Network Datasets Research Articles

Vertex clustering in diverse dynamic networks

We present theoretical and experimental results for spatiotemporal graph k-means (STGkM)—a new unsupervised method to cluster vertices within a dynamic network. STGkM finds both short-term dynamic clusters and a “long-lived” partition of vertices within a network whose topology is evolving over time; we first introduced this technique in a recent conference paper. Here, we update our algorithm with a more efficient relaxation scheme, provide additional theoretical results, compare its performance to several other methods, and demonstrate its capabilities on real, diverse datasets. We construct a theoretical foundation to distinguish STGkM from connected components and static clustering and prove results for the stochastic setting for the first time. In addition to our previous experiments on the United States House of Representatives dataset, we report new state-of-the-art empirical results on a dynamic scientific citation network and Reddit dataset. These findings demonstrate that STGkM is accurate, efficient, informative, and operates well in diverse settings. Finally, as previously noted, one of the main advantages of STGkM is that it has only one required parameter: k, the number of clusters; we therefore include an extended analysis of the range of this parameter and guidance on selecting its optimal value. Our data and code are available on Github; see: https://github.com/dynestic/stgkm.

Exploring Physical Activity Among Mexican American Immigrants in New York City Before and During the COVID-19 Pandemic: A Two-Wave Panel, Mixed-Method Analysis.

The COVID-19 pandemic was associated with decreases in moderate-to-vigorous physical activity and increases in sedentary time for children and adults, and there is some evidence that Latino populations were particularly affected. This article leverages a two-wave panel, mixed-method social network dataset collected before and during the pandemic with Mexican American immigrants living in New York City to examine social networks and other factors associated with physical activity and explore participants' perceptions about how the pandemic affected their physical activity. Participants (n = 49) completed in-person, egocentric social network interviews between January and June 2019 and virtual follow-up egocentric social network interviews between May and November 2021. Qualitative data collection with a subsample (n = 25) occurred between October and December 2022. Social network quantitative analyses found that age and female sex were negatively associated with participants' baseline physical activity level, but the proportion of alters (important persons identified by the participant) with whom the participant did physical activity at baseline was positively associated. Baseline physical activity level was inversely related to the change in physical activity, and the change in the proportion of alters who were close to the participant was negatively associated with physical activity change at follow-up. Qualitative results supported the quantitative findings, providing rich narratives regarding the importance of social support for physical activity during the pandemic. Social networks among Mexican American immigrants in New York City provided motivation and support for physical activity but became harder to sustain during a pandemic. The findings can inform strategies to create environments that are conducive to physical activity while also protecting public health.

Anomaly detection system based on deep learning for cyber physical systems on sensory and network datasets

Anomaly detection system based on deep learning for cyber physical systems on sensory and network datasets

Fast prediction of key parameters in FEBA using the COSINE subchannel code and artificial neural network

Numerical techniques have emerged as an essential tool for operators and designers to preemptively acquire key parameters in accidents analysis. However, due to insufficient experience, it is difficult for them to obtain satisfactory numerical results. Moreover, the uncertainty analysis and quantification necessitate the simulation of a substantial number of samples, which requires a significant amount of computational time. Therefore, the development of a fast prediction model becomes imperative. In this work, a prediction model based on the in-house COSINE subchannel code and Multi-Head Perceptron (MHP) is developed. The COSINE subchannel code is employed to provide data sets for training neural networks. Firstly, the numerical results of COSINE subchannel code are compared with experimental data to ensure the accuracy of data sets. Secondly, input features for neural networks are selected by evaluating the impact of input parameters on numerical results, and a series of simulations is carried out to generate data sets. Then, a comparative analysis was conducted between the Multi-Layer Perceptron (MLP) and Support Vector Regression (SVR) models, and the MLP model performs better. Subsequently, the MLP was compared with the MHP, demonstrating the advantage of MHP model. Based on this, the predictions are conducted using the MHP model and the distribution of key parameters is compared with that obtained by COSINE subchannel code. The results illustrate that developed MHP model is an efficient tool for predicting key parameters during the reflooding phase.

Model Parameter Prediction Method for Accelerating Distributed DNN Training

As the size of deep neural network (DNN) models and datasets increases, distributed training becomes popular to reduce the training time. However, a severe communication bottleneck in distributed training limits its scalability. Many methods aim to solve this communication bottleneck by reducing communication traffic, such as gradient sparsification and quantification. However, these methods either are at the expense of losing model accuracy or introducing lots of computing overhead. We observe that the data distribution between layers of neural network models is similar. Thus, we propose a model parameter prediction method (MP2) to accelerate distributed DNN training under parameter server (PS) framework, where workers push only a subset of model parameters to PS, and residual model parameters are locally predicted by an already-trained deep neural network model on PS. Some key challenges are addressed. First, we build a hierarchical parameters dataset by randomly sampling a subset of model from normal distributed trainings. Second, we build a neural network model with the structure of “convolution + channel attention + Max pooling” for predicting model parameters by using a prediction result-based evaluation method. For VGGNet, ResNet, and AlexNet models on CIFAR10 and CIFAR100 datasets, compared with Baseline, Top-k, deep gradient compression (DGC), and weight nowcaster network (WNN), MP2 can reduce traffic by up to 88.98%; and accelerates the training by up to 47.32% while not losing the model accuracy. MP2 has shown good generalization.

Sharpness-Aware Lookahead for Accelerating Convergence and Improving Generalization.

Lookahead is a popular stochastic optimizer that can accelerate the training process of deep neural networks. However, the solutions found by Lookahead often generalize worse than those found by its base optimizers, such as SGD and Adam. To address this issue, we propose Sharpness-Aware Lookahead (SALA), a novel optimizer that aims to identify flat minima that generalize well. SALA divides the training process into two stages. In the first stage, the direction towards flat regions is determined by leveraging a quadratic approximation of the optimization trajectory, without incurring any extra computational overhead. In the second stage, however, it is determined by Sharpness-Aware Minimization (SAM), which is particularly effective in improving generalization at the terminal phase of training. In contrast to Lookahead, SALA retains the benefits of accelerated convergence while also enjoying superior generalization performance compared to the base optimizer. Theoretical analysis of the expected excess risk, as well as empirical results on canonical neural network architectures and datasets, demonstrate the advantages of SALA over Lookahead. It is noteworthy that with approximately 25% more computational overhead than the base optimizer, SALA can achieve the same generalization performance as SAM which requires twice the training budget of the base optimizer.

Key Node Identification Method Based on Multilayer Neighbor Node Gravity and Information Entropy

In the field of complex network analysis, accurately identifying key nodes is crucial for understanding and controlling information propagation. Although several local centrality methods have been proposed, their accuracy may be compromised if interactions between nodes and their neighbors are not fully considered. To address this issue, this paper proposes a key node identification method based on multilayer neighbor node gravity and information entropy (MNNGE). The method works as follows: First, the relative gravity of the nodes is calculated based on their weights. Second, the direct gravity of the nodes is calculated by considering the attributes of neighboring nodes, thus capturing interactions within local triangular structures. Finally, the centrality of the nodes is obtained by aggregating the relative and direct gravity of multilayer neighbor nodes using information entropy. To validate the effectiveness of the MNNGE method, we conducted experiments on various real-world network datasets, using evaluation metrics such as the susceptible-infected-recovered (SIR) model, Kendall τ correlation coefficient, Jaccard similarity coefficient, monotonicity, and complementary cumulative distribution function. Our results demonstrate that MNNGE can identify key nodes more accurately than other methods, without requiring parameter settings, and is suitable for large-scale complex networks.

DGTNet:dynamic graph attention transformer network for traffic flow forecasting

Abstract Graph-based traffic flow prediction plays a crucial role in urban traffic management and planning. In this paper, we propose a novel Dynamic Graph Attention Transformer Network (DGTNet), which is designed to address the issue of inadequate integration of temporal and spatial dimensions in traditional models. DGTNet maintains temporal continuity while revealing the complex dynamic relationships between key nodes in the urban traffic system, capturing the periodic changes in the rhythm of city life. Specifically, this study adopts adaptive signal decomposition technology to decompose traffic data into multiple Intrinsic Mode Functions (IMFs), effectively capturing the dynamic changes in traffic flow. This decomposition method is key to the implementation of DGTNet’s dynamic graph construction, enabling the analysis of traffic flow at different time scales, thereby providing a new perspective for traffic flow prediction research. In the traffic prediction module, we comprehensively consider node, edge, and graph structural information, adopting a multi-head self-attention mechanism to achieve direct cross-modeling in both temporal and spatial dimensions. Finally, we introduce a position-wise feedforward network layer to integrate different types of data and capture nonlinear relationships. The experimental results, based on public transportation network datasets METR_LA, PEMS_BAY, PEMS03, and PEMS07, demonstrate that DGTNet exhibits notable enhancements in three evaluation indicators, namely the Mean Absolute Percentage Error (MAPE), the Root Mean Square Error (RMSE), and the Mean Absolute Error (MAE). The pertinent code has been made available for public access at https://github.com/chenjing0616/DGTNet.

Advancements of Exploiting Convolutional Neural Networks for Solving Differential Equations

Solving Partial Differential Equations (PDEs) is essential across various fields, including physics, mathematics, and engineering. This study explores innovative methods for solving PDEs using Convolutional Neural Networks (CNNs). Traditionally, solving PDEs through numerical methods like finite difference or spectral approaches is computationally intensive, particularly when addressing high-dimensional problems and complex boundary conditions. CNNs, with their ability to handle spatial data efficiently, offer a promising alternative. This research evaluates different neural network architectures, including data-driven, physics-driven, and hybrid approaches, and their role in enhancing the accuracy and scalability of PDE solutions. Data-driven methods rely on extensive datasets for training neural networks, while Physics-Informed Neural Networks (PINNs) integrate the underlying physical laws into the solution process. Hybrid approaches combine both strategies, leveraging data and physical principles. The paper emphasizes CNNs potential to transform traditional PDE-solving methods, making them more efficient and adaptable to complex, real-world applications. Future research directions include optimizing these methods for greater scalability and interdisciplinary applications.

Analysis of athletes’ technical action based on deep learning

In order to raise athletes’ technical proficiency and overall performance, this paper uses deep learning technology to precisely assess table tennis technical activities. The paper builds a hybrid neural network model for technical action analysis of table tennis players, based on Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) in deep learning. First, CNN extracts the spatial characteristics from the video frames. After that, an LSTM processes these data in a time series to accurately recognize and classify the movements of athletes. Lastly, through trials, the model is trained and evaluated using the Table Tennis Tracking Network (TTNET) dataset. The experimental findings demonstrate that: (1) In table tennis technical action analysis, the suggested deep learning model performs better than other comparative models. The efficacy of combining convolution with sequence learning is fully demonstrated by the CNN-LSTM hybrid model, which performs best in all indicators when compared to the CNN, LSTM, Multi-Objective Function (MOF), and Convolutional Neural Network—Long Short-Term Memory (CNN-LSTM) combined models. Its accuracy rate, precision rate, recall rate, and F1 score are 0.923, 0.918, 0.925, and 0.921, respectively. In contrast, the performances of LSTM and CNN are also excellent, but the performance of MOF model is relatively low; (2) In the classification of technical actions, the model has the highest classification accuracy of service, reaching 0.930, and the classification accuracy of other technical actions such as forehand stroke, backhand stroke and spike is also above 0.9, and the time sequence consistency index also maintains a high level, indicating that the model can effectively identify and analyze table tennis technical actions. In addition, the performance evaluation of real-time feedback shows that the model can achieve low processing time and feedback delay in video data processing with different lengths, which ensures the real-time and reliability in practical applications. These results show that the proposed model can not only provide accurate technical action recognition, but also provide timely and effective feedback in practical application, which has high practical value. The results of this paper prove the potential of deep learning technology in the analysis of athletes’ technical actions, and provide scientific basis and effective tools for the technical training and optimization of table tennis players.

The development of a routable pavement network dataset to support active travel

ABSTRACT Active travel (AT) networks have only recently become a priority for urban planners in many countries, and networks of spatial data and applications that use them are also lacking. Road networks have historically been used as a proxy for pedestrian networks, yet they are a poor one as human movement is much less constrained than that of motorised vehicles. This study presents a novel approach to producing an attribute-rich network for AT, together with a semi-automated method to produce a width-attributed pavement network from topographic mapping agency data (based on a case study of Ordnance Survey (OS) MasterMap data from the UK). A shortest path network routing application using Dijkstra’s algorithm is presented to demonstrate the possibilities of AT routing incorporating increased attribution. The paper concludes by firstly proposing additional steps to further enhance the attribution of the network and finally by describing the relevant policy implications of this work.

Influence of multiple spreaders through farthest first traversal

Identifying influential spreaders in complex networks is crucial for improving the efficiency of spreading processes. In this paper, we use a farthest first traversal to partition the network into communities and identify influential spreaders. The spreaders selected by this method satisfy the two criteria of being dispersed as well as influential in their neighborhood. Using an SIR-based epidemic spread on network datasets we examine the spreading ability of the influential spreaders. We compare the epidemic size when initial spreaders are selected from each community with ranked initial spreaders but no assurance of representation from each community. A larger epidemic size is observed when the initial influential spreaders are dispersed and selected from each community. In addition, the spread ability of influential spreaders selected by the proposed methods is similar to those obtained by creating non-overlapped communities using the Louvain algorithm, provided they are selected by using the same criteria.

Information Propagation in Hypergraph-Based Social Networks.

Social networks, functioning as core platforms for modern information dissemination, manifest distinctive user clustering behaviors and state transition mechanisms, thereby presenting new challenges to traditional information propagation models. Based on hypergraph theory, this paper augments the traditional SEIR model by introducing a novel hypernetwork information dissemination SSEIR model specifically designed for online social networks. This model accurately represents complex, multi-user, high-order interactions. It transforms the traditional single susceptible state (S) into active (Sa) and inactive (Si) states. Additionally, it enhances traditional information dissemination mechanisms through reaction process strategies (RP strategies) and formulates refined differential dynamical equations, effectively simulating the dissemination and diffusion processes in online social networks. Employing mean field theory, this paper conducts a comprehensive theoretical derivation of the dissemination mechanisms within the SSEIR model. The effectiveness of the model in various network structures was verified through simulation experiments, and its practicality was further validated by its application on real network datasets. The results show that the SSEIR model excels in data fitting and illustrating the internal mechanisms of information dissemination within hypernetwork structures, further clarifying the dynamic evolutionary patterns of information dissemination in online social hypernetworks. This study not only enriches the theoretical framework of information dissemination but also provides a scientific theoretical foundation for practical applications such as news dissemination, public opinion management, and rumor monitoring in online social networks.

PQKELP: Projected Quantum Kernel Embedding based Link Prediction in dynamic networks

In dynamic networks, where the network’s topology is constantly changing, link prediction is a challenging issue. The two major challenges in link prediction within a time-varying network are accuracy and efficiency. Although random walk techniques have introduced promising embedding-based approaches, they fall short of optimization. Quantum computing, on the other hand, enhances performance in high-dimensional spaces, yet faces concerns about the limited efficiency of few qubit simulators. Addressing these issues, Projected Quantum Kernels (PQK) presents an elegant solution to achieve quantum advantage by simple quantum projection using the kernel trick, followed by a projection back to the classical state with relabeled data. In this work, we propose Projected Quantum Kernel Embedding based Link Prediction (PQKELP), a projected quantum kernel (PQK) approach on random walk embedding-based features to solve the link prediction problem. Thereby achieving a two-fold improvement by combining embedding generation and quantum projection, resulting in quantum-enhanced embedded features that enhance link prediction performance. Extensive experiments and analyses were done with a set of well-known random walk methods, i.e., Node2Vec, DeepWalk, and Walklets, also with six classical machine learning techniques and on five well-known dynamic network datasets. The results including several performance metrics like accuracy, AUC, F1-score, and precision show that our proposed model is better than traditional link prediction methods, classical machine learning approaches, and even the most cutting-edge methods available currently.

Neural Correlates Of Irritability And Potential Moderating Effects Of Inhibitory Control

BackgroundIrritability affects up to 20% of youth and is a primary reason for referral to pediatric mental health clinics. Irritability is thought to be associated with disruptions in processing of reward, threat, and cognitive control; however, empirical study of these associations at both the behavioral and neural level have yielded equivocal findings that may be driven by small Ns and differences in study design. Understudied are associations between irritability and brain connectivity between cognitive control and reward or threat-processing circuits. Further, better inhibitory control has been linked to lower irritability and differential neural functioning among irritable youth, suggesting that good inhibitory control may serve as a protective factor. MethodsWe hypothesized that higher irritability scores would be associated with less positive (or negative) connectivity between cognitive control and threat-processing circuits and between cognitive control and reward-processing circuits in the Healthy Brain Network dataset (Release 10.0; N=4135). We further hypothesized that these associations would be moderated by inhibitory control such that weaker associations between irritability and connectivity would be detected in youth with better versus worse inhibitory control. Regression models tested if associations between irritability and between-network connectivity were moderated by inhibitory control. ResultsCounter to our hypothesis, we detected higher irritability associated with reduced connectivity between threat- and reward-processing and cognitive control networks only in five to nine year-old boys. Inhibitory control did not moderate associations of irritability with between-network connectivity. ConclusionsExploratory findings indicate reduced between-network connectivity may underlie difficulty regulating negative emotions, leading to greater irritability.

Synergistic Integration of Local and Global Information for Critical Edge Identification.

Identifying critical edges in complex networks is a fundamental challenge in the study of complex networks. Traditional approaches tend to rely solely on either global information or local information. However, this dependence on a single information source fails to capture the multi-layered complexity of critical edges, often resulting in incomplete or inaccurate identification. Therefore, it is essential to develop a method that integrates multiple sources of information to enhance critical edge identification and provide a deeper understanding and optimization of the structure and function of complex networks. In this paper, we introduce a Global-Local Hybrid Centrality method which integrates a second-order neighborhood index, a first-order neighborhood index, and an edge betweenness index, thus combining both local and global perspectives. We further employ the edge percolation process to evaluate the significance of edges in maintaining network connectivity. Experimental results on various real-world complex network datasets demonstrate that the proposed method significantly improves the accuracy of critical edge identification, providing theoretical and methodological support for the analysis and optimization of complex networks.

Revealing the closeness of publication ties in scientific cooperation taking into account scientific productivity based on the Time-Weighted PageRank Method with Citation Intensity

The object of this study is the processes related to the assessment of the closeness of publication ties among scientists and taking into account their productivity related to scientific activity. This is necessary to increase the efficiency of management of research projects. To this end, the PR, TWPR, TWPR-CI methods for calculating scientific productivity estimates of scientists were described. In particular, the TWPR-CI method gives preference to those scientists whose works were more intensively published and cited during the last period of time, which is important for the formation of the composition of the executors of scientific research projects. The method for calculating the closeness of publication ties among scientists or average asymmetric tie strength was also described. The verification of dependence between the evaluation of the closeness of publication ties among scientists and their scientific productivity was carried out based on the analysis of the citation network of scientific publications and the network of scientific cooperation. The networks are built on the basis of the open access Citation Network Dataset (ver. 14). The dataset contains information on more than 5 million scientific publications and more than 36 million citations to them. The correlation analysis revealed the presence of a weak inverse relationship between these estimates. However, the weakness of the connection allows us to state that for this case there is no established correlation between the assessment of scientific productivity and the assessment of the closeness of publication ties. That is, the hypothesis that the weak connection between scientists makes it possible to increase the productivity and innovativeness of their publications was not confirmed. The results allow for a systematic approach to the process of evaluation and planning of the results of research projects, as well as the formation of the composition of their executors

Sine Cosine Algorithm Based on Optimal Convolutional Autoencoder for Intrusion Detection and Classification Models

Network security comprises a multifaceted method that aims to protect computer networks from malicious activities, unauthorized access, and data breaches. The security mechanism is Intrusion Detection which is an important constituent that is employed to monitor and analyse the network traffic for recognizing and responding to intrusive or suspicious behavior. Innovative methods such as deep learning (DL) are employed to enhance the effectiveness of Intrusion Detection Systems (IDSs). DL is extremely implemented for IDS owing to its proficiency for automatically learning and extracting complex patterns and features from massive and multifaceted network datasets. Neural network (NN) models, permit the system to distinguish between anomalous patterns and normal network behaviors, increasing the accuracy of intrusion detection. The flexibility of DL methods to emerging cyberattacks with their adeptness to handle large - scale and various data, positions them as a strong and efficient tool for proactive and intelligent intrusion detection in existing cybersecurity settings. This article presents a Sine Cosine Algorithm with Optimal Convolutional Autoencoder for Intrusion Detection and Classification (SCAOCAE - IDC) method. The developed SCAOCAE - IDC system presents a wide - ranging strategy to improve the precision and effectiveness of IDSs. The method combines diverse advanced mechanisms like Min - Max scalar normalization for data preprocessing, Sine Cosine Algorithm (SCA) for feature selection (FS), Convolutional Autoencoder (CAE) for better feature extraction and classification, and Heap - Based Optimization (HBO) for hyperparameter tuning. The Min - Max scalar makes sure of robust data normalization, SCA increasingly chooses main features, CAE capably captures complex patterns in the data, and HBO fine - tunes hyperparameters for improved system performance. By employing the synergistic combination of such modules, the presented SCAOCAE - IDC algorithm indicates considerable outcomes for increasing the reliability and accuracy of IDSs and classification systems.

Improved identification of network anomalies through optimal CURE clustering

Abstract In this paper, we propose an advanced network anomaly behavior identification framework to overcome the constraints inherent in conventional rule- or signature-based approaches, which often struggle with emerging and previously unknown threats. Central to our framework is an Enhanced CURE (Cluster Updating and REfining) clustering algorithm, meticulously tailored and refined to incorporate a density-based methodology. This enhancement enables the algorithm to discern subtle shifts in network anomaly patterns with heightened precision. The implementation workflow commences with the application of the optimized CURE algorithm to analyze network data, followed by the deployment of a sophisticated anomaly degree ranking mechanism. This mechanism, through meticulous calculation of individual data points’ anomaly degrees and subsequent ranking, effectively isolates those deviating significantly from standard behavioral norms, incorporating a strategic threshold to filter out false positives. To validate our methodology’s efficacy and its superiority over existing techniques, experiments were conducted utilizing a substantial real-world network dataset. These tests affirm not only a marked increase in the accuracy of abnormal behavior recognition and a reduction in computational intricacy but also demonstrate the adaptability across diverse network ecosystems. Our approach has proven successful in pinpointing a wide array of network anomalies, encompassing malicious cyberattacks, fraudulent activities, unauthorized intrusions, and breaches of security protocols, thereby highlighting its comprehensive capability in bolstering network defense strategies. Despite the notable advancements and successful identification of various network anomalies, our framework currently lacks integration with real-time learning capabilities, limiting its immediate responsiveness to rapidly evolving attack patterns and necessitating ongoing research for dynamic updates and adaptive learning mechanisms.

A survey on feature extraction and learning techniques for link prediction in homogeneous and heterogeneous complex networks

Complex networks are commonly observed in several real-world areas, such as social, biological, and technical systems, where they exhibit complicated patterns of connectedness and organised clusters. These networks have intricate topological characteristics that frequently elude conventional characterization. Link prediction in complex networks, like data flow in telecommunications networks, protein interactions in biological systems, and social media interactions on platforms like Facebook, etc., is an essential element of network analytics and presents fresh research challenges. Consequently, there is a growing emphasis in research on creating new link prediction methods for different network applications. This survey investigates several strategies related to link prediction, ranging from feature extraction based to feature learning based techniques, with a specific focus on their utilisation in dynamic and developing network topologies. Furthermore, this paper emphasises on a wide variety of feature learning techniques that go beyond basic feature extraction and matrix factorization. It includes advanced learning-based algorithms and neural network techniques specifically designed for link prediction. The study also presents evaluation results of different link prediction techniques on homogeneous and heterogeneous network datasets, and provides a thorough examination of existing methods and potential areas for further investigation.