Dense Network Research Articles

Thermally Conductive Polydimethylsiloxane-Based Composite with a Three-Dimensional Vertically Aligned Thermal Network Incorporating Hexagonal Boron Nitride Nanosheets and Nanodiamonds.

Thermal interface materials play a pivotal role in efficiently transferring heat from heating devices to thermal management components, thereby reducing the risk of component degradation due to overheating. In this study, we propose a strategy for enhancing the out-of-plane thermal conductivity (TC) of composite materials by fabricating a three-dimensional (3D) thermal network within a polydimethylsiloxane (PDMS) matrix. Specifically, the composite material was designed to incorporate a dense thermal network comprising hexagonal boron nitride nanosheets (BNNSs) and nanodiamonds (NDs). The fabrication process commenced with the preparation of BNNSs through liquid-phase exfoliation, followed by the creation of a 3D BNNSs-NDs/polyimide aerogel thermal framework using a unidirectional solidification ice templating method and subsequent heat treatment. Vacuum impregnation and curing were then employed to finalize the production of the 3D BNNSs-NDs/PDMS composite material. Characterization analyses indicated that the addition of NDs filled the voids between BNNSs, leading to the densification of the thermal framework pore walls and the establishment of additional thermal pathways. Impressively, with concentrations of BNNSs and NDs of 17.99 and 7.71 wt %, respectively, the out-of-plane TC of the 3D BNNSs-NDs/PDMS composite material reached 1.623 W m-1 K-1, marking notable enhancements of 754.21% and 256.70% compared to those of pure PDMS and composites prepared via direct blending with randomly distributed BNNSs and NDs, respectively. Furthermore, the 3D BNNSs-NDs thermal framework improved the compressive strength and the dimensional stability of the composite material. Finite element simulations additionally confirmed the synergistic improvement of the TC achieved through the combination of BNNSs and NDs, demonstrating that the 3D BNNSs-NDs/PDMS composite material displayed superior heat conduction and a greater density of thermal pathways compared to those of its counterparts, including 3D BNNSs/PDMS and 3D NDs/PDMS composite materials. In summary, this work presents a strategy for enhancing the out-of-plane TC of polymer-based composite materials by incorporating vertically aligned thermal networks.

SRARDA: A lightweight adaptive residual dense attention generative adversarial network for image super-resolution

Image super-resolution (SR) is the task of inferring a high resolution (HR) image from one/multiple single low resolution (LR) input(s). Traditional networks are evaluated by pixel-level metrics such as Peak-Signal-to-Noise Ratio (PSNR) etc., which do not always align with human perception of image quality. They often produce excessively smooth images that lack high-frequency texture and appear unnatural. Therefore, in this paper, we propose a lightweight adaptive residual dense attention generative adversarial network (SRARDA) for image SR. Firstly, our generator adopts the residual in residual (RIR) structure but redesigns the basic module. By using dynamic residual connection (ARC) to dynamically adjust the importance of residual and main paths, we design a novel adaptive residual dense attention block (ARDAB) that enhances the feature extraction capability of the generator. In addition, we build a high-frequency filtering unit (HFU) to extract more high-frequency features from the LR space. Finally, to fully utilize the discriminator, we use WGAN to compute the difference between the HR image and the reconstructed image. Experiments demonstrate that SRARDA effectively addresses the issue of excessive smoothing in reconstructed images, while also enhancing visual quality.

Sodium alginate supramolecular nanofibers in synergy with surface crack engineering to prepare tough and highly sensitive hydrogels

Soft and wet hydrogels often struggle to achieve both toughness and high sensitivity simultaneously, limiting their usefulness in flexible devices. To tackle this challenge, we devised a strategy that combines supramolecular sodium alginate nanofibers, utilizing Zr4+ as physical crosslinkers, with surface crack engineering via the micro-phase separation of polyaniline, to create a physically and chemically dual crosslinked polyacrylamide (PAM)/sodium alginate (SA)/polyaniline (PANI) hydrogel with exceptional toughness and high sensitivity. Owing to the supramolecular sodium alginate nanofibers, the dual crosslinked hydrogel exhibited a tensile strength of 0.391 MPa, an elongation at break of 568.9 %, and a toughness of 1.020 MJ/m3. The in-situ polymerized polyaniline layer, confined within the dense network, introduced micro-cracks onto the hydrogel surface, resulting in a high gauge factor of 11.4 for the fabricated hydrogel. Furthermore, integrating this hydrogel into a triboelectric nanogenerator transformed it into self-powered sensors capable of detecting external forces and generating various signals without power supply. These findings suggest that the developed hydrogel held great potential in diverse fields, including human motion detection, human-machine interaction, and wearable electronic devices.

Natural Language Processing to assess structure and complexity of system requirements

AbstractThe development process of a system is shaped by numerous variables that influence its progress and outcome. As a result, complexity can increase throughout the development process, potentially leading to negative consequences, which makes the management of complexity critical. Most development processes begin with the definition of needs and requirements. In this paper, the authors present a novel approach that enables the automated extraction of structure from requirements specifications. The approach uses Natural Language Processing to elicit three structural layers from a set of requirements, which are subsequently analyzed using metrics to assess complexity. In a case study, the approach is demonstrated using a set of 79 requirements, within which 246 individual entities are identified. These entities and the requirements are structured and analyzed using network density and spectral entropy. The metrics allow for interpretation and insight generation, such as identifying an increase in the number of potentially problematic loops. The approach achieved a detection and structural accuracy of over 98% in the given case study and is planned to be expanded with future cases.

Multilateral collaborative governance of Arctic shipping safety: Examining the impact of the Arctic Council via collaborative network analysis

Global climate change has dramatically reduced Arctic sea ice, opening new potential shipping routes, which has garnered international attention. This study explores the role of the Arctic Council in the collaborative governance network focused on Arctic shipping safety. For this purpose, a collection of 49 collaborative events initiated by international organisations and the Arctic Council was used as the data sample. These events are divided into two groups, one containing all the events and the other excluding those initiated by the Arctic Council. As a result, two collaborative networks are developed and evaluated based on network density, average path length, network transitivity, centrality indices (degree, closeness, betweenness), and clustering analysis. The results demonstrate that the Arctic Council significantly enhances the network density and transitivity of the international collaboration network, and reshapes core-periphery dynamics and cohesive subgroup structures. For instance, under the influence of the Arctic Council, Canada, France, and Japan surpass the United Kingdom in centrality (degree, closeness, betweenness). Results of the clustering coefficients reveal the Arctic Council's impact on connections among states, leading to shifts in collaboration patterns. Additionally, subgroup cohesion and density coefficients confirm the Arctic Council's role in promoting collaboration among states and restructuring governance networks.

GIAE-DTI: Predicting Drug-Target Interactions Based on Heterogeneous Network and GIN-based Graph Autoencoder.

Accurate prediction of drug-target interactions (DTIs) is essential for advancing drug discovery and repurposing. However, the sparsity of DTI data limits the effectiveness of existing computational methods, which primarily focus on sparse DTI networks and have poor performance in aggregating information from neighboring nodes and representing isolated nodes within the network. In this study, we propose a novel deep learning framework, named GIAE-DTI, which considers cross-modal similarity of drugs and targets and constructs a heterogeneous network for DTI prediction. Firstly, the model calculates the cross-modal similarity of drugs and proteins from the relationships among drugs, proteins, diseases, and side effects, and performs similarity integration by taking the average. Then, a drug-target heterogeneous network is constructed, including drug-drug interactions, protein-protein interactions, and drug-target interactions processed by weighted K nearest known neighbors. In the heterogeneous network, a graph autoencoder based on a graph isomorphism network is employed for feature extraction, while a dual decoder is utilized to achieve better self-supervised learning, resulting in latent feature representations for drugs and targets. Finally, a deep neural network is employed to predict DTIs. The experimental results indicate that on the benchmark dataset, GIAE-DTI achieves AUC and AUPR scores of 0.9533 and 0.9619, respectively, in DTI prediction, outperforming the current state-of-the-art methods. Additionally, case studies on four 5-hydroxytryptamine receptor-related targets and five drugs related to mental diseases show the great potential of the proposed method in practical applications.

Short-term PV energy yield predictions within city neighborhoods for optimum grid management

The global shift towards the exploitation of renewable energy sources is a key aspect in the transition to carbon-free societies. Although harnessing of PV energy disclosed the numerous environmental and economic benefits of PV systems, the ever-increasing share of PV systems in power generation has revealed their adverse effect on the electricity grid due to their dependence on weather conditions. Uncertain environmental parameters, especially cloud cover, can adversely affect the stability of the energy generated by a PV system, causing dynamic changes in PV power generation in a given future time period. Therefore, accurate predictions of PV power output are a major challenge, as they can contribute to the optimal management and flexibility of the power grid, and to the improvement of the operating efficiency of PV power stations, thus enabling the development of flexible green power electricity grids across cities. In addition, precise forecasting can provide system/grid operators with information needed for efficient capacity management and scheduling and ensure grid stability.Most forecasting models provide promising results in terms of error performance; however, they need weather variables or satellite information as inputs that make prognosis significantly challenging. Hence, the current work utilizes a data-driven method for predicting PV power generation from distributed PV systems. The spatio-temporal model Sparse Gaussian Conditional Random Fields was considered. Although this model was used extensively for wind forecasting, in this work, it was used for solar PV power output predictions. Moreover, the effect of varying spatial distribution for aggregated multiple PV systems was investigated on the forecast model performance. The main objective was to provide the solar PV energy sector with important information for building a smart grid for cities or regions.Continuous data from a grid-connected dense network of 21 PV systems was used within a region of 38.485 km2 in Nicosia, Cyprus. It should be noted that the test set includes data only from winter months. The results showed that the predicted power output signal responds to the fluctuation and follows the trend of the actual power signal, even on days with large variations in actual PV power. The average nRMSE of all PV systems in each dataset studied ranges from 0.119 to 0.146, proving that the proposed method delivered significant results that can be compared with existing results from similar studies. Overall, the proposed method can provide important information that can assist in decision-making for the management, optimization, and visualization of grids across cities.

Superhydrophobic Cellulose Nanofiber Aerogels for Efficient Hemostasis with Minimal Blood Loss.

Reducing unnecessary blood loss in hemostasis is a major challenge for traditional hemostatic materials due to uncontrolled blood absorption. Tuning the hydrophilic and hydrophobic properties of hemostatic materials provides a road to reduce blood loss. Here, we developed a superhydrophobic aerogel that enabled remarkably reduced blood loss. The aerogel was fabricated with polydopamine-coated and fluoroalkyl chain-modified bacterial cellulose via a directional freeze-drying method. Primarily, the hydrophobic feature prevented blood from uncontrolled absorption by the material and overflowing laterally. Additionally, the aerogel had a dense network of channels that allowed it to absorb water from blood due to the capillary effect, and fluoroalkyl chains trapped the blood cells entering the channels to form a compact barrier via hydrophobic interaction at the bottom of the aerogel, causing quick fibrin generation and blood coagulation. The animal experiments reveal that the aerogel reduced the hemostatic time by 68% and blood loss by 87 wt % compared with QuikClot combat gauze. The study demonstrates the superiority of superhydrophobic aerogels for hemostasis and provides new insights into the development of hemostatic materials.

A Novel Topology Adaptation Strategy for Dynamic Sparse Training in Deep Reinforcement Learning.

Deep reinforcement learning (DRL) has been widely adopted in various applications, yet it faces practical limitations due to high storage and computational demands. Dynamic sparse training (DST) has recently emerged as a prominent approach to reduce these demands during training and inference phases, but existing DST methods achieve high sparsity levels by sacrificing policy performance as they rely on the absolute magnitude of connections for pruning and randomly generating connections. Addressing this, our study presents a generic method that can be seamlessly integrated into existing DST methods in DRL to enhance their policy performance while preserving their sparsity levels. Specifically, we develop a novel method for calculating the importance of connections within the model. Subsequently, we dynamically adjust the sparse network topology by dropping existing connections and introducing new connections based on their respective importance values. Through validation on eight widely used simulation tasks, our method improves two state-of-the-art (SOTA) DST approaches by up to 70% in episode return and average return across all episodes under various sparsity levels.

Efficient computation of Katz centrality for very dense networks via negative parameter Katz

Abstract Katz centrality (and its limiting case, eigenvector centrality) is a frequently used tool to measure the importance of a node in a network, and to rank the nodes accordingly. One reason for its popularity is that Katz centrality can be computed very efficiently when the network is sparse, ie having only O(n) edges between its n nodes. While sparsity is common in practice, in some applications one faces the opposite situation of a very dense network, where only O(n) potential edges are missing with respect to a complete graph. We explain why and how, even for very dense networks, it is possible to efficiently compute the ranking stemming from Katz centrality for unweighted graphs, possibly directed and possibly with loops, by working on the complement graph. Our approach also provides an interpretation, regardless of sparsity, of ‘Katz centrality with negative parameter’ as usual Katz centrality on the complement graph. For weighted graphs, we provide instead an approximation method that is based on removing sufficiently many edges from the network (or from its complement), and we give sufficient conditions for this approximation to provide the correct ranking. We include numerical experiments to illustrate the advantages of the proposed approach.

Lightweight multi-scale feature dense cascade neural network for scene understanding of intelligent autonomous platform

Understanding the surrounding environment is an essential part of intelligent autonomous platform to complete autonomous driving or other autonomous tasks. Thereinto, scene understanding is a key technology to understand environmental information. For intelligent autonomous platform, it not only requires high accuracy, but also has a strict requirement for the inference speed. Moreover, it should be able to perform normally in complex environments such as target occlusion, the influence of illumination, and objects at different sizes. For tackling these problems, in this study, a lightweight multi-scale feature dense cascade neural network (LMFDCNet) is proposed in real-time scene understanding in complex scenes. Furthermore, trilateral fusion module is adopted to enhance low-level detailed features or high-level semantic features of each branch to describe the object better in different environments. A multi-scale upsampling module that fuses contextual information at different scales allows objects to be represented at different size. The results indicate that LMFDCNet produces 99.1 % classification accuracy on the tobacco dataset. On the Cityscapes dataset, we achieve 74.2 % Mean Intersection over Union (MIoU) with 13.6 M parameters at the speed of 38.2 FPS on a single 1070Ti card. To test the performance on the real-time application in the real world, LMFDCNet is deployed on the autonomous platform for real-time vision task such as classification of tobacco leaf state in curing process and semantic segmentation in the driving process of autonomous platform. The results reveal that LMFDCNet perform well on the AI embedded device and could be applied into intelligent autonomous platform.

Complex network analysis of cryptocurrency market during crashes

This paper identifies the cryptocurrency market crashes and analyses its dynamics using the complex network. We identify three distinct crashes during 2017–20, and the analysis is carried out by dividing the time series into pre-crash, crash, and post-crash periods. Partial correlation based complex network analysis is carried out to study the crashes. Degree density (ρD), average path length (l̄), and average clustering coefficient (cc¯) are estimated from these networks. We find that both ρD and cc¯ are smallest during the pre-crash period, and spike during the crash suggesting the network is dense during a crash. Although ρD and cc¯ decrease in the post-crash period, they remain higher than pre-crash levels for the 2017–18 and 2018–19 crashes suggesting a market attempt to return to normalcy. We get l̄ is minimal during the crash period, suggesting a rapid flow of information. A dense network and rapid information flow suggest that during a crash uninformed synchronized panic sell-off happens. However, during the 2019–20 crash, the values of ρD, cc¯, and l̄ did not vary significantly, indicating minimal change in dynamics compared to other crashes. The findings of this study may guide investors in making decisions during market crashes.

Unraveling cereal physical barriers composed of cell walls and protein matrix: Insights from structural changes and starch digestion

Physical barriers composed of cell walls and protein matrix in cereals, as well as their cooking changes, play important roles in starch digestion. In this study, the physical barriers of native and cooked highland barley (HB), brown rice (BR), and oats (OA) kernels and their contribution to starch digestion were investigated. The resistant starch content was similar in cereal flours, but varied among cooked kernels (HB > BR > OA: 45.05 %, 10.30 %, and 24.71 %). The water adsorption, gelatinization enthalpy, and decrease in hardness of HB kernels were lower than those of OA and BR kernels. Microstructural observations of native kernels showed that HB had the thickest cell walls. After cooking, the lowest cell wall deformation and a dense continuous network developed from the protein matrix were observed in HB kernels. During digestion, undigested starch granules encapsulated by the stable cell walls and strong protein network were observed in HB kernels, but not in BR or OA kernels. Furthermore, the heavily milled HB kernels still had more resistant starch than the intact OA and BR kernels. Therefore, the physical barriers of HB kernels exhibited stronger inhibition of starch gelatinization and digestion. Differences in cereal physical barriers led to various inhibitory effects.

Low-rank sparse fully-connected tensor network for tensor completion

Fully-connected tensor network (FCTN) has recently drawn lots of attention in tensor completion due to its full description of all correlations between any two modes. However, the FCTN model has multiple ranks, and existing methods often ignore the difficulty brought about by rank selection, especially when the model is rank-sensitive. To overcome this drawback, this paper proposed a low-rank sparse FCTN for tensor completion. Specifically, we theoretically show that, for a tensor with FCTN structure, its subtensor can be represented as a coefficient sum of basic tensors, where the coefficients are remained in the FCTN’s factors. This means that factors’ sparsity can improve the robustness of FCTN to larger rank selection. Moreover, according to the relation between the target tensor and its FCTN’s factors, low-rank constrain is used to enhance rank robustness. Lastly, we optimize the proposed model by the alternating direction method of multipliers (ADMM) algorithm. Experimental results show that the low-rank sparse constraint effectively improves the rank robustness of the FCTN model, and achieves excellent results compared with other state-of-the-art completion methods.

Tracking aquatic non-native macroinvertebrate species in Germany using long-term data

Biological invasions pose a global challenge, threatening both biodiversity and human well-being. Projections suggest that as invasions increase, the financial costs associated with management and the ecological harm they cause will also escalate. Here, we examined whether long-term biomonitoring strategies were adequate to identify and track benthic aquatic non-native macroinvertebrate species by using the German subset (151 time series; 129 of which reported non-native species) of the currently most comprehensive European long-term dataset of 1816 macroinvertebrate community time series from 22 European countries. The detection of aquatic non-native species was directly linked to the availability of long-term sites and thus, monitoring effort, having identified the spatio-temporal occurrence of 32 non-native species. The available long-term monitoring site data were mostly concentrated in the western part of Germany, predominantly covering the Rhine River and its tributaries. The spatially biased network of long-term monitoring sites, therefore, naturally skews the detection and reporting of aquatic non-native species toward this area and underestimates Eastern and Southern regions, impeding the comprehension of invasion dynamics. However, based on the available data, we found that the absolute number of non-native species increased and the proportion of non-native species relative to native species decreased over time. This indicates complex ecological interactions between native and non-native species and underlines the value of long-term data for investigating invasion dynamics. Considering the value of comprehensive monitoring networks, a spatially biased network delays the application of management and mitigation plans, possibly worsening the ecological and economic effects of biological invasions in Germany. The results provided here indicate the disadvantages of biased datasets, but simultaneously underline the enormous potential of a dense network of long-term monitoring. Our results also highlight the urgent need to increase and diversify long-term biomonitoring efforts throughout Germany to cover the main freshwater resources and their connections where the introduction risk of non-native species is the highest. Centrally collating such data would provide a profound basis for the monitoring of spreading aquatic non-native species and could serve the implementation of national biosecurity efforts.

Insights into the fabrication and antibacterial effect of fibrinogen hydrolysate-carrageenan loading apigenin and quercetin composite hydrogels

Escherichia coli and Staphylococcus aureus are the most prevalent pathogenic bacteria, often resulting in the foodborne disease outbreaks through food spoilage and foodborne infections. To prevent and control food spoilage and foodborne infections induced by Escherichia coli and Staphylococcus aureus, the antibacterial hydrogels were fabricated using fibrinogen hydrolysate-carrageenan (AHs-C) and flavonoids (apigenin and quercetin), and the antibacterial effect of the composite hydrogels against Escherichia coli and Staphylococcus aureus was further investigated. The results of mechanical property exhibited that the composite hydrogels with 0.2 % of apigenin and quercetin (AHs-C-Ap/Que) showed the highest hardness and swelling property compared with the separate addition of apigenin or quercetin. Scanning electron microscopy and atomic force microscopy showed that the dense networks were formed in the hydrogels of AHs-C-Ap/Que., and the average roughness of AHs-C-Ap/Que. significantly increased to 30.70 nm compared with AHs-C. 1H NMR and FTIR spectra demonstrated that apigenin and quercetin were bound to AHs-C by hydrogen bond, hydrophobic interaction and Schiff base, where the interactions between Ap/Que. and AHs-C was stronger compared with the separate addition of apigenin or quercetin. The hydrogels of AHs-C-Ap/Que. showed the highest antibacterial capacity and antibacterial adhesion against Escherichia coli and Staphylococcus aureus. The antibacterial adhesion assay showed that 99 % removal ratios for E. coli and S. aureus were observed in AHs-C-Ap/Que. hydrogels, which showed a great potential to prevent food spoilage and foodborne infections.

Routing algorithm for sparse unstructured P2P networks using honey bee behavior

SummaryUnstructured peer‐to‐peer (P2P) networks pose unique challenges for efficient and scalable routing. In this study, we introduce a novel routing algorithm inspired by the foraging behavior of honey bees named Honey Bee Optimization in P2P Networks (HBO_P2P) to address the inherent limitations of routing in unstructured P2P networks, focusing on improving packet delivery, minimizing hop count, reducing message overhead, and optimizing overall throughput. To evaluate the performance of our proposed algorithm, we conducted comprehensive experiments comparing it with existing algorithms commonly used in P2P networks, namely, particle swarm optimization (PSO), genetic algorithm (GA), and ant colony optimization (ACO). After the simulation, we got the results as follows: Our algorithm outperforms ACO, GA, and PSO by exhibiting the highest number of data hops indicating potential efficiency in route optimization. Routing overhead is also minimal as compared to ACO, GA, and PSO. The average data packet delay is also low in our algorithm as compared to ACO, GA, and PSO. HBO_P2P achieves the highest throughput, nearly reaching 100 Mbps. While ACO and GA exhibit similar throughput of around 80 Mbps, and PSO has the lowest throughput, approximately 60 Mbps.

Information Support or Emotional Support? Social Support in Online Health Information Seeking among Chinese Older Adults.

Online Health Information Seeking (OHIS) serves as an alternative form of social capital that can help older adults alleviate offline medical-related stress. This study collected and analyzed user interaction data from Patient-to-Doctor and Patient-to-Peer platforms and compared the roles of social support between them. Significant differences were identified in the dimensions of social support (information, emotional, and companion) on the Patient-to-Peer platforms compared with Patient-to-Doctor platforms (p < 0.05). The overall and core-core network density values for social support on Patient-to-Peer platforms were higher than those on Patient-to-Doctor platforms. Patient-to-Doctor interactions focused on information support, displaying a more centralized and efficient network with structural holes pertaining to treatment effects. By contrast, Patient-to-Peer interactions provided more emotional support, with a dispersed and redundant network containing structural holes related to individual information. Companion support was found to be weaker on both platforms. Additionally, digital literacy, surrogate seeking, and altruistic information significantly explained the variances between the two platforms (p < 0.01), with surrogate seeking playing a crucial role. These findings enhance our understanding of OHIS disparities among older adults and their surrogates, offering valuable insights for developing effective support systems and regulatory frameworks for health information platforms.

Spectrum Sensing Method Based on STFT-RADN in Cognitive Radio Networks.

To address the common issues in traditional convolutional neural network (CNN)-based spectrum sensing algorithms in cognitive radio networks (CRNs), including inadequate signal feature representation, inefficient utilization of feature map information, and limited feature extraction capabilities due to shallow network structures, this paper proposes a spectrum sensing algorithm based on a short-time Fourier transform (STFT) and residual attention dense network (RADN). Specifically, the RADN model improves the basic residual block and introduces the convolutional block attention module (CBAM), combining residual connections and dense connections to form a powerful deep feature extraction structure known as residual in dense (RID). This significantly enhances the network's feature extraction capabilities. By performing STFT on the received signals and normalizing them, the signals are converted into time-frequency spectrograms as network inputs, better capturing signal features. The RADN is trained to extract abstract features from the time-frequency images, and the trained RADN serves as the final classifier for spectrum sensing. Experimental results demonstrate that the STFT-RADN spectrum sensing method significantly improves performance under low signal-to-noise ratio (SNR) conditions compared to traditional deep-learning-based methods. This method not only adapts to various modulation schemes but also exhibits high detection probability and strong robustness.

How Local Pollution and Transboundary Air Pollution Impact Air Quality in Jakarta?

Jakarta’s rapid population growth and industrialization have severely impacted air quality, with limited monitoring stations and increasingly complex pollutant sources influenced by meteorological factors. Source apportionment studies are critical for addressing urban air quality issues because knowing the source of pollution is an important step in overcoming urban air quality problems. This study uses the Geographically Weighted Regression (GWR) and Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) models to analyze the distribution and sources of NO2, SO2, and CO in Jakarta and surrounding areas. Data includes spatial information from road networks, industries, and the energy sector during the wet (December) and dry (June) months of 2023. GWR results show that network density significantly contributes to pollutant concentrations, particularly during the dry month, while industrial and energy sectors display seasonal variations. The HYSPLIT model indicates that in December, air masses carrying pollutants mainly originate from the northwest and north, and in June, from the southeast and southwest. These patterns underscore the regional influence of transboundary air pollution from industrial and energy plants. The combined GWR and HYSPLIT findings provide a comprehensive understanding of air pollution dynamics and the critical role of meteorological factors in pollutant distribution.