Dynamic Network Research Articles

In the domain of medical image segmentation, models utilizing convolutional neural network (CNN) and Transformer have been t extensively studied and widely implemented. However, the self-attention mechanism in Transformer is incapable of adapting its focus to target structures at varying scales, resulting in discontinuities in segmentation. The objective of this study is to propose a multi-directional dynamic modeling network for medical image segmentation. We propose a Cross-axis Mamba attention (CMA) to capture global info and establish long-range dependencies. It integrates both global context and local details, enhancing segmentation performance. We also introduce an Edge Feature Enhancement Model (EFCN) to improve edge feature detection. We evaluated the method on the ISIC2018 dataset, as well as the CVC-300 and Kvasir-SEG datasets. The dice similarity coefficient and intersection-over-union (IoU) metrics achieved values of 91.12 and 85.07, 90.35 and 83.43, and 94.14 and 89.62, respectively. These results outperform those of advanced models such as VM-Unet and Swin-UMamba. The experimental results indicate that the proposed method has good generalization ability and robustness. It also provides important support for clinical diagnosis and treatment.

ABSTRACTDetecting antifragile communities is crucial for understanding complex network structures and dynamics. Communities can be categorized as fragile, robust or antifragile based on their inherent properties. This paper proposes a theoretical framework to identify community states using complexity measures. Initially, single‐node and multi‐node elimination perturbations are applied to selected communities within networks. Subsequently, the community's entropy before and after each perturbation and the resulting complexity variations are computed. To validate this framework, five distinct synthetic and three real‐world complex networks, each spanning multiple snapshots, are analysed. The findings reveal that communities with an initial entropy exceeding 0.5 out of 1 demonstrate antifragility, while those with initial entropy below 0.5 exhibit fragility in response to perturbations. Communities that become more complex are considered antifragile, whereas those that experience a decrease in complexity are deemed fragile. Our analysis also reveals a positive correlation between antifragility and both network density and average degree.

Cannabis is often consumed in polysubstance contexts, yet research comparing its isolated use to polydrug use is limited. Such comparisons are essential for understanding substance-specific versus interactive effects on brain activity. This study investigates how cannabis-only and polydrug use relate to alterations in resting-state EEG functional connectivity, using the Directed Transfer Function (DTF) to assess frequency-specific information flow in the brain. Three age- and sex-matched groups were studied: individuals with cannabis-only use (n=27), polydrug use (n=29), and non-drug-using controls (n=30). Resting-state EEG was recorded under eyes-open and eyes-closed conditions. DTF was applied to quantify directed connectivity and frequency-dependent propagation of neural signals across brain regions. Individuals with cannabis-only use exhibited localized changes in neural propagation, particularly in frontal areas and within theta, beta, and gamma bands, while alpha-band activity remained relatively stable. In contrast, individuals with polydrug use showed more widespread and complex alterations in connectivity, including reduced alpha activity and increased frontal and prefrontal engagement, suggesting heightened excitation and possible compensatory mechanisms. This study is the first to characterize frequency-specific EEG connectivity patterns associated with cannabis and polydrug use. The findings highlight the importance of studying substance use patterns and considering cannabis within a polysubstance context. A longitudinal study is needed to assess the effects of use duration, frequency, and dosage.

ABSTRACT Scaling out of agroecology requires social capital through networking of organizations to deliver different services including capacity building and resources. However, the role of networks in scaling agroecology for climate adaptation remains underexplored. This paper examines the role of networks in advancing agroecology for climate adaptation in Uganda. This paper explores two research questions; 1) What are the existing social networks among agroecology organizations in Uganda? 2) What challenges hinder networking among agroecology organizations in Uganda? Focus group discussions, small group interviews, and in-depth interviews were employed to explore Participatory Ecological Land Use Management (PELUM) network structure and connectivity, drivers for networking, networking challenges, and strategies for strengthening agroecology networks. Thematic and network analysis mapped networking dynamics among 25 PELUM member organizations across 17 Ugandan districts. Findings reveal a core periphery network structure, where central hubs collaborate with many organizations while peripheral ones operate in silos. Collaborative initiatives, such as farmer cooperatives, policy advocacy joint activities, and knowledge exchanges empower member organizations and farming communities. However, systemic collaboration challenges persist, including competition for funding, leadership deficits, geographic disparities, and conceptual misalignments. Network analysis suggests that regional clustering could bridge gaps and enhance networking. The study recommends establishment of consortium-based programming, formal partnerships, capacity building, and use of digital tools to strengthen collaboration and adoption of agroecology. In addition, we recommend that funding agencies and policymakers should incentivize consortium projects, integrate gender-sensitive frameworks, and support multi-stakeholder initiatives that strengthen agroecology for climate adaptation and food sovereignty.

Ethereum blockchain is the market leading platform for decentralized applications and smart contracts that have powered the new age of financial ecosystem. In order to improve security and performance, identify influential nodes, and understand network dynamics on Ethereum it is critical to identify influential nodes in Ethereum. This study explore machine learning techniques for discovery of these nodes using graph based algorithms, centrality measures and clustering methods. It studies the impact of a node in terms of frequency of usage, connectivity and computational power for a node. Finally, this study compare performance of proposed methodology combining supervised learning and graph neural networks to their traditional counterparts and demonstrate approach outperforms existing methods. The study demonstrate that highly influential nodes engage in unique patterns of behavior, which are detectable and categorizable. This study contribute to understanding of the network structure of Ethereum, along with a scalable approach to monitoring and optimising blockchain ecosystems. Moreover the study discuss the implications for network robustness, fraud detection and protocol enhancements, and demonstrate the promise of machine learning for blockchain analytics.

Most start-ups fail, and early-stage ventures face even lower survival rates. Identifying high-potential start-ups remains a critical challenge for venture capital (VC) investors and policymakers. While predictive models exist, the evolving relationships between VC investors, start-ups, and management teams in dynamic networks are underexplored. We propose a method to predict whether a start-up will succeed within 5 years of its first funding round. Using a 40-year global VC dataset, we model the VC ecosystem as a dynamic bipartite network linking start-ups to individuals (investors/managers). Our approach incrementally updates graph embeddings through unsupervised self-attention to incorporate new nodes, edges, and their neighbors. Node embeddings are further fine-tuned via link prediction and classification tasks, while temporal dependencies are captured to form sequential representations. The model identifies early-stage start-ups with twice the success likelihood of those chosen by professional investors. Key factors including networking and education align with VC literature. Additionally, we provide model complexity analysis and open source our implementation to support practical applications and future research.

In tactical communication networks, highly dynamic topologies and frequent data exchanges create complex spatiotemporal dependencies among link states. However, most existing intelligent routing algorithms rely on simplified model architectures and fail to capture these spatiotemporal correlations, resulting in limited situational awareness and poor adaptability under dynamic network conditions. To address these challenges, this study proposes an intelligent path selection method—Deep Reinforcement Learning with Spatiotemporal-aware Link State Guidance Algorithm (DRLSGA). The algorithm builds upon the Proximal Policy Optimization (PPO) framework to develop an intelligent decision-making model and integrates a link state feature extraction module that combines Gated Recurrent Units (GRU) and a Graph Attention Network (GAT). This design enables the model to learn long-term temporal dependencies and spatial structural relationships from sequential link state data, thereby enhancing perception and decision-making capability. An attention mechanism is further introduced to highlight salient features within link state sequences, while an optimal routing strategy is derived through a deep reinforcement learning-based training process. Experimental results demonstrate that, compared with the existing DRL-ST algorithm, DRLSGA reduces average end-to-end latency by at least 2.07%, lowers the packet loss rate by 1.65%, and increases average throughput by up to 2.59% under high-traffic conditions. Moreover, the proposed algorithm exhibits stronger adaptability to highly dynamic network topologies.

We investigate the effects of dynamic and stable network structures on the mechanical behavior of poly(vinyl alcohol) hydrogels prepared through diverse methods, including freeze-thawing (F), salting out (S), dry annealing (A), and their combinations (F-S, F-A, F-S-A). Utilizing in situ small-angle X-ray/neutron scattering (SAXS/SANS) and rheometry, we elucidate the structural evolution and mechanical response mechanisms of these hydrogels. Salting out induces a dynamic network (generally refers to the network with highly reversible cross-link points such as hydrogen bonds), enhancing energy dissipation and self-healing, while annealing forms a stable network (generally refers to the network with low-reversibility cross-link points such as crystals), improving strength and stiffness. The synergistic effect of salting out and annealing achieves a balanced network structure, optimizing crystal formation, uniformity, and mechanical performance. This tunable network design offers a universal strategy for adaptive hydrogels in biomedical devices. Additionally, the process of fracture and recrystallization of crystals within the structure is closely related to the yielding, hardening, fracture, and other behaviors of the sample, revealing a direct correlation between microstructural evolution and macroscopic mechanical properties.

BackgroundThe emergent dynamics of complex gene regulatory networks govern various cellular processes. However, understanding these dynamics is challenging due to the difficulty of parameterizing the computational models for these networks, especially as the network size increases. Here, we introduce a simulation library, Gene Regulatory Interaction Network Simulator (GRiNS), to address these challenges.ResultsGRiNS integrates popular parameter-agnostic simulation frameworks, RACIPE and Boolean Ising formalism, into a single Python library capable of leveraging GPU acceleration for efficient and scalable simulations. GRiNS extends the ordinary differential equations (ODE) based RACIPE framework with a more modular design, allowing users to choose parameters, initial conditions, and time-series outputs for greater customisability and accuracy in simulations. For large networks, where ODE-based simulation formalisms do not scale well, GRiNS implements Boolean Ising formalism, providing a simplified, coarse-grained alternative, significantly reducing the computational cost while capturing key dynamical behaviours of large regulatory networks.ConclusionGRiNS enables parameter-agnostic modeling of gene regulatory networks to study their dynamic and steady-state behaviors in a scalable and efficient manner. The documentation and installation instructions for GRiNS can be found at https://moltenecdysone09.github.io/GRiNS/.

ABSTRACT Digital literacy (DL) is increasingly recognized as a fundamental skill for participating in modern society. However, rural communities often face challenges due to limited access to resources, technology, and varying literacy levels. This research proposes an affordable, scalable, and adaptable embedded system to support DL education in rural areas, aligning with broader rural revitalization efforts. The system collects learner interaction data—such as usage patterns, DL task performance, and demographic information—to understand learner needs and progress. Data preprocessing techniques, including normalization and handling missing values, were applied to ensure high data quality for model training. Central to the system is the Cheetah Optimised Dynamic Recurrent Neural Network (CO-DRNN), which adapts to learners’ evolving needs, simulates learning progression, and delivers personalized educational content. The Cheetah Optimization component improves the DRNN’s efficiency and flexibility. Evaluation results show significant improvements in DL skills, with the CO-DRNN achieving 99.43% accuracy, 88.54% precision, 96.78% recall, and an F1-score of 86.21%. The model outperforms existing approaches, demonstrating its potential as a powerful tool for DL education. By equipping rural residents with essential digital skills, the system contributes to increased employability, digital inclusion, and community engagement, ultimately supporting sustainable rural development and participation in the digital economy.

Polymeric foams have garnered significant interest in advanced engineering applications due to their unique porous architectures. Ice templating represents a promising strategy for creating polymeric foams with tunable pore morphology, yet faces practical constraints including reliance on energy‐intensive lyophilization and/or insufficient mechanical properties of the resulting foams. To address this challenge, a strategy is developed to prepare mechanically robust polymeric foams by photo‐crosslinking ice‐templated emulsions. By incorporating dynamic hindered urea bonds into the emulsion, dynamic network reconfiguration is achieved, which dramatically improves mechanical properties. The resultant foam (≈65% porosity) exhibits tensile strength and breaking strains of 7.9 MPa and 533%, with toughness of 15.8 MJ m−3. In addition, this strategy permits the construction of intricate 3D architectures using 3D‐printed sacrificial thermoplastic templates, expanding their potential applications in demanding engineering scenarios.

Dynamic graph neural networks, i.e., DyGNNs, have been extensively explored in literature to handle structural and temporal properties in graphs. On the one hand, there naturally exist distribution shifts in real-world scenarios relevant to dynamic graphs. On the other hand, the dynamics may further bring extra uncertainties to patterns in dynamic graphs. However, existing DyGNNs merely exploit variant patterns with respect to labels under distribution shifts, failing to accurately make predictions when there exist distribution shifts together with uncertain patterns from training data to test data. To deal with this issue, in this paper we propose to handle spatio-temporal distribution shifts in dynamic graphs via the discovery and utilization of invariant patterns, taking uncertainties in patterns into account, where the invariant patterns include structures and features whose predictive abilities are stable across distribution shifts. Nevertheless, we face the following key challenges: i) How to discover the complex invariant and variant spatio-temporal patterns involving time-varying topological structures and node-level features; ii) How to utilize the invariant and variant patterns to deal with the spatio-temporal distribution shifts in dynamic graphs; iii) How to handle the pattern uncertainties upon capturing the hidden invariance and variance with a theoretical guarantee. To tackle these challenges, we propose the Information Bottleneck guided Disentangled Dynamic Graph ATtention network (IB-D2GAT). Our proposed IB-D2GAT model is able to effectively handle spatio-temporal distribution shifts with uncertainties in dynamic graphs through discovering variant and invariant spatio-temporal patterns via information bottleneck. Specifically, we propose a disentangled spatio-temporal attention network to capture the invariant and variant patterns. Next, guided by the information bottleneck principle, we propose the distribution-based invariance optimization strategy which injects stochasticity into the invariant pattern identification so as to prevent the variant information from influencing the prediction, thus eliminating the spurious impacts of variant patterns. We further theoretically show that our proposed tailored invariance optimization strategy can lead to accurately capturing the invariant patterns with stable predictive abilities and therefore is capable of handling distribution shifts. Experiments on multiple real-world datasets and one synthetic dataset demonstrate the superiority of our method over state-of-the-art baselines under distribution shifts.

With the arrival of sixth-generation (6G) wireless systems comes radical potential for the deployment of autonomous aerial vehicle (AAV) swarms in mission-critical applications, ranging from disaster rescue to intelligent transportation. However, 6G-supporting AAV environments present challenges such as dynamic three-dimensional topologies, highly restrictive energy constraints, and extremely low latency demands, which substantially degrade the efficiency of conventional routing protocols. To this end, this work presents a Q-learning-enhanced ad hoc on-demand distance vector (QL-AODV). This intelligent routing protocol uses reinforcement learning within the AODV protocol to support adaptive, data-driven route selection in highly dynamic aerial networks. QL-AODV offers four novelties, including a multipath route set collection methodology that retains up to ten candidate routes for each destination using an extended route reply (RREP) waiting mechanism, a more detailed RREP message format with cumulative node buffer usage, enabling informed decision-making, a normalized 3D state space model recording hop count, average buffer occupancy, and peak buffer saturation, optimized to adhere to aerial network dynamics, and a light-weighted distributed Q-learning approach at the source node that uses an ε-greedy policy to balance exploration and exploitation. Large-scale simulations conducted with NS-3.34 for various node densities and mobility conditions confirm the better performance of QL-AODV compared to conventional AODV. In high-mobility environments, QL-AODV offers up to 9.8% improvement in packet delivery ratio and up to 12.1% increase in throughput, while remaining persistently scalable for various network sizes. The results prove that QL-AODV is a reliable, scalable, and intelligent routing method for next-generation AAV networks that will operate in intensive environments that are expected for 6G.

Does the structure of dynamic symptom networks depend on baseline psychopathology in students?

In materio computing offers the potential for widespread embodied intelligence by leveraging the intrinsic dynamics of complex systems for efficient sensing, processing, and interaction. While individual devices offer basic data processing capabilities, networks of interconnected devices can perform more complex and varied tasks. However, designing such networks for dynamic tasks is challenging in the absence of physical models and accurate characterization of device noise. We introduce the Noise-Aware Dynamic Optimization (NADO) framework for training networks of dynamical devices, using Neural Stochastic Differential Equations (Neural-SDEs) as differentiable digital twins to capture both the dynamics and stochasticity of devices with intrinsic memory. Our approach combines backpropagation through time with cascade learning, enabling effective exploitation of the temporal properties of physical devices. We validate this method on networks of spintronic devices across both temporal classification and regression tasks. By decoupling device model training from network connectivity optimization, our framework reduces data requirements and enables robust, gradient-based programming of dynamical devices without requiring analytical descriptions of their behaviour.

IntroductionUrban communities, as the basic unit of urban governance, play a crucial role in responding to public health emergencies (PHEs). This study aims to investigate the resilience measurement and optimization strategies of urban communities in responding to PHEs in order to improve their resilience.MethodsThe study constructed a resilience assessment framework and identified 31 key influencing factors to measure the resilience of case communities in Nanjing. Through sensitivity analysis, static optimization strategies were proposed from social, environmental, and economic levels. Dynamic Bayesian network inference simulation and importance analysis were used to propose dynamic optimization strategies from pre, during, and long-term perspectives.ResultsThrough the combination of dynamic and static strategies, community managers promote resilience building from both short-term and long-term perspectives.DiscussionThe study provides a valuable reference for comprehensively improving the emergency management system.

A focal ischemic lesion is thought to alter neuronal activity beyond the area of structural damage, thereby interfering with the whole network architecture. Here, we used a combination of transcranial magnetic stimulation and electroencephalography in conjunction with dynamic connectivity analyses and graph theory to study alterations and reorganization of cortical connectivity in a cohort of 41 patients longitudinally after stroke. We found a link between an increase in low-frequency coupling in the delta band and alterations in neural information processing in the first weeks after stroke and their relevance for motor outcome >3 months later. We demonstrated that stroke enhances slow activity and delta coupling between frontocentral and parietal regions. In addition, we observed a loss of the physiological network architecture with a decrease in small-worldness and modularity in the delta frequency, implying that a focal ischemic lesion interferes with both cortical information integration and functional segregation within the first weeks after stroke. While we found a link between bifrontal coupling in the alpha spectrum and the degree of the motor deficit in the early post-acute phase, the amount of small-worldness disruption early after stroke indicated the motor outcome in the follow-up session. In contrast, recovery of motor function and cortical reorganization after >3 months post-stroke were paralleled by the normalization of increased low-frequency coupling and a reinstatement of the complex network structure featuring a modular and small-world topology.

TMS-induced modulation of brain networks and its associations to rTMS treatment for depression: a concurrent fMRI-EEG-TMS study.

This paper introduces a novel approach to stability analysis and control for TCP-based Cyber-Physical Systems (CPSs) using the Linear Parameter Varying (LPV) approach considering communication network delays and dynamics. The closed-loop CPS is modelled considering the transmission control protocol (TCP) as the network protocol. By integrating network dynamics and delay, the entire closed-loop system is represented as a delay dependent LPV system. Subsequently, a sufficient stability condition for the augmented system is established in terms of Linear Matrix Inequalities (LMIs), and the maximum allowable transfer interval (MATI) by considering network parameteres is calculated. Additionally, an LPV state feedback controller is designed for the augmented system which adaptively compensates the effect of network dynamics and parameteres on the plant. In this way, the proposed stabilising controller for CPSs is able to deal with the potential destabilisation effects related to network delays. Finally, the proposed approach is evaluated using a widely used application case study in networked control systems literature.

Existing studies have examined various measures to mitigate the ripple effect of supply chain disruptions, but few have focused on the optimal timing for supplier replacement, particularly in cruise supply chains. This study proposes a novel analytical framework that, for the first time, integrates causal dynamic Bayesian networks, Do-calculus, and mathematical programming to assess supplier replacement timing for controlling disruption propagation. First, a supply chain disruption ripple effect model is constructed: a dynamic Bayesian network captures the ripple effect without supplier replacement, while the causal dynamic Bayesian network and Do-calculus capture the ripple effect under supplier replacement. Building on this model, three models of supply chain risk, service level, and cost are developed. Then, a multi-objective non-convex mixed-integer programming model is formulated to determine the optimal supplier replacement timing, aiming to minimise risk, maximise service level, and minimise cost. Finally, the empirical analysis of cruise supply chain operations shows that risk threshold settings influence the timing and frequency of supplier replacement. This leads to a nonlinear relationship among cost, risk, and service level. Specifically, this relationship manifests as phased improvements, temporary fluctuations due to increased strategic complexity, and diminishing marginal returns as cost inputs rise.