Search Results (785)

Despite the remarkable success of Deep Neural Networks (DNNs) in Remote Sensing Image (RSI) object detection, they remain vulnerable to adversarial attacks. Numerous adversarial attack methods have been proposed for RSI; however, adding a single large-scale adversarial patch to certain high-value targets, which are typically large in physical scale and irregular in shape, is both costly and inflexible. To address this issue, we propose a strategy of using multiple compact patches. This approach introduces two fundamental challenges: (1) how to optimize patch placement for a synergistic attack effect, and (2) how to retain strong adversarial potency within size-constrained mini-patches. To overcome these challenges, we introduce the Spatially Adaptive and Distillation-Enhanced Mini-Patch Attack (SDMPA) framework, which consists of two key modules: (1) an Adaptive Sensitivity-Aware Positioning (ASAP) module, which resolves the placement challenge by fusing the model’s attention maps from both an explainable and an adversarial perspective to identify optimal patch locations, and (2) a Distillation-based Mini-Patch Generation (DMPG) module, which tackles the potency challenge by leveraging knowledge distillation to transfer adversarial information from large teacher patches to small student patches. Extensive experiments on the RSOD and MAR20 datasets demonstrate that SDMPA significantly outperforms existing patch-based attack methods. For example, against YOLOv5n on the RSOD dataset, SDMPA achieves an Attack Success Rate (ASR) of 88.3% using only three small patches, surpassing other patch attack methods. Full article

Foreign object intrusion in railway perimeter areas poses significant risks to train operation safety. To address the limitation of current visual detection technologies that overly focus on target identification while lacking quantitative risk assessment, this paper proposes a railway intrusion risk quantification method integrating track semantic segmentation and spatiotemporal features. An improved BiSeNetV2 network is employed to accurately extract track regions, while physical-constrained risk zones are constructed based on railway structure gauge standards. The lateral spatial distance of intruding objects is precisely calculated using track gauge prior knowledge. A lightweight detection architecture is designed, adopting ShuffleNetV2 as the backbone to reduce computational complexity, with an incorporated Dilated Transformer module to enhance global context awareness and sparse feature extraction, significantly improving detection accuracy for small-scale objects. The comprehensive risk assessment formula integrates object category weights, lateral risk coefficients in intrusion zones, longitudinal distance decay factors, and dynamic velocity compensation. Experimental results demonstrate that the proposed method achieves 84.9% mean average precision (mAP) on our proprietary dataset, outperforming baseline models by 3.3%. By combining lateral distance detection with multidimensional risk indicators, the method enables quantitative intrusion risk assessment and graded early warning, providing data-driven decision support for active train protection systems and substantially enhancing intelligent safety protection capabilities. Full article

While Forgetting Factor Recursive Least Square (FFRLS) algorithms with evaluation mechanisms have been developed to address SOC-dependent parameter mapping shifts and their efficacy has been proven in Li-ion batteries, their applicability to lithium–sulfur (Li-S) batteries remains uncertain due to different electrochemical characteristics. This study critically evaluates the applicability of a Fisher information matrix-constrained FFRLS framework for online parameter identification in Li-S battery equivalent circuit network (ECN) models. Experimental validation using distinct drive cycles showed that the identification results of polarization-related parameters are significantly biased between different current excitations, and root mean square error (RMSE) variations diverge by 100%, with terminal voltage estimation errors more than 0.05 V. The parametric uncertainty under variable excitation profiles and voltage plateau estimation deficiencies confirms the inadequacy of such approaches, constraining model-based online identification viability for Li-S automotive applications. Future research should therefore prioritize hybrid estimation architectures integrating electrochemical knowledge with data-driven observers, alongside excitation capturing specifically optimized for Li-S online parameter observability requirements and cell nonuniformity and aging condition consideration. Full article

The iron and steel industry is one of the main industrial contributors to air pollution. The aim of our study is to analyze modern studies on air pollution by the iron and steel industry, as a result of which the geography and research directions and the degree of development of current issues will be assessed, and the most cited articles and journals will be identified. A review of contemporary research (2018–2024) was conducted on the basis of articles with a digital object identifier (DOI) using machine learning methodologies (VOSviewer software version 1.6.20). The number of articles selected was 80. The heat map of study density clearly showed that the geographic distribution of studies was extremely uneven. A total of 65% of the studies were conducted in China, 9% in Nigeria, 6% in Russia, 3% in Poland, and 3% in Turkey. The remaining 14% of articles represent a series of single studies conducted in 11 countries. The revealed geographical imbalance between countries with developed production and the number of studies conducted in them shows a significant shortcoming in monitoring research. Most of the studies (20%) were devoted to the assessment of multicomponent emissions. A special place among them was occupied by the inventory of emissions using various methods. The next main directions in terms of the number of articles were aimed at studying the toxic metal emissions (19%), at the analysis of organic emissions (19%), at the modeling and forecasting of emissions (18%), and at particulate matter studies (15%). The main features of the articles for each direction are briefly noted. Citation analysis made it possible to compile a rating of articles of greatest scientific interest and the most authoritative journals. Citation network analysis revealed important insights into the structure of scientific communication in the monitoring of atmospheric pollution from the iron and steel industry. The results of our review will contribute to the consolidation of scientists, the identification of gaps in scientific knowledge, and the improvement of environmental policy and technological solutions. Full article

The Qinghai-Tibet Plateau is a sensitive area of global climate change and an “Asian water tower” and lakes in Qinghai-Tibet Plateau changes are of great significance to the regional hydrological cycle and ecological balance. However, the existing research mostly focuses on a single lake or short-term monitoring, and lacks a systematic review of the evolution of knowledge structure and interdisciplinary dynamics. Based on 354 literatures from CNKI (China National Knowledge Infrastructure) and Web of Science, this study used CiteSpace 6.3.R1 software to construct a scientific knowledge map of lake changes in the Qinghai-Tibet Plateau under the background of climate change for the first time. By analyzing the number of publications, research hotspots, institutional cooperation networks and keyword emergence rules, the core triangle structure of ”climate change–Qinghai-Tibet Plateau–lake” was revealed, and the three stages of sedimentary reconstruction (2002–2008), glacier–lake coupling (2005–2014) and human–land system comprehensive research (2015–2025) were divided. The study found that the scientific literature written in Chinese and the scientific literature written in English focused on empirical cases and model simulations, respectively, The research frontiers focused on hot karst lakes (burst intensity 3.71), lake water level (2.97) and carbon cycle (2.13). The research force is centered on the Chinese Academy of Sciences, forming a cluster of institutions in the northwest region, but international cooperation only accounts for 12.3%. Future research needs to deepen multi-source data fusion, strengthen cross-regional comparison, and build an international cooperation network to cope with the complex challenges of plateau lake systems under climate change. This study provides a scientific basis for the paradigm shift and future direction of plateau lake research. Full article

Geological elements are fundamental components of the Earth’s ecosystem, and accurately identifying their spatial distribution is essential for analyzing environmental processes, guiding land-use planning, and promoting sustainable development. Remote sensing technologies, combined with artificial intelligence algorithms, offer new opportunities for the efficient interpretation of geological features. However, in areas with dense vegetation coverage, the information directly extracted from single-source optical imagery is limited, thereby constraining interpretation accuracy. Supplementary inputs such as synthetic aperture radar (SAR), topographic features, and texture information—collectively referred to as sensitive features and prior knowledge—can improve interpretation, but their effectiveness varies significantly across time and space. This variability often leads to inconsistent performance in general-purpose models, thus limiting their practical applicability. To address these challenges, we construct a geological element interpretation dataset for Northwest China by incorporating multi-source data, including Sentinel-1 SAR imagery, Sentinel-2 multispectral imagery, sensitive features (such as the digital elevation model (DEM), texture features based on the gray-level co-occurrence matrix (GLCM), geological maps (GMs), and the normalized difference vegetation index (NDVI)), as well as prior knowledge (such as base geological maps). Using five mainstream deep learning models, we systematically evaluate the performance improvement brought by various sensitive features and prior knowledge in remote sensing-based geological interpretation. To handle disparities in spatial resolution, temporal acquisition, and noise characteristics across sensors, we further develop a multi-source complement-driven network (MCDNet) that integrates an improved feature rectification module (IFRM) and an attention-enhanced fusion module (AFM) to achieve effective cross-modal alignment and noise suppression. Experimental results demonstrate that the integration of multi-source sensitive features and prior knowledge leads to a 2.32–6.69% improvement in mIoU for geological elements interpretation, with base geological maps and topographic features contributing most significantly to accuracy gains. Full article

Gear wear degrades transmission performance, necessitating highly reliable fault diagnosis methods. To address the limitations of existing approaches—where dynamic models rely heavily on prior knowledge, while data-driven methods lack interpretability—this study proposes an integrated bidirectional verification framework combining dynamic modeling and deep learning for interpretable gear wear diagnosis. First, a dynamic gear wear model is established to quantitatively reveal wear-induced modulation effects on meshing stiffness and vibration responses. Then, a deep network incorporating Gradient-weighted Class Activation Mapping (Grad-CAM) enables visualized extraction of frequency-domain sensitive features. Bidirectional verification between the dynamic model and deep learning demonstrates enhanced meshing harmonics in wear faults, leading to a quantitative diagnostic index that achieves 0.9560 recognition accuracy for gear wear across four speed conditions, significantly outperforming comparative indicators. This research provides a novel approach for gear wear diagnosis that ensures both high accuracy and interpretability. Full article

Quasi-solid aqueous zinc batteries (QSAZBs) have wide applications in the energy storage field due to their advantages of high safety, cost-effectiveness, and eco-friendliness. Despite prolific research output in the field of QSAZBs, existing reviews predominantly focus on experimental advancements, with limited synthesis of global research trends, interdisciplinary connections, or knowledge gaps. Herein, we review the research on QSAZBs via bibliometric analysis using the VOSviewer software (version 1.6.20). First, the data from qualitatively evaluated publications on QSAZBs from 2016 and 2024 are integrated. In addition, the annual trends, leading countries/regions and their international collaborations, institutional research and patent distribution, and important keyword cluster analyses in QSAZB research are evaluated. The results reveal that China dominates in terms of publication output (71.16% of total papers), and Singapore exhibits the highest citation impact (103.2 citations/paper). International collaboration networks indicate the central role of China, with strong ties to Singapore, the USA, and Australia. Keyword clustering indicates core research priorities: cathode materials (MnO₂ and V₂O₅), quasi-solid electrolyte optimization (hydrogels and graphene composites), and interfacial stability mechanisms. By mapping global trends and interdisciplinary linkages, this work provides insights to accelerate QSAZBs’ transition from laboratory breakthroughs to grid-scale and wearable applications. Full article

Forest fire detection is an essential application in environmental surveillance since wildfires cause devastating damage to ecosystems, human life, and property every year. The effective and accurate detection of fire is necessary to allow for timely response and efficient management of disasters. Traditional techniques for fire detection often experience false alarms and delayed responses in various environmental situations. Therefore, developing robust, intelligent, and real-time detection systems has emerged as a central challenge in remote sensing and computer vision research communities. Despite recent achievements in deep learning, current forest fire detection models still face issues with generalizability, lightweight deployment, and accuracy trade-offs. In order to overcome these limitations, we introduce a novel technique (FireNet-KD) that makes use of knowledge distillation, a method that maps the learning of hard models (teachers) to a light and efficient model (student). We specifically utilize two opposing teacher networks: a Vision Transformer (ViT), which is popular for its global attention and contextual learning ability, and a Convolutional Neural Network (CNN), which is esteemed for its spatial locality and inductive biases. These teacher models instruct the learning of a Swin Transformer-based student model that provides hierarchical feature extraction and computational efficiency through shifted window self-attention, and is thus particularly well suited for scalable forest fire detection. By combining the strengths of ViT and CNN with distillation into the Swin Transformer, the FireNet-KD model outperforms state-of-the-art methods with significant improvements. Experimental results show that the FireNet-KD model obtains a precision of 95.16%, recall of 99.61%, F1-score of 97.34%, and mAP@50 of 97.31%, outperforming the existing models. These results prove the effectiveness of FireNet-KD in improving both detection accuracy and model efficiency for forest fire detection. Full article

Blockchain technology’s increasing adoption across diverse sectors necessitates robust security measures to mitigate rising fraudulent activities. This paper presents a comprehensive bibliometric analysis of anomaly detection research in blockchain networks from 2017 to 2024, conducted under the PRISMA paradigm. Using CiteSpace 6.4.R1, we systematically map the knowledge domain based on 363 WoSCC-indexed articles. The analysis encompasses collaboration networks, co-citation patterns, citation bursts, and keyword trends to identify emerging research directions, influential contributors, and persistent challenges. The study reveals geographical concentrations of research activity, key institutional players, the evolution of theoretical frameworks, and shifts from basic security mechanisms to sophisticated machine learning and graph neural network approaches. This research summarizes the state of the field and highlights future directions essential for blockchain security. Full article

This paper presents a meta-analysis of social research on water, offering a novel methodological contribution to the study of emerging interdisciplinary research fields. We propose and implement a mixed methods framework that integrates quantitative network analysis with qualitative research, aiming to enhance both to give access to new emerging empirical fields and enhance the analytical depth of empirical social research. Drawing on a dataset of publications from the Web of Science over four distinct time intervals, we identify thematic clusters through keyword co-occurrence networks that reveal the evolving structure and internal dynamics of the field. Our findings show a clear trend toward increasing interdisciplinarity, responsiveness to global events, and contemporary challenges such as the emergence of COVID-19 and the continued centrality of topics related to water management and evaluation. By uncovering latent structures, our approach not only maps the field’s development but also lays the foundation for targeted qualitative analysis of articles representative of identified clusters. This methodological design contributes to the broader discourse on mixed methods research in the social sciences by demonstrating how computational tools can enhance the transparency and reliability of qualitative inquiry without sacrificing its interpretive richness. Furthermore, this study opens new avenues for critically reflecting on the epistemic culture of social water research, particularly in relation to its proximity to applied science and governance-oriented perspectives. The proposed method holds potential relevance for both academic researchers and decision makers in the water sector, offering a means to systematically access dispersed knowledge and identify underrepresented subfields. Overall, the study showcases the potential of mixed methods designs for navigating and structuring complex interdisciplinary research landscapes. Full article

With the ongoing advancement of industrialization and rapid urbanization, the emission of volatile organic compounds (VOCs) has increased significantly. As key precursors of PM_2.5 and ozone formation, VOCs pose a growing threat to the health of ecosystems. Due to their complex and dynamic transformation processes across air, water, and soil media, the ecological risks associated with VOCs have attracted increasing attention from both the scientific community and policy-makers. This study systematically reviews the core literature on the ecological impacts of VOCs published between 2005 and 2024, based on data from the Web of Science and Google Scholar databases. Utilizing three bibliometric tools (CiteSpace, VOSviewer, and Bibliometrix), we conducted a comprehensive visual analysis, constructing knowledge maps from multiple perspectives, including research trends, international collaboration, keyword evolution, and author–institution co-occurrence networks. The results reveal a rapid growth in the ecological impact of VOCs (EIVOCs), with an average annual increase exceeding 11% since 2013. Key research themes include source apportionment of air pollutants, ecotoxicological effects, biological response mechanisms, and health risk assessment. China, the United States, and Germany have emerged as leading contributors in this field, with China showing a remarkable surge in research activity in recent years. Keyword co-occurrence and burst analyses highlight “air pollution”, “exposure”, “health”, and “source apportionment” as major research hotspots. However, challenges remain in areas such as ecosystem functional responses, the integration of multimedia pollution pathways, and interdisciplinary coordination mechanisms. There is an urgent need to enhance monitoring technology integration, develop robust ecological risk assessment frameworks, and improve predictive modeling capabilities under climate change scenarios. This study provides scientific insights and theoretical support for the development of future environmental protection policies and comprehensive VOCs management strategies. Full article

Traditional Chinese medicine (TCM), a holistic medical system rooted in dialectical theories and natural product-based therapies, has served as a cornerstone of healthcare systems for millennia. While its empirical efficacy is widely recognized, the polypharmacological mechanisms stemming from its multi-component nature remain poorly characterized. The conventional trial-and-error approaches for bioactive compound screening from herbs raise sustainability concerns, including excessive resource consumption and suboptimal temporal efficiency. The integration of artificial intelligence (AI) and multi-omics technologies with network pharmacology (NP) has emerged as a transformative methodology aligned with TCM’s inherent “multi-component, multi-target, multi-pathway” therapeutic characteristics. This convergent review provides a computational framework to decode complex bioactive compound–target–pathway networks through two synergistic strategies, (i) NP-driven dynamics interaction network modeling and (ii) AI-enhanced multi-omics data mining, thereby accelerating drug discovery and reducing experimental costs. Our analysis of 7288 publications systematically maps NP-AI–omics integration workflows for natural product screening. The proposed framework enables sustainable drug discovery through data-driven compound prioritization, systematic repurposing of herbal formulations via mechanism-based validation, and the development of evidence-based novel TCM prescriptions. This paradigm bridges empirical TCM knowledge with mechanism-driven precision medicine, offering a theoretical basis for reconciling traditional medicine with modern pharmaceutical innovation. Full article

Physics-informed neural networks (PINNs) have emerged as a transformative methodology integrating deep learning with scientific computing. This review establishes a three-dimensional analytical framework to systematically decode PINNs’ development through methodological innovation, theoretical breakthroughs, and cross-disciplinary convergence. The contributions include threefold: First, identifying the co-evolutionary path of algorithmic architectures from adaptive optimization (neural tangent kernel-guided weighting achieving 230% convergence acceleration in Navier-Stokes solutions) to hybrid numerical-deep learning integration (5× speedup via domain decomposition) and second, constructing bidirectional theory-application mappings where convergence analysis (operator approximation theory) and generalization guarantees (Bayesian-physical hybrid frameworks) directly inform engineering implementations, as validated by $72\%$ cost reduction compared to FEM in high-dimensional spaces ($p<0.01,n=15$ benchmarks). Third, pioneering cross-domain knowledge transfer through application-specific architectures: TFE-PINN for turbulent flows ($5.12\pm 0.87\%$ error in NASA hypersonic tests), ReconPINN for medical imaging ($▵\mathrm{SSIM}=+0.18\pm 0.04$ on multi-institutional MRI), and SeisPINN for seismic systems ($0.52\pm 0.18$ km localization accuracy). We further present a technological roadmap highlighting three critical directions for PINN 2.0: neuro-symbolic, federated physics learning, and quantum-accelerated optimization. This work provides methodological guidelines and theoretical foundations for next-generation scientific machine learning systems. Full article

The significant burden of polypharmacy in clinical settings contrasts sharply with the narrow research focus on drug–drug interactions (DDIs), revealing an important gap in understanding the complexity of real-world multi-drug regimens. The present study addresses this gap by conducting a high-resolution, multidimensional bibliometric and network analysis of 19,151 DDI publications indexed in the Web of Science Core Collection (1975–2025). Using advanced tools, including VOSviewer version 1.6.20, Bibliometrix 5.0.0, and AI-enhanced terminology normalization, global research trajectories, knowledge clusters, and collaborative dynamics were systematically mapped. The analysis revealed an exponential growth in publication volume (from 55 in 1990 to 1194 in 2024), with output led by the United States and a marked acceleration in Chinese contributions after 2015. Key pharmacological agents frequently implicated in DDI research included CYP450-dependent drugs such as statins, antiretrovirals, and central nervous system drugs. Thematic clusters evolved from mechanistic toxicity assessments to complex frameworks involving clinical risk management, oncology co-therapies, and pharmacokinetic modeling. The citation impact peaked at 3.93 per year in 2019, reflecting the increasing integration of DDI research into mainstream areas of pharmaceutical science. The findings highlight a shift toward addressing polypharmacy risks in aging populations, supported by novel computational methodologies. This comprehensive assessment offers insights for researchers and academics aiming to navigate the evolving scientific landscape of DDIs and underlines the need for more nuanced system-level approaches to interaction risk assessment. Future studies should aim to incorporate patient-level real-world data, expand bibliometric coverage to underrepresented regions and non-English literature, and integrate pharmacogenomic and time-dependent variables to enhance predictive models of interaction risk. Cross-validation of AI-based approaches against clinical outcomes and prospective cohort data are also needed to bridge the translational gap and support precision dosing in complex therapeutic regimens. Full article