Large-scale Networks Research Articles

Jaynes machine: The universal microstructure of deep neural networks

Despite the recent stunning progress in large-scale deep neural network applications, our understanding of their microstructure, ‘energy’ functions, and optimal design remains incomplete. Here, we present a new game-theoretic framework, called statistical teleodynamics, that reveals important insights into these key properties. The optimally robust design of such networks inherently involves computational benefit-cost trade-offs that physics-inspired models do not adequately capture. These trade-offs occur as neurons and connections compete to increase their effective utilities under resource constraints during training. In a fully trained network, this results in a state of arbitrage equilibrium, where all neurons in a given layer have the same effective utility, and all connections to a given layer have the same effective utility. The equilibrium is characterized by the emergence of two lognormal distributions of connection weights and neuronal output as the universal microstructure of large deep neural networks. We call such a network the Jaynes Machine. Our theoretical predictions are shown to be supported by empirical data from seven large-scale deep neural networks. We also show that the Hopfield network and the Boltzmann Machine are the same special case of the Jaynes Machine.

A Remote Monitoring System for Wind Speed and Direction Based on Non-Contact Triboelectric Nanogenerator

Wind speed and wind direction sensors are crucial sensor categories in the Internet of Things (IoT), particularly vital in fields such as meteorological monitoring, construction engineering, transportation engineering, and ocean engineering. However, power supply remains a key limiting factor for the widespread application of these sensors in large-scale sensor networks. In this paper, a self-powered, non-contact wind speed and direction sensor based on the triboelectric nanogenerator (SD-TENG) is proposed. The optimized wind speed measurement structure has a start-up wind speed as low as 0.2m/s and exhibits good linearity in the range of 0.2-29m/s. Additionally, the sensor demonstrates high temperature and humidity resistance, with a voltage attenuation of 2.4% at 45°C ambient temperature and 9.8% at 95% relative humidity. For wind direction measurement, the design of non-uniform electrodes enhances the recognition capability of different channels. To meet the demands of remote monitoring, we have designed an advanced signal processing circuit that can directly convert the raw output of a wind speed sensor into wind speed information and upload it to a cloud platform via a host. This system not only records and displays wind speed data in real-time but also features historical data storage and alarm functionalities, enhancing the intelligence and automation levels of wind speed monitoring. Additionally, users can access and analyze wind speed data through both computer and mobile devices, ensuring the system's efficiency and reliability. This work provides crucial technical support for the advancement of smart cities, clean energy, and environmental monitoring, thereby promoting the application and dissemination of IoT technology in the environmental field.

Drainage estimation across mountainous regions from large-scale soil moisture observations

Drainage is a crucial soil hydrological process that governs the partitioning of rainfall into runoff, groundwater recharge, soil water storage and evapotranspiration. Despite its significance, the drainage process is poorly understood due to the difficulty in direct measurements and insufficient understanding of its underlying physical mechanisms. To address these challenges, we present an innovative, physically-based, data-driven approach, SM2D (Soil Moisture to Drainage), to estimate drainage. SM2D was applied and examined using soil moisture data from a large-scale observation network over mountainous areas during 2014–2020. The soil moisture threshold governing drainage initiation proves to be significantly lower than the commonly employed field capacity metric in hydrological models. This threshold is influenced by factors such as mean soil moisture, bulk density, residual soil moisture, soil organic carbon, and parameters n and α of soil retention curve. Notably, field capacity has minimal impact on this threshold. Additionally, our analysis reveals that the drainage process is more influenced by the Soil Water Storage Increment (SWSI) than by mean soil moisture (MSM) that has traditionally been recognized as a key factor in drainage control. In comparison to commonly used exponential equations and those in models such as the Soil & Water Assessment Tool (SWAT), SM2D demonstrates superior performance in estimating drainage. The exponential equation derived from the SWSI outperforms those derived from other soil moisture metrics, including the commonly utilized MSM, challenging prevailing norms in drainage equations. SM2D holds the potential to generate extensive drainage datasets from satellite or large-scale soil moisture observations, advancing large-scale hydrological studies.

Age-related differences of the time-varying features in the brain functional connectivity and cognitive aging.

Brain functional modular organization changes with age. Considering the brain as a dynamic system, recent studies have suggested that time-varying connectivity provides more information on brain functions. However, the spontaneous reconfiguration of modular brain structures over time during aging remains poorly understood. In this study, we investigated the age-related dynamic modular reconfiguration using resting-state functional MRI data (615 participants, aged 18-88 years) from Cam-CAN. We employed a graph-based modularity analysis to investigate modular variability and the transition of nodes from one module to another in modular brain networks across the adult lifespan. Results showed that modular structure exhibits both linear and nonlinear age-related trends. The modular variability is higher in early and late adulthood, with higher modular variability in the association networks and lower modular variability in the primary networks. In addition, the whole-brain transition matrix showed that the times of transition from other networks to the dorsal attention network were the largest. Furthermore, the modular structure was closely related to the number of cognitive components and memory-related cognitive performance, suggesting a potential contribution to flexibility cognitive function. Our findings highlighted the notable dynamic characteristics in large-scale brain networks across the adult lifespan, which enhanced our understanding of the neural substrate in various cognitions during aging. These findings also provided further evidence that dedifferentiation and compensation are the outcomes of functional brain interactions.

Dynamic thermal simulation of a tree-shaped district heating network based on discrete event simulation

The computational complexity involved in modelling district heating (DH) networks impedes the integration of network operations into comprehensive DH system studies. We developed a flexible, accurate, and fast dynamic thermal simulation model utilising discrete event simulation (DES). This model is versatile, suitable for any tree-shaped DH network with a central heating plant and can estimate node temperatures and calculate pipe heat losses. The speed of the model is improved via using variable time steps and by incorporating two advanced techniques: lazy evaluation and a customised priority queue. To further improve the computational speed, we developed a technique to eliminate redundant sampling points. This model was tested and demonstrated excellent consistency with actual measurements. Remarkably, reducing sampling points can speed up the simulation by a factor of three without compromising the temperature accuracy. A 72-day simulation of a network with 102 pipes was completed within 0.219 s. Our findings highlight the significant potential of the DES model for large-scale dynamic network simulations and offer a promising solution for DH network simulations and system optimisation.

Multidimensional patterns of bird diversity and its driving forces in the Yangtze river basin of China

Biodiversity is fundamental to human well-being and economic development. The Yangtze River, the largest river in China, faces biodiversity loss due to habitat degradation, climate change, and other anthropogenic threats. However, the long-term changes in the region's biodiversity remain poorly understood. Here, we constructed an optimized living planet index (LPIO) by combining Partial Least Squares Structural Equation Modeling and Random Forest Modeling. Using data from a monitoring network of 536 sites, we observed an increasing trend in terrestrial bird diversity and functional complexity across the entire watershed from 2011 to 2020. Our findings indicate that a large-scale ecological restoration program has contributed to increases in terrestrial and aquatic bird diversity in the Yangtze River Basin. In contrast, bird diversity in the downstream area has decreased by 2.83 ​%, largely due to a rapid decline in wetland birds. The degradation of wetland habitats and insufficient conservation measures have negatively impacted bird diversity in the downstream region. This suggests that although there have been significant improvements in terrestrial bird diversity, more effective wetland restoration is necessary for biodiversity conservation. We recommend optimizing the national large-scale biodiversity monitoring network and increasing the number of upstream monitoring sites.

Global sustainable closed-loop supply chain considering Incoterms rules and advertisement impacts

Industrial information integration plays a crucial role in modern supply chains by ensuring the smooth flow of data across all stages, including recovery, recycling, and disposal, which is essential for the successful implementation of a closed-loop supply chain (CLSC) model. Building on this, our paper addresses a global CLSC problem by incorporating International Commercial Terms (Incoterms) and international transportation modes, bridging global supply chain operations with sustainability criteria. This innovative approach advances the development of a globally sustainable CLSC by focusing on the integration of economic, environmental, and social factors, i.e., the triple bottom line of sustainability. Specifically, we address environmental concerns through the introduction of carbon taxation and enhance social sustainability by exploring the impact of advertising on customer satisfaction. To further refine this model, we classify customers based on their sustainability engagement and apply a fuzzy programming approach to account for uncertainty in customer demand influenced by advertising. To solve this complex global CLSC model, we conduct a thorough analysis of constraints and develop a robust Lagrangian relaxation reformulation. While the initial solution may result in infeasibility, we propose a heuristic algorithm that ensures feasible solutions. Our efficient Lagrangian-based heuristic, incorporating an adaptive strategy, is capable of solving large-scale networks with an approximate 10% optimality gap. Ultimately, this research provides both a comprehensive framework for practitioners to improve the environmental performance and global operations of their supply chains, as well as significant theoretical contributions to the field of industrial information systems.

MQGA: A quantitative analysis of brain network hubs using multi-graph theoretical indices

Recent advancements in large-scale network studies have shown that connector hubs and provincial hubs are vital for coordinating complex cognitive tasks by facilitating information transfer between and within specialized modules. However, current methods for identifying these hubs often lack standardized measurement criteria, hindering quantitative analysis. This study proposes a novel computational method utilizing multi-graph theoretical index calculations to quantitatively analyze hub attributes in brain networks. Using benchmark network, random simulation network (N = 100), resting fMRI data from the ADHD-200 NYU dataset (HC = 110, ADHD = 146), and the Peking dataset (HC = 120, ADHD = 83), we introduce the Multi-criteria Quantitative Graph Analysis (MQGA) method, which employs betweenness centrality, degree centrality, and participation coefficient to determine the connector (con) hub index and provincial (pro) hub index. The method's accuracy, reliability, and stability were validated through correlation analysis of hub indices and labels, vulnerability tests, and consistency analysis across subjects. Results indicate that as network sparsity increases, the con hub index increases while the pro hub index decreases, with the optimal hub node index at 4 % sparsity. Vulnerability tests revealed that removing con nodes had a greater impact on network integrity than removing pro nodes. Both con and pro exhibited stability in consistency analyses, but con was more stable. The stability of hub scores in disease groups was significantly lower than in the healthy control group. High con values were found in the precuneus, postcentral gyrus, and precentral gyrus, whereas high pro values were identified in the precentral gyrus, postcentral gyrus, superior parietal lobule, precuneus, and superior temporal gyrus. This approach enhances the accuracy and sensitivity of hub node identification, facilitating precise comparisons and producing consistent, replicable results, advancing our understanding of brain network hub nodes, their roles in cognitive processes, and their implications for brain disease research.

Digital Twin Enhanced Multi-Agent Reinforcement Learning for Large-Scale Mobile Network Coverage Optimization

Digital Twin Enhanced Multi-Agent Reinforcement Learning for Large-Scale Mobile Network Coverage Optimization

축제 산업생태계 네트워크 구조 분석 : 명예 문화관광축제를 중심으로

The demand for the industrialization of festivals has become increasingly prominent both academically and in policy discussions. Approaching festivals from an industrial perspective is regarded as a key strategy for improving supply chain efficiency and expanding the festival business ecosystem. The demand for the industrialization of festivals is increasingly strong, both academically and in policy discussions. This study aims to analyze the structural characteristics (connectivity and cohesion) and positional characteristics (influence) of South Korea's festival industry through social network analysis, grounded in industry ecosystem theory. The research utilizes 595 expenditure data points from four major cultural tourism festivals in Korea. The data were preprocessed in Excel, using pivot tables to identify relationships between attributes. After data cleaning, the dataset was divided into node and edge files, which were analyzed using Gephi 0.10, an open-source Java-based software optimized for large-scale network analysis. This multidimensional approach was critical in visualizing and optimizing the network data. The findings reveal that Korea’s festival business ecosystem displays characteristics of early-stage industrialization, characterized by low network density and limited two-way interactions between actors. While cohesive clusters within the network show diversity, the influence of major networks remains low, highlighting the need for focused efforts to strengthen them. In terms of positional characteristics, festival organizations, venue construction industries, and content development companies were identified as influential actors. Academically, this study contributes to the empirical identification of the network structure within the festival business ecosystem, providing foundational data for understanding its dynamic changes. Practically, it offers policy recommendations, including fostering key actors to enhance cohesion and expanding opportunities for inter-company collaboration to further develop the ecosystem.

Real-Time Pipeline Fault Detection in Water Distribution Networks Using You Only Look Once v8.

Detecting faulty pipelines in water management systems is crucial for ensuring a reliable supply of clean water. Traditional inspection methods are often time-consuming, costly, and prone to errors. This study introduces an AI-based model utilizing images to detect pipeline defects, focusing on leaks, cracks, and corrosion. The YOLOv8 model is employed for object detection due to its exceptional performance in detecting objects, segmentation, pose estimation, tracking, and classification. By training on a large dataset of labeled images, the model effectively learns to identify visual patterns associated with pipeline faults. Experiments conducted on a real-world dataset demonstrate that the AI-based model significantly outperforms traditional methods in detection accuracy. The model also exhibits robustness to various environmental conditions such as lighting changes, camera angles, and occlusions, ensuring reliable performance in diverse scenarios. The efficient processing time of the model enables real-time fault detection in large-scale water distribution networks implementing this AI-based model offers numerous advantages for water management systems. It reduces dependence on manual inspections, thereby saving costs and enhancing operational efficiency. Additionally, the model facilitates proactive maintenance through the early detection of faults, preventing water loss, contamination, and infrastructure damage. The results from the three conducted experiments indicate that the model from Experiment 1 achieves a commendable mAP50 of 90% in detecting faulty pipes, with an overall mAP50 of 74.7%. In contrast, the model from Experiment 3 exhibits superior overall performance, achieving a mAP50 of 76.1%. This research presents a promising approach to improving the reliability and sustainability of water management systems through AI-based fault detection using image analysis.

Advancing Artificial Intelligence: The Potential of Brain-Inspired Architectures and Neuromorphic Computing for Adaptive, Efficient Systems

Brain-inspired AI architecture, also known as neuromorphic computing, seeks to emulate the structure and functionality of the human brain to create more efficient, adaptive, and intelligent systems. Unlike traditional AI models that rely on conventional computing frameworks, brain-inspired architectures leverage neural networks and synapse-like connections to perform computations more similarly to biological brains. This approach offers significant advantages, including lower power consumption, improved learning capabilities, and enhanced problem-solving efficiency, particularly in tasks that require complex pattern recognition and cognitive processes. This paper explores the key components of brain-inspired AI architectures, such as spiking neural networks (SNNs) and neuromorphic hardware, and reviews the latest advancements in this field. We examine their applications across diverse domains, including robotics, autonomous systems, and medical diagnostics, where brain-like adaptability and real-time learning are critical. Additionally, we analyze the challenges associated with scaling these architectures, including hardware constraints and the complexity of accurately mimicking human brain functionality. The potential for combining brain-inspired AI with current machine learning models is also discussed, highlighting future directions for achieving more advanced, efficient, and human-like artificial intelligence systems. This research contributes to the growing body of knowledge on neuromorphic computing and its promise in shaping the future of AI technologies.Moreover, brain-inspired AI architectures have the potential to surpass traditional AI systems in terms of real-time decision-making and learning efficiency, particularly in environments that require adaptive behavior. The use of spiking neural networks (SNNs) in these architectures allows for more biologically plausible models of neuron activity, which can lead to advancements in sensory processing and autonomous decision-making. As neuromorphic hardware continues to evolve, integrating it with existing AI frameworks could enhance both performance and scalability. However, replicating the brain's full complexity remains a significant challenge, particularly in terms of creating energy-efficient hardware capable of supporting large-scale neural networks. Despite these challenges, the development of brain-inspired AI promises to bridge the gap between artificial and human intelligence, offering transformative possibilities for various industries.

A novel optimization framework for natural gas transportation pipeline networks based on deep reinforcement learning

Natural gas is an emerging and reliable energy source in transition to a low-carbon economy. The natural gas transportation pipeline network systems are crucial when transporting natural gas from the production endpoints to processing or consuming endpoints. Optimizing the operational efficiency of compressor stations within pipeline networks is an effective way to reduce energy consumption and carbon emissions during transportation. This paper proposes an optimization framework for natural gas transportation pipeline networks based on deep reinforcement learning (DRL). The mathematical simulation model is derived from mass balance, hydrodynamics principles of gas flow, and compressor characteristics. The optimization control problem in steady state is formulated into a one-step Markov decision process (MDP) and solved by DRL. The decision variables are selected as the discharge ratio of each compressor. By the comprehensive comparison with dynamic programming (DP) and genetic algorithm (GA) in three typical element topologies (a linear topology with gun-barrel structure, a linear topology with branch structure, and a tree topology), the proposed method can obtain 4.60% lower power consumption than GA, and the time consumption is reduced by 97.5% compared with DP. The proposed framework could be further utilized for future large-scale network optimization practices.

FDG-PET patterns associate with survival in patients with prion disease.

Prion disease classically presents with rapidly progressive dementia, leading to death within months of diagnosis. Advances in diagnostic testing have improved recognition of patients with atypical presentations and protracted disease courses, raising key questions surrounding the relationship between patterns of neurodegeneration and survival. We assessed the contribution of fluorodeoxyglucose (FDG-PET) imaging for this purpose. FDG-PET were performed in 40 clinic patients with prion disease. FDG-PET images were projected onto latent factors generated in an external dataset to yield patient-specific eigenvalues. Eigenvalues were input into a clustering algorithm to generate data-driven clusters, which were compared by survival time. Median age at FDG-PET was 65.3 years (range 23-85). Median time from FDG-PET to death was 3.7 months (range 0.3-19.0). Four data-driven clusters were generated, termed "Neocortical" (n = 7), "Transitional" (n = 12), "Temporo-parietal" (n = 13), and "Deep nuclei" (n = 6). Deep nuclei and transitional clusters had a shorter survival time than the neocortical cluster. Subsequent analyses suggested that this difference was driven by greater hypometabolism of deep nuclei relative to neocortical areas. FDG-PET-patterns were not associated with demographic (age and sex) or clinical (CSF total-tau, 14-3-3) variables. Greater hypometabolism within deep nuclei relative to neocortical areas associated with more rapid decline in patients with prion disease and vice versa. FDG-PET informs large-scale network physiology and may inform the relationship between spreading pathology and survival in patients with prion disease. Future studies should consider whether FDG-PET may enrich multimodal prion disease prognostication models.

Optimizing data transmission in 6G software defined networks using deep reinforcement learning for next generation of virtual environments

Data transmission of Virtual Reality (VR) plays an important role in delivering a powerful VR experience. This increasing demand on both high bandwidth and low latency. 6G emerging technologies like Software Defined Network (SDN) and resource slicing are acting as promising technologies for addressing the transmission requirements of VR users. Efficient resource management becomes dominant to ensure a satisfactory user experience. The integration of Deep Reinforcement Learning (DRL) allows for dynamic network resource balancing, minimizing communication latency and maximizing data transmission rates wirelessly. Employing slicing techniques further aids in managing distributed resources across the network for different services as enhanced Mobile Broadband (eMBB) and Ultra-Reliable and Low Latency Communications (URLLC). The proposed VR-based SDN system model for 6G cellular networks facilitates centralized administration of resources, enhancing communication between VR users. This innovative solution seeks to contribute to the effective and streamlined resource management essential for VR video transmission in 6G cellular networks. The utilization of Deep Reinforcement Learning (DRL) approaches, is presented as an alternative solution, showcasing significant performance and feature distinctions through comparative results. Our results show that implementing strategies based on DRL leads to a considerable improvement in the resource management process as well as in the achievable data rate and a reduction in the necessary latency in dynamic and large scale networks.

Design and Performance Evaluation of an Authentic End-to-End Communication Model on Large-Scale Hybrid IPv4-IPv6 Virtual Networks to Detect MITM Attacks

After the end of IPv4 addresses, the Internet is moving towards IPv6 address architecture quickly with the support of virtualization techniques worldwide. IPv4 and IPv6 protocols will co-exist long during the changeover process. Some attacks, such as MITM attacks, do not discriminate by appearance and affect IPv4 and IPv6 address architectures. In an MITM attack, the attacker secretly captures the data, masquerades as the original sender, and sends it toward the receiver. The receiver replies to the attacker because the receiver does not authenticate the source. Therefore, the authentication between two parties is compromised due to an MITM attack. The existing authentication schemes adopt complicated mathematical procedures. Therefore, the existing schemes increase computation and communication costs. This paper proposes a lightweight and authentic end-to-end communication model to detect MITM attacks using a pre-shared symmetric key. In addition, we implement and analyze the performance of our proposed security model on Linux-based virtual machines connected to large-scale hybrid IPv4-IPv6 virtual networks. Moreover, security analyses prove the effectiveness of our proposed model. Finally, we compare the performance of our proposed security model with existing models in terms of computation cost and communication overhead.

HARNESSING GROUP THEORY FOR SIMPLIFIED ELECTRIC CIRCUIT ANALYSIS

Electric circuit analysis is fundamental to advancing modern technology, yet traditional methods often become complex and inefficient when applied to large-scale or intricate networks. Group theory, a branch of abstract algebra renowned for its capacity to exploit symmetry, has shown promise as a solution to this challenge. By applying group theoretical principles to circuit analysis, particularly through symmetry operations and transformation groups, engineers can simplify complex systems, reducing computational demands and revealing deeper insights into circuit behavior. Through systematic symmetry-based simplifications, group theory can decrease the number of equations required by up to 40% in certain configurations, enhancing both analytical efficiency and practical implementation. However, challenges include the mathematical complexity inherent to group theory, the limitations in analyzing highly irregular or nonlinear circuits, and the need for compatible computational tools. Recommendations include the integration of group theoretical approaches into standard circuit analysis software, focused training for electrical engineers, and further research to address the application of these methods in emerging technologies such as quantum and neuromorphic computing. Therefore, this review aims to elucidate the practical potential of group theory in transforming electric circuit analysis, presenting a streamlined approach that balances theoretical rigor with engineering feasibility.

Development of a Large-Scale Horizontal Curve Inventory for Safety Analysis Using Open GIS Data

Abstract Horizontal curves are crucial for the safety and mobility of roadways. To enhance the understanding of these curves, transportation agencies are investigating methods to accurately determine their spatial locations, geometric attributes, and conditions within their extensive road networks. Leveraging the widespread availability of Global Positioning System (GPS) and Geographic Information System (GIS) data, significant advancements have been made in developing automated algorithms that identify horizontal curves from GPS trajectories and GIS basemaps. This study introduces a novel dataset—the Horizontal Curve Inventory—that encapsulates detailed features of horizontal curves. We employed a modified automated cycle regression algorithm designed for large-scale road network analysis to extract horizontal curve data. This method was applied nationally to process an expansive network utilizing open-source GIS basemaps, encompassing over 300,000 miles across interstate, U.S. highway, and state highway systems. The inventory includes comprehensive data on curve radius, rotation angle, start and end point coordinates, and the curve’s center points. These data are essential for supporting safety research pertaining to road alignment. Our approach demonstrated robust performance across various open GIS data sources and provides the first accurate, efficient, and comprehensive inventory of horizontal curves for all major highway systems in the United States. The dataset is available at http://aichengbo.com/research/national-horizontal-curve-inventory/.

Comparison of whole-brain task-modulated functional connectivity methods for fMRI task connectomics

Higher brain functions require flexible integration of information across widely distributed brain regions depending on the task context. Resting-state functional magnetic resonance imaging (fMRI) has provided substantial insight into large-scale intrinsic brain network organisation, yet the principles of rapid context-dependent reconfiguration of that intrinsic network organisation are much less understood. A major challenge for task connectome mapping is the absence of a gold standard for deriving whole-brain task-modulated functional connectivity matrices. Here, we perform biophysically realistic simulations to control the ground-truth task-modulated functional connectivity over a wide range of experimental settings. We reveal the best-performing methods for different types of task designs and their fundamental limitations. Importantly, we demonstrate that rapid (100 ms) modulations of oscillatory neuronal synchronisation can be recovered from sluggish haemodynamic fluctuations even at typically low fMRI temporal resolution (2 s). Finally, we provide practical recommendations on task design and statistical analysis to foster task connectome mapping.

Endogenous opioid receptor system mediates costly altruism in the human brain

Functional neuroimaging studies suggest that a large-scale brain network transforms others’ pain into its vicarious representation in the observer, potentially modulating helping behavior. However, the neuromolecular basis of individual differences in vicarious pain and helping is poorly understood. We investigated the role of the endogenous μ-opioid receptor (MOR) system in altruistic costly helping. MOR density was measured using [11C]carfentanil. In a separate fMRI experiment, participants could donate money to reduce a confederate’s pain from electric shocks. Participants were generally willing to help, and brain activity was observed in amygdala, anterior insula, anterior cingulate cortex (ACC), striatum, primary motor cortex, primary somatosensory cortex and thalamus when witnessing others’ pain. Haemodynamic responses were negatively associated with MOR availability in emotion circuits. However, MOR availability positively associated with the ACC and hippocampus during helping. These findings suggest that the endogenous MOR system modulates altruism in the human brain.