proposed framework is designed for general multi-agent coordination problems in civilian aerial applications. Full article

This paper presents an adaptive optimal consensus tracking control scheme for canonical nonlinear multi-agent systems (MASs) with unknown dynamics, employing an actor–critic reinforcement learning (RL) framework. The scheme integrates a sliding mode mechanism to suppress tracking errors and ensure consensus tracking between the followers and the leader. Additionally, optimal control is designed to find a Nash equilibrium in a graphical game. To address the intractability of obtaining an analytical solution for the coupled Hamilton–Jacobi–Bellman (HJB) equation, a policy iteration algorithm is utilized. Within this algorithm, a critic neural network (NN) approximates the gradient of the optimal value function, while an actor NN approximates the optimal control policy. Together, these networks form a compact actor–critic (AC) architecture that achieves optimal consensus tracking. Furthermore, the proposed method guarantees the boundedness of all closed-loop signals while ensuring consensus tracking. Finally, two simulations are conducted to verify the effectiveness and advantages of the proposed method. Full article

In IIoT edge computing, multi-edge server collaborative scheduling faces two core issues due to random task arrivals, heterogeneous resources, and complex topology: traditional model-driven methods cannot make dynamic decisions in dynamic environments, and conventional MARL fails to characterize inter-node topological dependencies and load correlations. To address this, this paper investigates the joint optimization of task offloading, computing resource allocation, and SFC orchestration in IIoT, constructs a cloud-edge-end collaborative architecture, and models the problem as a POMDP to minimize the overall system cost under multiple constraints. A graph-guided value-decomposition MARL method is proposed, which extracts spatial topology and neighborhood-load features of edge nodes via a GNN and combines them with the QMIX framework to realize multi-agent centralized training and distributed execution. Simulations show that the algorithm converges stably under different server scales and task loads, significantly outperforms benchmark algorithms, and can suppress performance degradation in high-load scenarios, demonstrating its robustness and scalability in complex industrial environments. Full article

Glucagon-like peptide-1 (GLP-1) and dual GIP/GLP-1 receptor agonists, originally introduced for the management of type 2 diabetes mellitus and obesity, are increasingly recognized for their broader actions within the central nervous system, with emerging implications in neuropsychiatry and neurodegeneration. This review integrates current preclinical and clinical evidence, emphasizing their pharmacodynamic profile, central receptor distribution, and the molecular pathways linking metabolic signaling to neural function. Evidence suggests that GLP-1 receptor activation across key brain regions involved in energy balance and reward modulates multiple neurotransmitter systems, including dopamine and serotonin, as well as glutamatergic and GABAergic transmission, thereby influencing behavior, affective processes, and cognitive function. In parallel, these agents exhibit neuroprotective properties through improved neuronal insulin sensitivity, attenuation of neuroinflammatory pathways, and support of neuroplasticity, alongside effects on limiting pathological protein aggregation. Dual GIP/GLP-1 agonism may further potentiate these central actions through complementary metabolic and synaptic mechanisms. Although pharmacovigilance data have identified isolated neuropsychiatric adverse events, current clinical evidence does not support a consistent causal association. Collectively, incretin-based therapies represent a promising translational approach at the interface of metabolic and neuropsychiatric disorders, warranting further investigation into their long-term central safety, therapeutic efficacy, and clinical relevance. Full article

With the rapid proliferation of large language models (LLMs), distinguishing machine-generated text from human-authored content has become increasingly critical for ensuring content authenticity, academic integrity, and trust in information systems. However, detecting text generated by LLMs remains a challenging problem, particularly in zero-shot settings where labeled data and domain-specific tuning are unavailable. To address this challenge, in this paper, we propose a novel Collaborative Multi-Agent Zero-Shot Detection framework (CMA-ZSD). In contrast to existing methods based on watermarking, statistical heuristics, or neural classifiers, our CMA-ZSD employs three functionally heterogeneous agents that perform differentiated perturbations of the input text. By jointly modeling semantic consistency, grammatical normalization, and feature-level reconstruction, our method captures intrinsic asymmetries between human-authored and LLM-generated text. A semantic similarity evaluation mechanism, combined with majority voting, enables robust and interpretable detection decisions that balance individual agent autonomy with collective consensus. Extensive experiments across 11 domains demonstrate the effectiveness of our method, with its zero-shot detection achieving accuracy comparable to domain-finetuned models in specific domains such as Finance and Reddit-dli5. Full article

Background/Objectives: Constipation is a common gastrointestinal problem in older adults and is associated with reduced quality of life, functional decline, frailty, and an increased risk of delirium and cognitive impairment. Its pathogenesis is multifactorial, involving age-related changes in gastrointestinal motility, neural regulation, comorbidities, and polypharmacy. However, this condition has traditionally been regarded as a localized gastrointestinal disorder, which may not fully reflect its systemic clinical significance in older populations. While prior narrative reviews have described multifactorial contributors to constipation, none have formally applied a geriatric syndrome framework to integrate these dimensions. This review proposes a three-criterion operational definition—multifactorial pathogenesis, association with functional decline and frailty, and contribution to adverse systemic outcomes—to characterize constipation in older adults as a “systemic geriatric syndrome,” and evaluates available evidence against each criterion. Methods: A narrative literature search was conducted using PubMed to identify relevant studies published between 1 January 2023, and 31 December 2025. MeSH terms included “Constipation” [Major Topic] and “Aged” [MeSH Terms]. Eligible articles included English-language original studies, systematic reviews, and clinical or epidemiological studies involving individuals aged ≥65 years. Results: Diagnosis in older adults is often complicated by secondary causes, including medications and neurological disorders, as well as atypical symptom presentations in individuals with cognitive impairment. Key pathophysiological mechanisms include reductions in interstitial cells of Cajal, impaired smooth muscle contractility, dysfunction of the enteric and autonomic nervous systems, and gut microbiota dysbiosis, which may promote chronic low-grade inflammation. Major contributing factors include physical inactivity, sarcopenia, dehydration, inappropriate defecation posture, and polypharmacy, particularly opioids and anticholinergic agents. Importantly, these factors interact through the brain–gut–microbiota axis, contributing not only to gastrointestinal dysfunction but also to systemic outcomes such as frailty, cognitive decline, and increased healthcare burden, thereby supporting a multidimensional disease framework. Conclusions: The available evidence collectively supports the plausibility of framing constipation in older adults as a systemic geriatric syndrome, though formal validation of this classification requires further longitudinal and mechanistic research. Comprehensive and individualized management strategies, extending beyond simple laxative use, are essential to reduce complications and preserve functional health in aging populations. Further studies are required to validate this framework. Full article

Coordinated control of multiple intersections remains a major challenge in urban traffic systems due to complex local–global dependencies. To address this issue, the MPNNLight framework is proposed, a message-passing graph neural network for cooperative multi-intersection traffic signal control. The core innovation of this work is the proposed Attention-Wavelet Spatial Transformer (AWSformer), a message computation module that combines self-attention mechanisms and 2D discrete wavelet transform (2D-DWT) to adaptively compute inter-agent messages across spatial and frequency domains. This design enables efficient extraction of multi-scale dependencies and enhances coordination among intersections. Experiments on diverse datasets demonstrate that MPNNLight achieves superior performance and efficiency, compared with existing GNN-based traffic signal control methods. Full article

Traumatic brain injury (TBI) represents a significant global health challenge with limited effective treatments. The secondary injury phase, characterized by persistent neuroinflammation, is a major contributor to long-term neurological deficits. Conventional therapies face substantial hurdles, including the blood–brain barrier (BBB), short therapeutic windows, and poor neuroregenerative capacity. Innovative biomaterials offer a promising platform to overcome these limitations by providing localized Drug Deliv., immunomodulation, and structural support for neural regeneration. This review outlines the pathological mechanisms of neuroinflammation and repair obstacles following TBI. It then systematically categorizes and discusses the mechanisms of various biomaterials—including natural, synthetic, nano-scale, composite, and intelligent materials—in modulating neuroinflammation. Furthermore, we elaborate on strategies for promoting neural repair, such as constructing regenerative scaffolds, delivering therapeutic agents (e.g., neurotrophic factors, stem cells, and exosomes), and remodeling the regenerative microenvironment. Special emphasis is placed on the emerging application of exosome delivery systems. Finally, we address the challenges in clinical translation and present future perspectives on smart materials, multi-modal systems, and personalized therapies, highlighting the transformative potential of biomaterials in TBI management. Full article

As a crucial component of multimodal sensing in modern AI agents, remote sensing images have attracted significant attention, for which neural representation is a promising direction. Implicit Neural Representations (INRs) using Multi-Layer Perceptrons (MLPs) have the ability to model images by learning an implicit mapping from pixel coordinates to pixel intensities. This paper revisits the ReLU activation function, a widely adopted non-linearity known for its dead region on the negative axis, within the context of MLP-based INRs. We introduce the Dead-Free Linear Unit (DeLU), a novel activation function that leverages a linearly transformed absolute value to eliminate inactive regions. By combining dead-free non-linearity with adaptive linear scaling, DeLU enhances the expressiveness of INR architectures, particularly those employing periodic activations. Extensive experiments across multiple remote sensing datasets, including LandCover.ai, LoveDA, INRIA, UAVid, and ISPRS Potsdam, validate the efficacy of our proposed method. Full article

Climate change disproportionately threatens low- and middle-income countries, yet integrated assessments combining socio-economic fragility with physical hazards remain limited. This study quantifies multi-dimensional climate vulnerability and derives optimized adaptation policies for six representative nations (Bangladesh, Colombia, Kenya, Morocco, Pakistan, Vietnam) by fusing socio-economic indicators with climate risk data (2000–2024). A computational framework integrating unsupervised learning, dimensionality reduction, and predictive modeling was employed. Principal Component Analysis synthesized eight indicators into a Compound Vulnerability Score (CVS), while K-Means and DBSCAN identified distinct vulnerability regimes. XGBoost quantified driver importance, and Graph Neural Networks captured systemic interconnections. XGBoost identified projected drought risk (31.2%), precipitation change (18.1%), and poverty headcount (14.3%) as primary drivers. Graph networks demonstrated significant risk amplification in African nations (Morocco SRS: 0.728–0.874; Kenya SRS: 0.504–0.641) versus damping in Asian countries. A Reinforcement Learning (RL) agent was trained using Deep Q-Networks with experience replay to optimize intervention portfolios under budget constraints. The RL policy achieved a 23% reduction in systemic risk compared to uniform allocation baselines, generating context-specific priorities: drought management for Morocco (score 50) and Pakistan (40); poverty alleviation for Kenya (40); coastal protection for Bangladesh (40); agricultural resilience for Vietnam (35); and institutional capacity building for Colombia (50). In conclusion, socio-economic fragility non-linearly amplifies climate hazards, with poverty and drought risk constituting critical vulnerability multipliers. The AI-driven framework demonstrates that targeted interventions in high-sensitivity systems maximize systemic risk reduction. This integrated approach provides a replicable, evidence-based foundation for strategic adaptation finance allocation in an increasingly uncertain climate future. Full article

The development of novel antiepileptic agents requires early identification of pharmacokinetic limitations to mitigate risks at later stages. This study aimed to perform in silico profiling of a library containing 448 novel 2H,5H-chromeno[4’,3’:4,5]thiopyrano[2,3-d]thiazol-2-one derivatives to select lead compounds with an optimal balance of safety and efficacy. The study was conducted using the ADMET-AI platform, based on a graph neural network, to predict physicochemical, pharmacokinetic, and toxicological properties. The methodology involved calculating drug-likeness descriptors for primary screening and a comparative statistical analysis of the top 20 selected structures against 16 approved antiepileptic drugs and four reference compounds. Based on drug-likeness descriptors and predicted ADMET (absorption, distribution, metabolism, excretion, toxicity) related parameters, 20 structures were prioritized for further analysis. Their predicted profiles suggested high intestinal absorption and blood–brain barrier (BBB) permeability, which may be relevant for central nervous system (CNS) directed agents. In comparison with the reference thiazolidinones, the prioritized compounds showed comparatively more favorable predicted mutagenicity and carcinogenicity profiles. Elevated predicted risks of hepatotoxicity and cardiotoxicity were observed for several structures, indicating the need for further structural optimization. The results suggest that the thiopyranothiazolidinone scaffold merits further anticonvulsant-oriented investigation at the stage of early compound prioritization. Experimental validation will be required to confirm the actual pharmacokinetic, toxicological, and anticonvulsant properties of the prioritized compounds. Full article

Geometric problem-solving has always been a great challenge in the field of deductive reasoning and artificial intelligence. Symmetry is a defining characteristic of geometric shapes and properties. Consequently, the application of symmetry principles to geometric reasoning arises as a natural choice. To address the efficiency degradation and limited generalization, we propose FGeo-ISRL, a neural-symbolic inverse search framework whose core is the synergistic integration of a task-fine-tuned large language model and Monte Carlo Tree Search. Under the formal framework of FormalGeo, geometric theorems are iteratively applied starting from the given conditions and the target conclusion, in order to infer the necessary supporting premises. The large language model simultaneously serves as a policy network and a value network, guiding theorem application decisions and evaluating intermediate proof states, whereas the Monte Carlo Tree Search performs structured exploration over the state space, both training for policy refinement and inference for online search. The reinforcement learning agent is trained with a hybrid reward scheme, combining immediate feedback from the value difference and a sparse success reward. Experiments demonstrate the effectiveness and correctness of FGeo-ISRL. It not only achieves a Single-Step Theorem Accuracy of 90.2% and a Geometric Problem-Solving Accuracy of 83.14%, but also ensures that every step of the proof process remains readable, verifiable, and traceable. Full article

With the rapid development of e-commerce and smart manufacturing, automated warehouse systems have become critical infrastructure for modern logistics. In China’s vast market, the dynamic scheduling of Rail-Guided Vehicles (RGVs) faces significant challenges due to complex task uncertainties, hierarchical supply chain structures, and real-time collision avoidance requirements. Traditional rule-based methods and static optimization models often fail to adapt to such dynamic environments. To address these issues, this paper proposes a novel hybrid deep reinforcement learning framework integrating a Dynamic Graph Neural Network (DGNN) and a Transformer model. The DGNN captures the spatiotemporal dependencies of the warehouse network topology, while the Transformer mechanism enhances long-range feature extraction for task prioritization. Furthermore, we design a centralized Deep Q-network (DQN) framework with parameterized action spaces to coordinate multiple RGVs collaboratively. While the system manages multiple physical vehicles, the learning architecture employs a single-agent global scheduler to avoid the non-stationarity issues inherent in multi-agent reinforcement learning. Experimental results based on real-world data from a large-scale electronics manufacturing warehouse demonstrate that our method reduces average task completion time by 18.5% and improves system throughput by 22.3% compared to state-of-the-art baselines. The proposed approach demonstrates potential for intelligent warehouse management in dynamic industrial scenarios. Full article

In polar environments, icing on ship deck surfaces severely compromises navigation safety. Conventional electric trace heating systems operate in continuous heating mode, resulting in high energy consumption. This study proposes an intelligent periodic heating control method that integrates physics-informed neural networks (PINNs) and deep reinforcement learning (DRL) for energy-efficient anti-icing of circular pipe components on polar vessels. Using a polar coupled environment simulation platform, experiments were conducted on electric heating anti-icing for circular pipe components. Temperature data under various heating modes were collected, and a physically constrained PINN temperature prediction model was constructed, achieving high prediction accuracy with limited samples (test set R² = 0.9091; 5-fold cross-validation R² = 0.8877 ± 0.0312). The DRL agent trained in this virtual environment autonomously optimized the heating strategy, yielding optimal cycle parameters: heating ratio D = 0.722 and cycle duration τ = 88 s. While maintaining surface temperatures above 0 °C against a −10 °C ambient baseline, this strategy achieved a unit energy consumption of 0.27 kJ/°C, representing a 63% reduction compared to conventional continuous heating. This study provides a data-physics fusion control approach for polar vessel anti-icing systems, demonstrating strong potential for engineering applications. Full article

The shrinkage deformation is a significant risk to self-compacting concrete (SCC)-filled steel tube structures. It was essential to understand the concrete autogenous shrinkage strain before being regulated in order to determine compensation shrinkage measures. In this study, A PSO-BPNN model was constructed, which is based on the Particle Swarm Optimization-Back Propagation Neural Networks (PSO-BPNN), and the autogenous shrinkage strain of SCC was predicted based on PSO-BPNN before being regulated. Moreover, some experiments about compensating for shrinkage by expansion and by a combination of expansion and contraction were investigated. Based on this prediction, a series of experiments was conducted on the regulation of the shrinkage deformation of SCC for an actual bridge project. The results indicated that a good consistency of PSO-BPNN between predicted and measured values, demonstrating that PSO-BPNN is a model with high accuracy in predicting concrete autogenous shrinkage strain before regulation, and as a guidance for regulation to compensate for shrinkage. The prediction error was less than 10% for 28-day self-shrinkage, and the experimental workload was reduced. The PSO-BPNN is a convenient tool for predicting the shrinkage of SCC, enabling the determination of dosages of expansion agent and reducing shrinkage agent to achieve SCC’s shrinkage regulation. Full article