Search Results (78)

As artificial intelligence continues to evolve, reinforcement learning (RL) has shown remarkable potential for solving complex sequential decision problems and is now applied in diverse areas, including robotics, autonomous vehicles, and financial analytics. Among the various RL paradigms, multi-agent reinforcement learning (MARL) stands out for its ability to manage cooperative and competitive interactions within multi-entity systems. However, mainstream MARL algorithms still face critical challenges in training stability and policy generalization due to factors such as environmental non-stationarity, policy coupling, and inefficient sample utilization. To mitigate these limitations, this study introduces an enhanced algorithm named MATD3_AHD, developed by extending the MATD3 framework, which integrates TD3 and MADDPG principles. The goal is to improve the learning efficiency and overall policy effectiveness of agents operating in complex environments. The proposed method incorporates three key mechanisms: (1) an Adaptive Exploration Policy (AEP), which dynamically adjusts the perturbation magnitude based on TD error to improve both exploration capability and training stability; (2) a Hierarchical Sampling Policy (HSP), which enhances experience utilization through sample clustering and prioritized replay; and (3) a Dynamic Delayed Update (DDU), which adaptively modulates the actor update frequency based on critic network errors, thereby accelerating convergence and improving policy stability. Experiments conducted on multiple benchmark tasks within the Multi-Agent Particle Environment (MPE) demonstrate the superior performance of MATD3_AHD compared to baseline methods such as MADDPG and MATD3. The proposed MATD3_AHD algorithm outperforms baseline methods—by an average of 5% over MATD3 and 20% over MADDPG—achieving faster convergence, higher rewards, and more stable policy learning, thereby confirming its robustness and generalization capability. Full article

Gear defect detection directly affects the operational reliability of critical equipment in fields such as automotive and aerospace. Gear defect detection technology based on machine vision, leveraging the advantages of non-contact measurement, high efficiency, and cost-effectiveness, has become a key support for quality control in intelligent manufacturing. However, it still faces challenges including difficulties in semantic alignment of multimodal data, the imbalance between real-time detection requirements and computational resources, and poor model generalization in few-shot scenarios. This paper takes the paradigm evolution of gear defect detection technology as the main line, systematically reviews its development from traditional image processing to deep learning, and focuses on the innovative application of intelligent algorithms. A research framework of “technical bottleneck-breakthrough path-application verification” is constructed: for the problem of multimodal fusion, the cross-modal feature alignment mechanism based on Transformer network is deeply analyzed, clarifying its technical path of realizing joint embedding of visual and vibration signals by establishing global correlation mapping; for resource constraints, the performance of lightweight models such as MobileNet and ShuffleNet is quantitatively compared, verifying that these models reduce Parameters by 40–60% while maintaining the mean Average Precision essentially unchanged; for small-sample scenarios, few-shot generation models based on contrastive learning are systematically organized, confirming that their accuracy in the 10-shot scenario can reach 90% of that of fully supervised models, thus enhancing generalization ability. Future research can focus on the collaboration between few-shot generation and physical simulation, edge-cloud dynamic scheduling, defect evolution modeling driven by multiphysics fields, and standardization of explainable artificial intelligence. It aims to construct a gear detection system with autonomous perception capabilities, promoting the development of industrial quality inspection toward high-precision, high-robustness, and low-cost intelligence. Full article

As one of the primary fatal diseases globally, cancer represents a severe threat to human health because of its high incidence and fatality rates. While traditional treatments including surgery, radiation, and conventional pharmacotherapy demonstrate therapeutic effects, they commonly suffer from issues like severe side effects, high rates of relapse, and immunosuppression. The advent of immune checkpoint inhibitors and targeted drugs has undoubtedly revolutionized cancer management and improved survival; however, a significant proportion of patients still encounter obstacles such as acquired resistance, an immunosuppressive tumor microenvironment, and poor drug delivery to avascular tumor regions. Recent integration of engineered bacteria with nanomaterials has offered novel strategies for cancer immunotherapy. Engineered bacteria feature natural tumor tropism, immune-stimulating properties, and programmability, while nanomaterials are characterized by high drug payload, tunable release profiles, and versatile functionality. This article reviews the application of hybrid systems integrating engineered bacteria and nanomaterials in cancer immunotherapy, exploring their potential for drug delivery, immunomodulation, targeted treatment, and smart responsiveness. The construction of an “intelligent drug factory” through the merger of bacterial biological traits and sophisticated nanomaterial functions enables precise manipulation of the tumor microenvironment and potent immune activation, thereby establishing a novel paradigm for the precise treatment of solid tumors. However, its clinical translation faces challenges such as long-term biosafety, genetic stability, and precise spatiotemporal control. Synergistic integration with therapies such as radiotherapy, chemotherapy, and immunotherapy represents a promising direction worthy of exploration. Full article

In the past decades, significant advancements in the biological and genetic characterization of acute leukemias and optimization of risk-adapted multi-agent treatment protocols have dramatically improved cure rates and quality of life for children with acute lymphoblastic leukemia (ALL). Despite these optimal results, patients with relapsed or chemotherapy-refractory (R/R) disease or with high-risk genetic features still face unsatisfactory outcomes. Further intensification of conventional chemotherapy has reached its limits in achieving the desired efficacy without undue side effects, necessitating innovative approaches to improve cure rates while continuing to minimize the toxicities associated with chemotherapy and hematopoietic stem cell transplantation. In the era of precision medicine, two key therapeutic strategies have emerged in hemato-oncology: molecularly targeted therapies and immunotherapies. Antibody-based and cellular immunotherapies have undoubtedly reshaped the landscape of childhood ALL treatment and have significant potential to play leading roles in current and future frontline regimens; these important therapies are well delineated in recent reviews. Molecularly targeted small molecule inhibitor therapies remain a cornerstone of precision medicine, supported by recent advancements in next-generation sequencing, which have enabled the application of transcriptomic and genomic profiling data to risk stratification and therapy optimization. Clinical trials for children with ALL have been instrumental in refining therapies and improving outcomes, a paradigm that remains critical as treatment strategies become increasingly complex. This comprehensive review focuses upon molecularly targeted therapy approaches for childhood ALL and aims to summarize findings from completed clinical trials to highlight the current landscape of ongoing and upcoming trials and to provide insights into future directions for the precision-driven optimization of pediatric B-ALL and T-ALL treatment. Full article

Introduction/Background: Maternal depression is a significant risk factor for infant emotion regulation (ER), often linked to detrimental mother–infant interactions. Individual effects of maternal depression and maternal sensitivity are known, but their combined influence on infant ER across different emotional contexts remains underexplored. This study investigates concurrent relations among maternal depression, maternal sensitivity, and infant ER in low- and high-arousal contexts in a matched sample of primarily White educated mothers. Methods: We examined 5-month-old infants of clinically depressed and nondepressed mothers. Maternal sensitivity was coded from home observations; infant ER behaviors (e.g., gaze aversion, object-attend, self-soothing) were assessed through observation during modified Still-Face Paradigm (SFP) and fear-eliciting tasks. Results: Clinically depressed mothers exhibited lower maternal sensitivity than nondepressed mothers. Infants of depressed mothers used adaptive ER strategies less—specifically, lower monitoring and gaze aversion in the SFP, and lower gaze aversion and object-attend in the Fear task. Maternal sensitivity moderated the association between maternal depression and infant gaze aversion during the SFP and both gaze avert and object-attend during the Fear task. There was a context-specific regulatory difference for self-soothing; only infants of depressed mothers used self-soothing significantly more during the high-arousal Fear task. Conclusions: These findings underscore the interplay between maternal clinical depression and sensitivity in affecting infant ER. Maternal sensitivity acts as a crucial buffer against the adverse effects of maternal depression on infant ER. The results also indicate that infant emotion regulation varies in different contexts of low and high arousal. Interventions that target maternal sensitivity could significantly improve emotion regulation in infants of depressed mothers. Full article

Diabetes and cardiovascular diseases (CVDs) are two strongly associated conditions that mutually influence each other. This review aims to follow the historical timeline of their association by highlighting the changing paradigm in CV risk management and treatment strategies in type 2 diabetes (T2D). While the discovery of insulin was a breakthrough in reducing life-threatening complications like diabetic ketoacidosis, patients with diabetes still faced a poor prognosis in terms of macrovascular outcomes, especially CVDs. Initial efforts in improving outcomes by tightly controlling glycemia proved insufficient, highlighting the complex relationship between the two diseases. After decades of focusing solely on glucose-lowering strategies, rosiglitazone, a promising new drug was developed, ultimately raising the flag about the potential higher risk of CV complications like myocardial infarction associated with its use. This turning point shifted the focus towards CV safety of novel glucose-lowering drugs, mandating for the development of cardiovascular outcome trials. Several drug classes, like sodium-glucose co-transporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP-1 RAs), exceeded expectations by not only providing safety but also benefits in patients with T2D and CVDs and becoming the new standard of care in T2D management. The historical evidence linking T2D and CVDs has shaped regulatory requirements for cardiovascular outcome trials, guideline recommendations, and current therapeutic strategies. These insights highlight the importance of early interventions and a multidisciplinary approach to optimize patient outcomes. Full article

The successful integration of Terrestrial and Non-Terrestrial Networks (T/NTNs) in 6G is poised to revolutionize demanding domains like Earth Observation (EO) and Intelligent Transportation Systems (ITSs). Still, it requires Distributed Machine Learning (DML) frameworks that are scalable, private, and efficient. Existing methods, such as Federated Learning (FL) and Split Learning (SL), face critical limitations in terms of client computation burden and latency. To address these challenges, this paper proposes a novel hierarchical DML paradigm. We first introduce Federated Split Transfer Learning (FSTL), a foundational framework that synergizes FL, SL, and Transfer Learning (TL) to enable efficient, privacy-preserving learning within a single client group. We then extend this concept to the Generalized FSTL (GFSTL) framework, a scalable, multi-group architecture designed for complex and large-scale networks. GFSTL orchestrates parallel training across multiple client groups managed by intermediate servers (RSUs/HAPs) and aggregates them at a higher-level central server, significantly enhancing performance. We apply this framework to a unified T/NTN architecture that seamlessly integrates vehicular, aerial, and satellite assets, enabling advanced applications in 6G ITS and EO. Comprehensive simulations using the YOLOv5 model on the Cityscapes dataset validate our approach. The results show that GFSTL not only achieves faster convergence and higher detection accuracy but also substantially reduces communication overhead compared to baseline FL, and critically, both detection accuracy and end-to-end latency remain essentially invariant as the number of participating users grows, making GFSTL especially well suited for large-scale heterogeneous 6G ITS deployments. We also provide a formal latency decomposition and analysis that explains this scaling behavior. This work establishes GFSTL as a robust and practical solution for enabling the intelligent, connected, and resilient ecosystems required for next-generation transportation and environmental monitoring. Full article

Background: Non-communicable diseases (NCDs) have become a major threat to global public health, with the disease burden particularly severe in developing countries, China being one of them. The preventive and control effects of traditional treatment methods on NCDs are limited, and innovative strategies are urgently needed. In recent years, vaccine technology has expanded from the field of infectious diseases to non-communicable diseases (NCDs). Therapeutic vaccines have shown the potential to intervene in chronic diseases through immunomodulation, but their research and development (R & D), as well as promotion, still face multiple challenges. Methods: This article systematically reviews the current development status of NCD vaccines worldwide and points out the imbalance in their matching with disease burden: current research focuses on the field of cancer, while there is a lack of targeted vaccines for high-burden diseases such as hypertension and chronic kidney disease; the progress of independent R & D in China lags behind, and there are implementation obstacles such as uneven distribution of medical resources between urban and rural areas and low public willingness to be vaccinated. Results: By analyzing the biological mechanisms of NCD vaccines and non-biological challenges, phased solutions are proposed: In the short term, focus on target discovery and improvement of vaccine accessibility. In the medium term, strengthen multi-center clinical trials and international technology sharing. In the long term, build a digital health monitoring system and a public–private partnership financing model. Conclusions: The breakthrough of NCD vaccines requires interdisciplinary collaboration and systematic policy support. Their successful application will reshape the paradigm of chronic disease prevention and control, providing a new path for global health equity. Full article

The rapid advancement of general large models has significantly impacted and introduced new concepts to the traditional “one task, one model” research paradigm in construction automation. In this paper, we evaluate the performance of existing large models and those developed on large model platforms, using building material counting as an example. We compare three categories of large AI models for building material counting, including multimodal large models, purely visual large models, and secondary models developed on platforms. Through this research, we aim to explore the accuracy and practicality of these models in real-world construction scenarios. The results indicate that directly applying general large models faces challenges in processing photos with complex shapes or backgrounds, failing to provide accurate counting results. Additionally, while purely visual large models excel in instance segmentation tasks, their application to the specific counting of building materials requires additional programming work. To address these issues, this study explores solutions based on large model secondary development platforms and trains a model using EasyDL as an example. Leveraging deep learning techniques, this model achieves effective counting of building materials through five steps: data preparation, model type selection, model training, model validation, and model deployment. Although models developed based on large model platforms are presently less accurate than specialized models, they still represent a highly promising approach. Full article

The offshore implementation of vertical-axis wind turbines (VAWTs) presents a promising new paradigm for advancing marine wind energy utilization, owing to their omnidirectional wind acceptance, compact structural design, and potential for lower maintenance costs. However, VAWTs still face major aerodynamic challenges, particularly due to the pitching motion, where the angle of attack varies cyclically with the blade azimuth. This leads to strong unsteady effects and susceptibility to dynamic stalls, which significantly degrade aerodynamic performance. To address these unresolved issues, this study conducts a comprehensive investigation into the dynamic stall behavior and wake vortex evolution induced by Darrieus-type pitching motion (DPM). Quasi-three-dimensional CFD simulations are performed to explore how variations in blade geometry influence aerodynamic responses under unsteady DPM conditions. To efficiently analyze geometric sensitivity, a surrogate model based on a radial basis function neural network is constructed, enabling fast aerodynamic predictions. Sensitivity analysis identifies the curvature near the maximum thickness and the deflection angle of the trailing edge as the most influential geometric parameters affecting lift and stall behavior, while the blade thickness is shown to strongly impact the moment coefficient. These insights emphasize the pivotal role of blade shape optimization in enhancing aerodynamic performance under inherently unsteady VAWT operating conditions. Full article

Cancer research faces significant biological, technological, and systemic limitations that hinder the development of effective therapies and improved patient outcomes. Traditional preclinical models, such as 2D and 3D cell cultures, murine xenografts, and organoids, often fail to reflect the complexity of human tumor architecture, microenvironment, and immune interactions. This discrepancy results in promising laboratory findings not always translating effectively into clinical success. A core obstacle is tumor heterogeneity, characterized by diverse genetic, epigenetic, and phenotypic variations within tumors, which complicates treatment strategies and contributes to drug resistance. Hereditary malignancies and cancer stem cells contribute strongly to generating this complex panorama. Current early detection technologies lack sufficient sensitivity and specificity, impeding timely diagnosis. The tumor microenvironment, with its intricate interactions and resistance-promoting factors, further promotes treatment failure. Additionally, we only partially understand the biological processes driving metastasis, limiting therapeutic advances. Overcoming these barriers involves not only the use of new methodological approaches and advanced technologies, but also requires a cultural effort by researchers. Many cancer studies are still essentially observational. While acknowledging their significance, it is crucial to recognize the shift from deterministic to indeterministic paradigms in biomedicine over the past two to three decades, a transition facilitated by systems biology. It has opened the doors of deep metabolism where the functional processes that control and regulate cancer progression operate. Beyond biological barriers, systemic challenges include limited funding, regulatory complexities, and disparities in cancer care access across different populations. These socio-economic factors exacerbate research stagnation and hinder the translation of scientific innovations into clinical practice. Overcoming these obstacles requires multidisciplinary collaborations, advanced modeling techniques that better emulate human cancer, and innovative technologies for early detection and targeted therapy. Strategic policy initiatives must address systemic barriers, promoting health equity and sustainable research funding. While the complexity of cancer biology and systemic challenges are formidable, ongoing scientific progress and collaborative efforts inspire hope for breakthroughs that can transform cancer diagnosis, treatment, and survival outcomes worldwide. Full article

Industrial image anomaly detection plays a critical role in intelligent manufacturing by automatically identifying defective products through visual inspection. While unsupervised approaches eliminate dependency on annotated anomaly samples, current teacher–student framework-based methods still face two fundamental limitations: insufficient discriminative capability for structural anomalies and suboptimal anomaly feature decoupling efficiency. To address these challenges, we propose a Synthetic-Anomaly Contrastive Distillation (SACD) framework for industrial anomaly detection. SACD comprises two pivotal components: (1) a reverse distillation (RD) paradigm whereby a pre-trained teacher network extracts hierarchically structured representations, subsequently guiding the student network with inverse architectural configuration to establish hierarchical feature alignment; (2) a group of feature calibration (FeaCali) modules designed to refine the student’s outputs by eliminating anomalous feature responses. During training, SACD adopts a dual-branch strategy, where one branch encodes multi-scale features from defect-free images, while a Siamese anomaly branch processes synthetically corrupted counterparts. FeaCali modules are trained to strip out a student’s anomalous patterns in anomaly branches, enhancing the student network’s exclusive modeling of normal patterns. We construct a dual-objective optimization integrating cross-model distillation loss and intra-model contrastive loss to train SACD for feature alignment and discrepancy amplification. At the inference stage, pixel-wise anomaly scores are computed through multi-layer feature discrepancies between the teacher’s representations and the student’s refined outputs. Comprehensive evaluations on the MVTec AD and BTAD benchmark demonstrate that our method is effective and superior to current knowledge distillation-based approaches. Full article

Cooperative multi-agent reinforcement learning (MARL) has emerged as a powerful paradigm for addressing complex real-world challenges, including autonomous robot control, strategic decision-making, and decentralized coordination in unmanned swarm systems. However, it still faces challenges in learning proper coordination among multiple agents. The lack of effective knowledge sharing and experience interaction mechanisms among agents has led to substantial performance decline, especially in terms of low sampling efficiency and slow convergence rates, ultimately constraining the practical applicability of MARL. To address these challenges, this paper proposes a novel framework termed Reward redistribution and Experience reutilization based Coordination Optimization (RECO). This innovative approach employs a hierarchical experience pool mechanism that enhances exploration through strategic reward redistribution and experience reutilization. The RECO framework incorporates a sophisticated evaluation mechanism that assesses the quality of historical sampling data from individual agents and optimizes reward distribution by maximizing mutual information across hierarchical experience trajectories. Extensive comparative analyses of computational efficiency and performance metrics across diverse environments reveal that the proposed method not only enhances training efficiency in multi-agent gaming scenarios but also significantly strengthens algorithmic robustness and stability in dynamic environments. Full article

With the widespread application of unmanned aerial vehicles (UAVs), the safety detection system of UAVs has created an urgent need for anomaly detection technology. As a direct representation of system health status, flight data contain critical status information, driving data-driven methods to gradually replace traditional dynamic modeling as the mainstream paradigm. The former effectively circumvent the problems of nonlinear coupling and parameter uncertainty in complex dynamic modeling. However, data-driven methods still face two major challenges: the scarcity of anomalous flight data and the difficulty in extracting strong spatio-temporal coupling among flight parameters. To address these challenges, we propose an unsupervised anomaly detection method based on the causality-enhanced graph neural network (CEG). CEG innovatively introduces a causality model among flight parameters, achieving targeted extraction of spatial features through a causality-enhanced graph attention mechanism. Furthermore, CEG incorporates a trend-decomposed temporal feature extraction module to capture temporal dependencies in high-dimensional flight data. A low-rank regularization training paradigm is designed for CEG, and a residual adaptive bidirectional smoothing strategy is employed to eliminate the influence of noise. Experimental results on the ALFA dataset demonstrate that CEG outperforms state-of-the-art methods in terms of Precision, Recall, and F1 score. The proposed method enables accurate and robust anomaly detection on a wide range of anomaly types such as engines, rudders, and ailerons, validating its effectiveness in handling the unique challenges of UAV anomaly detection. Full article

The construction industry is undergoing a paradigm shift with the advent of Construction 5.0 (C5.0), which integrates artificial intelligence (AI), the Internet of Things (IoT), digital twins, blockchain, and robotics to enhance productivity, sustainability, and resilience. This study conducts a systematic literature review and bibliometric analysis of 78 scholarly sources published between 2022 and 2025, using data from Scopus and following the PRISMA method. Keyword co-occurrence mapping, citation analysis, and content review are utilized to identify key advancements, emerging trends, and adoption challenges in C5.0. Seven core technologies are examined through the lenses of sustainability, human–robot collaboration (HRC), and resilience, revealing a rapidly expanding yet still nascent research domain. While C5.0 presents transformative potential, its widespread implementation faces significant barriers. A critical evaluation of these challenges is conducted, alongside strategic pathways to facilitate adoption and maximize impact. Furthermore, the leading countries and seminal contributions in the field are highlighted to guide future research efforts. By addressing knowledge gaps and industry trends, this study provides practical insights for policymakers, researchers, and industry professionals, contributing to the development of innovative frameworks that enhance efficiency, sustainability, and resilience in the era of Industry 5.0. Full article