Design Of Network Structure Research Articles

Research and Application of ROM Based on Res-PINNs Neural Network in Fluid System

In the design of fluid systems, rapid iteration and simulation verification are essential, and reduced-order modeling techniques can significantly improve computational efficiency and accuracy. However, traditional Physics-Informed Neural Networks (PINNs) often face challenges such as vanishing or exploding gradients when learning flow field characteristics, limiting their ability to capture complex fluid dynamics. This study presents an enhanced reduced-order model (ROM): Physics-Informed Neural Networks based on Residual Networks (Res-PINNs). By integrating a Residual Network (ResNet) module into the PINN architecture, the proposed model improves training stability while preserving physical constraints. Additionally, the model’s ability to capture and learn flow field states is further enhanced by the design of a symmetric parallel neural network structure. To evaluate the effectiveness of the Res-PINNs model, two classic fluid dynamics problems—flow around a cylinder and Vortex-Induced Vibration (VIV)—were selected for comparative testing. The results demonstrate that the Res-PINNs model not only reconstructs flow field states with higher accuracy but also effectively addresses limitations of traditional PINN methods, such as vanishing gradients, exploding gradients, and insufficient learning capacity. Compared to existing approaches, the proposed Res-PINNs provide a more stable and efficient solution for deep learning-based reduced-order modeling in fluid system design.

RCFormer: radiation correction method for degraded multispectral UAV images using vision transformers

ABSTRACT In modern remote sensing technology, Unmanned Aerial Vehicles (UAVs) have become crucial data acquisition platforms. However, low-altitude remote sensing images from UAVs are often affected by atmospheric scattering and absorption, particularly under conditions like haze, which severely degrade image quality. Traditional radiation correction methods cannot accurately capture the true reflectance of ground objects under these conditions. This paper introduces RCFormer, an innovative Transformer-based network architecture. It combines the self-attention mechanism of Swin Transformer with a novel network structural design, especially incorporating parallel convolutional structures to enhance the extraction capability of surface features. Additionally, with the integration of atmospheric characteristic refinement module (ACRM) and Multi-scale Feature Fusion Gate (MFFG), the proposed RCFormer not only captures atmospheric information from deep features but also effectively fuses features across different scales, thereby improving the accuracy, transferability and robustness of the radiation correction model. Finally, quantitative and qualitative evaluations using real-world data demonstrate that RCFormer outperforms other dehazing networks in both degraded and haze-free images while also offering good efficiency and ease of operation.

Integrating Formal and Relational Contracting: The Link Between Network Structure and Contract Design in Interlocal Collaboration Agreements

ABSTRACTLocal governments increasingly rely on contractual relationships with third parties for public service delivery, blending formal agreements with relational mechanisms to reduce risks and uncertainties. While most research on these contracts has focused on relationships between two parties, this work adopts a network perspective to analyze how the structure of these networks—especially close‐knit connections—influences contract design. We analyzed 6576 agreements between local governments in Iowa related to public safety from 2007 to 2017 and selected 300 to qualitatively code for contractual provisions. Our findings show that when there are strong shared connections between third parties, contracts tend to be less formal. These results expand the typical focus on two‐party relationships, showing how larger networks can shape the balance between formal and relational contracts in public service management.

CC4S: Encouraging Certainty and Consistency in Scribble-Supervised Semantic Segmentation.

Deep learning-based solutions have achieved impressive performance in semantic segmentation but often require large amounts of training data with fine-grained annotations. To alleviate such requisition, a variety of weakly supervised annotation strategies have been proposed, among which scribble supervision is emerging as a popular one due to its user-friendly annotation way. However, the sparsity and diversity of scribble annotations make it nontrivial to train a network to produce deterministic and consistent predictions directly. To address these issues, in this paper we propose holistic solutions involving the design of network structure, loss and training procedure, named CC4S to improve Certainty and Consistency for Scribble-Supervised Semantic Segmentation. Specifically, to reduce uncertainty, CC4S embeds a random walk module into the network structure to make neural representations uniformly distributed within similar semantic regions, which works together with a soft entropy loss function to force the network to produce deterministic predictions. To encourage consistency, CC4S adopts self-supervision training and imposes the consistency loss on the eigenspace of the probability transition matrix in the random walk module (we named neural eigenspace). Such self-supervision inherits the category-level discriminability from the neural eigenspace and meanwhile helps the network focus on producing consistent predictions for the salient parts and neglect semantically heterogeneous backgrounds. Finally, to further improve the performance, CC4S uses the network predictions as pseudo-labels and retrains the network with an extra color constraint regularizer. From comprehensive experiments, CC4S achieves comparable performance to those from fully supervised methods and shows promising robustness under extreme supervision cases.

AutoLDT: a lightweight spatio-temporal decoupling transformer framework with AutoML method for time series classification

Time series classification finds widespread applications in civil, industrial, and military fields, while the classification performance of time series models has been improving with the recent development of deep learning. However, the issues of feature extraction effectiveness, model complexity, and model design uncertainty constrain the further development of time series classification. To address the above issues, we propose a Lightweight Spatio-Temporal Decoupling Transformer framework based on Automated Machine Learning technique (AutoLDT). The framework proposes a novel lightweight Transformer with fuzzy position encoding, TS-separable linear self-attention mechanism, and convolutional feedforward network, which mine the temporal and spatial features, as well as the local and global relationship of time series. Fuzzy positional encoding integrates fuzzy ideas to enhance the generalization performance of model information mining. TS-separable linear self-attention mechanism and convolutional feedforward network achieve feature extraction in a lightweight way by decoupling temporal and spatial features of time series. Notably, we adopt the Covariance Matrix Adaptation Evolution Strategy and global adaptive pruning technique to realize automated network structure design, which further improves the model training efficiency and automation, and avoids the uncertainty problem of network design. Finally, we validate the effectiveness of the proposed framework on publicly available UCR and UEA time series datasets. The experimental results show that the proposed framework not only improves the model classification performance in a lightweight way but also dramatically improves the model training efficiency.

Jellyfish-Inspired Polyurea Ionogel with Mechanical Robustness, Self-Healing, and Fluorescence Enabled by Hyperbranched Cluster Aggregates.

Ionogels are promising for soft iontronics, with their network structure playing a pivotal role in determining their performance and potential applications. However, simultaneously achieving mechanical toughness, low hysteresis, self-healing, and fluorescence using existing network structures is challenging. Drawing inspiration from jellyfish, we propose a novel hierarchical crosslinking network structure design for in situ formation of hyperbranched cluster aggregates (HCA) to fabricate polyurea ionogels to overcome these challenges. Leveraging the disparate reactivity of isocyanate groups, we induce the in situ formation of HCA through competing reactions, enhancing toughness and imparting the clustering-triggered emission of ionogel. This synergy between supramolecular interactions in the network and plasticizing effect in ionic liquid leads to reduced hysteresis of the ionogel. Furthermore, the incorporation of NCO-terminated prepolymer with dynamic oxime-urethane bonds (NPU) enables self-healing and enhances stretchability. Our investigations highlight the significant influence of HCA on ionogel performance, showcasing mechanical robustness including high strength (3.5 MPa), exceptional toughness (5.5 MJ m-3), resistance to puncture, and low hysteresis, self-healing, as well as fluorescence, surpassing conventional dynamic crosslinking approaches. This network design strategy is versatile and can meet the various demands of flexible electronics applications.

Automatic architecture design for distributed quantum computing

Abstract In distributed quantum computing (DQC), quantum hardware design mainly focuses on providing as many as possible high-quality inter-chip connections. Meanwhile, quantum software tries its best to reduce the required number of remote quantum gates between chips. However, this “hardware first, software follows” methodology may not fully exploit the potential of DQC. Inspired by classical software–hardware co-design, this paper explores the design space of application-specific DQC architectures. More specifically, we propose AutoArch, an automated quantum chip network (QCN) structure design tool. With qubits grouping followed by a customized QCN design, AutoArch can generate a near-optimal DQC architecture suitable for target quantum algorithms. Experimental results show that the DQC architecture generated by AutoArch can outperform other general QCN architectures when executing target quantum algorithms.

Ultra-high stretchability and shape fixation rate shape memory polyurethanes based on cyclic polytetrahydrofuran molecular rings

Chemical cross-linking is commonly used to prevent slippage between molecular chains in shape memory polymers (SMPs) to improve shape return. However, chemical cross-linking makes SMPs less stretchable, and the disordered network structure reduces the ability of SMPs to maintain temporary shapes. To obtain ultra-high stretchability and better shape memory properties, a cyclic polymer (C-PTHF-OH) was introduced into the shape memory polyurethane (PUCX) network, and the PUCX network topology was controlled by adjusting the content of C-PTHF-OH molecular rings. PUC0.5 exhibited the highest shape fixation (99.9 %) and shape recovery (98.4 %), and the higher the content of the C-PTHF-OH molecular ring, the higher the elongation at break of the prepared PUCX, with a slight decrease in tensile strength. Compared to PUC0 (2000 % elongation at break and 32 MPa tensile strength) prepared from the linear polymer, PUC0.5 showed up to 2150 % elongation at break and 31 MPa tensile strength. This study provides new ideas for the design of network structures for SMPs and is a new paradigm introduced into the SMPs network by cyclic topological polymers.

Jellyfish‐Inspired Polyurea Ionogel with Mechanical Robustness, Self‐Healing, and Fluorescence Enabled by Hyperbranched Cluster Aggregates

AbstractIonogels are promising for soft iontronics, with their network structure playing a pivotal role in determining their performance and potential applications. However, simultaneously achieving mechanical toughness, low hysteresis, self‐healing, and fluorescence using existing network structures is challenging. Drawing inspiration from jellyfish, we propose a novel hierarchical crosslinking network structure design for in situ formation of hyperbranched cluster aggregates (HCA) to fabricate polyurea ionogels to overcome these challenges. Leveraging the disparate reactivity of isocyanate groups, we induce the in situ formation of HCA through competing reactions, enhancing toughness and imparting the clustering‐triggered emission of ionogel. This synergy between supramolecular interactions in the network and plasticizing effect in ionic liquid leads to reduced hysteresis of the ionogel. Furthermore, the incorporation of NCO‐terminated prepolymer with dynamic oxime–urethane bonds (NPU) enables self‐healing and enhances stretchability. Our investigations highlight the significant influence of HCA on ionogel performance, showcasing mechanical robustness including high strength (3.5 MPa), exceptional toughness (5.5 MJ m−3), resistance to puncture, and low hysteresis, self‐healing, as well as fluorescence, surpassing conventional dynamic crosslinking approaches. This network design strategy is versatile and can meet the various demands of flexible electronics applications.

Enhancing accuracy and interpretability of multi-steps water demand prediction through prior knowledge integration in neural network architecture

In the field of water supply management, multi-steps water demand forecasting plays a crucial role. While there have been many studies related to multi-steps water demand forecasting based on deep learning, little attention has been paid to the interpretability of forecasting models. Aiming to improve both the forecasting accuracy and interpretability of the model, a novel urban water demand forecasting neural network (UWDFNet) was presented in this paper. Compared with traditional deep learning models, it innovatively considered domain-specific prior knowledge from water supply management and incorporated the correlation relationship between different input variables into the design of the neural network structure, and verified the consistency between the knowledge learned by the model and prior knowledge through interpretability analysis. Additionally, a systematic performance evaluation was conducted and proved that UWDFNet possesses better accuracy and stability compared to other baseline models(e.g., gated recurrent unit network (GRUN), GRUN with a corrected Network (GRUN+CORRNet), GRUN+PID, GRUN+Kmeans).

Dual-scale shifted window attention network for medical image segmentation

Swin Transformer is an important work among all the attempts to reduce the computational complexity of Transformers while maintaining its excellent performance in computer vision. Window-based patch self-attention can use the local connectivity of the image features, and the shifted window-based patch self-attention enables the communication of information between different patches in the entire image scope. Through in-depth research on the effects of different sizes of shifted windows on the patch information communication efficiency, this article proposes a Dual-Scale Transformer with double-sized shifted window attention method. The proposed method surpasses CNN-based methods such as U-Net, AttenU-Net, ResU-Net, CE-Net by a considerable margin (Approximately 3% ∼\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\sim$$\\end{document} 6% increase), and outperforms the Transformer based models single-scale Swin Transformer(SwinT)(Approximately 1% increase), on the datasets of the Kvasir-SEG, ISIC2017, MICCAI EndoVisSub-Instrument and CadVesSet. The experimental results verify that the proposed dual scale shifted window attention benefits the communication of patch information and can enhance the segmentation results to state of the art. We also implement an ablation study on the effect of the shifted window size on the information flow efficiency and verify that the dual-scale shifted window attention is the optimized network design. Our study highlights the significant impact of network structure design on visual performance, providing valuable insights for the design of networks based on Transformer architectures.

Congestion aware clustered WSN based on an improved ant colony algorithm

Conventional works carried out in Wireless Sensor Networks (WSN) mostly focussed on energy oriented services and very less significant measures given to delay oriented and congestion aware services. Hence the proposed mechanism specially focuses on network structural design by placing rendezvous location for each cluster as well as route segmentation for controlling the congestion occurrence and unwanted delay. Here Congestion Aware Clustering with Improved Ant Colony Algorithm (CAC_IACA) is proposed. This mechanism involves two steps (i) identifying the best route by following the Ant Colony Optimization (ACO) algorithm and (ii) data segmentation using rendezvous mobile nodes. The Rendezvous nodes are present in each cluster to reduce the congestion rate on receiver side during data transmission. This proposed methodology mainly concentrates on reducing coverage cost for 3D environmental monitoring. Simulation results are analysed and the efficiency of the proposed scheme proves 26.54 % better than the conventional method.

First-passage properties of bundled networks.

Bundled networks, obtained by attaching a copy of a fiber structure to each node on the base structure, serve as important realistic models for the geometry and dynamics of nontranslationally invariant systems in condensed matter physics. Here, we analyze the first-passage properties, including the mean first-passage time, the mean-trapping time, the global-mean first-passage time (GFPT), and the stationary distribution, of a biased random walk within such networks, in which a random walker moves to a neighbor on base with probability γ and to a neighbor on fiber with probability 1-γ when the walker at a node on base. We reveal the primary properties of both the base and fiber structure, which govern the first-passage characteristics of the bundled network. Explicit expressions between these quantities in the bundled networks and the related quantities in the component structures are presented. GFPT serves as a crucial indicator for evaluating network transport efficiency. Unexpectedly, bases and fibers with similar scaling of GFPT can construct bundled networks exhibiting different scaling behaviors of GFPT. Therefore, bundled networks can be tailored to accommodate specific dynamic property requirements by choosing a suitable base and fiber structure. These findings contribute to advancing the design and optimization of network structures.

Geoeconomic Dynamics in a New Economic Global Order from West to East

Abstract The global economic landscape is in a state of flux, with the COVID-19 pandemic starkly highlighting the interdependent yet unequal and unstable nature of the global political economy. Moving beyond traditional geopolitical considerations of conflict, state-centricism, and security concerns, the current environment is witnessing the emergence of new hegemonic rivalries involving established, rising, and aspiring powers. This shift is not merely theoretical but practical, characterized more by geoeconomic dynamics than traditional geopolitical factors. Moreover, we attribute the diversity in state strategies to a blend of the network's structural design, on one hand, and the domestic institutions and norms of states seeking to leverage these network structures, on the other hand. The full exploitation of the advantages of weaponized interdependence is reserved for states with physical or legal control over hub nodes. Entities possessing such power currently include the United States, the European Union, and, increasingly, China, all these being, capable to enjoy the benefits of weaponized interdependence, even though other actors may still play influential roles. This research investigates the origins, outcomes, and key players shaping geoeconomics on a global scale within the context of a transforming global order. The often-mentioned normative force in foreign policies of the EU, the largest singularly unified economic area worldwide, tends to be overshadowed by its lack of homogeneity and perspective diversity. This gap is filled by emerging regional and global powers. Furthermore, Argentina's decision not to join the group starting from 2024 raises the question which side tends to gain majority within Western vs. Eastern challenge: whether the discounted Argentina or the newly included top oil exporters Saudi Arabia, Iran, Egypt, Ethiopia, and the United Arab Emirates.

Metallic surface defect recognition network based on global feature aggregation and dual context decoupled head

Surface defect detection is a crucial inspection phase in ensuring industrial product quality and reliability, particularly in the context of metallic manufacturing components. While existing deep learning-based approaches have demonstrated some effectiveness, the design of certain network structures still lacks consideration for industrial scenarios. This paper proposed a complex metallic surface defect detection neural network which incorporates two innovatively designed modules along with other targeted enhancements. Firstly, the spatial pyramid pooling (SPP) structure is redesigned for richer global-local information fusion. A global feature fusion and redistribution module (GFAR) is designed to fully leverage feature information across different scales. With the combination of a novel global positional attention mechanism (GPAM), GFAR is capable of distinguishing defects from complex backgrounds. Moreover, the asymmetric convolution blocks (ACB) are employed to process the final features, enhancing the representational capacity for intricate defect shapes. Subsequently, the detection head is decoupled in both spatial and task domains to addresses the problems posed by inconsistent focus between classification and regression tasks. Finally, the reassignment of loss functions is undertaken for each respective task to enhance their alignment with the specific requirements of the defect detection task. Extensive ablation experiments are conducted on two steel surface defect datasets, NEU-DET and GC10-DET, to demonstrate the effectiveness of each module. The defects detection accuracy of our method respectively reaches 80.9 % and 84.8 % for the two datasets, which outperforms the majority of state-of-the-art detection neural networks.

Hybrid-Supervised Dual-Search: Leveraging Automatic Learning for Loss-Free Multi-Exposure Image Fusion

Multi-exposure image fusion (MEF) has emerged as a prominent solution to address the limitations of digital imaging in representing varied exposure levels. Despite its advancements, the field grapples with challenges, notably the reliance on manual designs for network structures and loss functions, and the constraints of utilizing simulated reference images as ground truths. Consequently, current methodologies often suffer from color distortions and exposure artifacts, further complicating the quest for authentic image representation. In addressing these challenges, this paper presents a Hybrid-Supervised Dual-Search approach for MEF, dubbed HSDS-MEF, which introduces a bi-level optimization search scheme for automatic design of both network structures and loss functions. More specifically, we harness a unique dual research mechanism rooted in a novel weighted structure refinement architecture search. Besides, a hybrid supervised contrast constraint seamlessly guides and integrates with searching process, facilitating a more adaptive and comprehensive search for optimal loss functions. We realize the state-of-the-art performance in comparison to various competitive schemes, yielding a 10.61% and 4.38% improvement in Visual Information Fidelity (VIF) for general and no-reference scenarios, respectively, while providing results with high contrast, rich details and colors. The code is available at https://github.com/RollingPlain/HSDS_MEF.

Multi-network PVA/SA aerogel fiber: Unveiling superior mechanical, thermal insulation, and extreme condition stability

Aerogels are highly sought-after in thermal insulation textiles due to their exceptional porosity and low density. However, Current research on aerogels mainly centers on their fixed two- and three-dimensional structures, lacking design flexibility, which significantly restricts their practicality. In contrast, one-dimensional aerogel fibers, while offering design flexibility, face complex fabrication, challenges in continuous molding, and pore collapse issues. In this study, utilizing polyvinyl alcohol (PVA) and sodium alginate (SA) as raw materials, a cost-effective wet spinning, freeze-thaw cycling, and freeze-drying technique were employed to continuously produce multi-functional PVA/SA aerogel fibers with a stable network pore structure through the design of a multi-crosslinked interlocking network structure. The optimal preparation process for the aerogel fibers was explored and characterized. The resulting aerogel fibers exhibited a stable hierarchical pore structure at the microscale, featuring large pores, nested mesopores, and micropores. These fibers possessed a high specific surface area (115.88 m²/g), high porosity (92.42 %), low density (0.0632 g/cm³), low thermal conductivity (0.0378 W/(m·K)), and flame resistance (LOI: 26.8 %). Furthermore, their stable internal network scaffold structure endowed them with outstanding mechanical properties (3.6 cN/tex) and stability in extreme environments. These composite aerogel fibers offer great potential for high-performance thermal insulation textiles and products.

Fractal and first-passage properties of a class of self-similar networks.

A class of self-similar networks, obtained by recursively replacing each edge of the current network with a well-designed structure (generator) and known as edge-iteration networks, has garnered considerable attention owing to its role in presenting rich network models to mimic real objects with self-similar structures. The generator dominates the structural and dynamic properties of edge-iteration networks. However, the general relationships between these networks' structural and dynamic properties and their generators remain unclear. We study the fractal and first-passage properties, such as the fractal dimension, walk dimension, resistance exponent, spectral dimension, and global mean first-passage time, which is the mean time for a walker, starting from a randomly selected node and reaching the fixed target node for the first time. We disclose the properties of the generators that dominate the fractal and first-passage properties of general edge-iteration networks. A clear relationship between the fractal and first-passage properties of the edge-iteration networks and the related properties of the generators are presented. The upper and lower bounds of these quantities are also discussed. Thus, networks can be customized to meet the requirements of fractal and dynamic properties by selecting an appropriate generator and tuning their structural parameters. The results obtained here shed light on the design and optimization of network structures.

Large-Scale Binary Matrix Optimization for Multimicrogrids Network Structure Design

Large-Scale Binary Matrix Optimization for Multimicrogrids Network Structure Design

Robust optimization of a closed-loop supply chain network based on an improved genetic algorithm in an uncertain environment

Aiming at the network structure design problems of facility location, path planning, and flow distribution in an e-commerce closed-loop supply chain network under an uncertain environment, the uncertainties of both the forward and reverse logistics demands and the impact of facility disruptions are analyzed. A robust optimization model is established using the robust peer-to-peer optimization method. An algorithm is designed that combines the minimum spanning tree algorithm with a two-layer adaptive genetic algorithm (Prim-DMGA) to solve the model. The results show that the Prim-DMGA not only requires less computation time, but also generates solutions of higher quality. In addition, the application of the robust optimization model reduces the adverse effects of uncertain factors on the e-commerce closed-loop supply chain network, thereby improving its robustness.