Dynamic Graph Model Research Articles

A Bayesian decision support system for counteracting activities of terrorist groups

Abstract We present an integrating decision support system designed to aid security analysts’ monitoring of terrorist groups. The system comprises of (i) a dynamic network model of the level of bilateral communications between individuals and (ii) dynamic graphical models of those individual’s latent threat states. These component models are combined in a statistically coherent manner to provide measures of the imminence of an attack by the terrorist group. Domain knowledge provides the structures of the models, values of parameters and prior distributions over latent variables. Inference of the values is performed using time-series of observed data and the statistical dependencies assumed between said data and model variables. The work draws on social network and graphical models used in sociological, military, and medical fields.

Graphodynamic System with Information Entropy Control as a Means of Control of Production Systems

Ensuring the quality of products and increasing the level of safety in the operation of production systems are the main tasks of all industries. The second task is the most relevant for technogenic systems of increased danger, which include, among other things, the production systems of the oil and gas industry. The management of production systems and the solution of selected tasks is carried out by the decision maker, using the parameters obtained during the diagnosis and monitoring of the system. The formation of a complex of diagnostic parameters of the production system is implemented by the decision maker, taking into account the loss of information, characterized by incomplete control of the parameters. When identifying the state of the production system, the decision maker must take into account not only the values of diagnostic parameters, but also the value of the information entropy of the system that they form, since "entropy characterizes the uncertainty of control, i.e. its management quality". Thus, the identification of the state of the production system and its management are only possible with the mandatory control of the information entropy of the system by the decision maker. The latter confirms the relevance of the synthesis of control systems with control of the state of their information entropy. Consider the production system as a graphodynamic system. The data characterizing its input, output, internal state and control action will be presented in the form of graphs. Through the analysis of diagnostic parameters, the decision maker identifies the state of the production system. A certain time is spent on obtaining and analyzing diagnostic parameters, which in some cases can be significant and lead to an increase in the value of the information entropy of the controlled system, which leads to a mismatch between the actual state of the system and its image processed by the decision maker. In the conditions of big production systems, taking into account funding constraints or other reasons, the frequency of monitoring the diagnostic parameters of the system may be violated. This leads to a significant increase in the information entropy of the controlled system and, accordingly, to a low quality of its control. The proposed method for controlling a collector-beam system for collecting products from a gas condensate field makes it possible to improve the quality of its control by implementing control of its state of information entropy. The developed graph-dynamic model of data storage of a collector-beam system for collecting gas condensate field products allows the decision maker to analyze its state in any period of operation.

Dynamic Meta-path Guided Temporal Heterogeneous Graph Neural Networks

Graph Neural Networks (GNNs) have become the de facto standard for representation learning on topological graphs, which usually derive effective node representations via message passing from neighborhoods. Although GNNs have achieved great success, previous models are mostly confined to static and homogeneous graphs. However, there are multiple dynamic interactions between different-typed nodes in real-world scenarios like academic networks and e-commerce platforms, forming temporal heterogeneous graphs (THGs). Limited work has been done for representation learning on THGs and the challenges are in two aspects. First, there are abundant dynamic semantics between nodes while traditional techniques like meta-paths can only capture static relevance. Second, different semantics on THGs have often mutually evolved with each other over time, making it more difficult than dynamic homogeneous graph modeling. To address this problem, here we propose the Dynamic Meta-path guided temporal heterogeneous Graph Neural Networks (DyMGNNs). To handle the dynamic semantics, we introduce the concept of dynamic meta-path which is a common base for temporal semantic search engines, and then adopt the temporal importance sampling to extract neighborhoods with temporal bias. Focusing on mutual evolution, we design the heterogeneous mutual evolution attention mechanism, which can model the fine-grained interplay of semanticlevel preferences for each node. Extensive experiments on three real-world datasets for node classification and temporal link prediction demonstrate that our method consistently outperforms state-of-the-art alternatives.

Stability and dynamics of a spectral graph model of brain oscillations.

We explore the stability and dynamic properties of a hierarchical, linearized, and analytic spectral graph model for neural oscillations that integrates the structural wiring of the brain. Previously, we have shown that this model can accurately capture the frequency spectra and the spatial patterns of the alpha and beta frequency bands obtained from magnetoencephalography recordings without regionally varying parameters. Here, we show that this macroscopic model based on long-range excitatory connections exhibits dynamic oscillations with a frequency in the alpha band even without any oscillations implemented at the mesoscopic level. We show that depending on the parameters, the model can exhibit combinations of damped oscillations, limit cycles, or unstable oscillations. We determined bounds on model parameters that ensure stability of the oscillations simulated by the model. Finally, we estimated time-varying model parameters to capture the temporal fluctuations in magnetoencephalography activity. We show that a dynamic spectral graph modeling framework with a parsimonious set of biophysically interpretable model parameters can thereby be employed to capture oscillatory fluctuations observed in electrophysiological data in various brain states and diseases.

Functional connectivity directionality between large-scale resting-state networks across typical and non-typical trajectories in children and adolescence.

Mental disorders often emerge during adolescence and have been associated with age-related differences in connection strengths of brain networks (static functional connectivity), manifesting in non-typical trajectories of brain development. However, little is known about the direction of information flow (directed functional connectivity) in this period of functional brain progression. We employed dynamic graphical models (DGM) to estimate directed functional connectivity from resting state functional magnetic resonance imaging data on 1143 participants, aged 6 to 17 years from the healthy brain network (HBN) sample. We tested for effects of age, sex, cognitive abilities and psychopathology on estimates of direction flow. Across participants, we show a pattern of reciprocal information flow between visual-medial and visual-lateral connections, in line with findings in adults. Investigating directed connectivity patterns between networks, we observed a positive association for age and direction flow from the cerebellar to the auditory network, and for the auditory to the sensorimotor network. Further, higher cognitive abilities were linked to lower information flow from the visual occipital to the default mode network. Additionally, examining the degree networks overall send and receive information to each other, we identified age-related effects implicating the right frontoparietal and sensorimotor network. However, we did not find any associations with psychopathology. Our results suggest that the directed functional connectivity of large-scale resting-state brain networks is sensitive to age and cognition during adolescence, warranting further studies that may explore directed relationships at rest and trajectories in more fine-grained network parcellations and in different populations.

TYRE: A dynamic graph model for traffic prediction

In this paper, we study the problem of traffic forecasting, which aims to predict the future traffic state of the road network. One key challenge is that the previous approaches lack discussion of capturing temporal dependencies, as well as spatial dependencies among locations in the traffic network. In addition, the long-term traffic prediction is not satisfied. In this paper, we propose a Traffic dYnamic gRaph modEl — TYRE — which is composed of a Graph Convolutional Network with Gated and Attention mechanisms. TYRE can learn the ‘importance’ of all adjacent and distant locations, control the aggregation of adjacent and distant neighbourhood information, and learn the temporal dependencies to support long effective historical sizes. We demonstrate the validity and effectiveness of our approach on two different traffic datasets (i.e., PeMSD4 and PeMSD8). The result shows that compared to the related approaches, our model that captures temporal and spatial dependence yields substantially improved performance. When predicting traffic conditions for the next 120 min, on PeMSD8 dataset, our model shows almost 6.6% RMSE improvement, 10.9% MAE improvement, and 2.1% MAPE improvement over the previous state of the art. All source codes of this work will be publicly available at https://github.com/wzhtxy/Traffic-dYnamic-gRaph-modEl

Deep auto‐encoder based clustering algorithm for graph‐based web page recommendation system

SummaryWeb‐page recommendation system (RS) plays a major role in smart web systems. The core technology is the RS, which helps users find their interests or discover item preferences. With the rapid proliferation of information technology, the present era witnessed a remarkable growth in the collection and generation of web data. Every day, it is difficult to project the appropriate information to the user and adds severe complexity to the decision making procedure. Moreover, the existing RS techniques have limitations like huge computation time, high computational cost, over specialization, and sparsity problems. This work presents an effective graph‐based web page recommendation model to overcome these difficulties. Initially, the input data is preprocessed utilizing Natural Language Tool Kit executed in Python. The snapshot‐based dynamic graph model obtains the user‐web page relationship. Weight calculation of user‐item and the user‐user edge is also processed with this graph model. The user‐user edge similarity is used to obtain the user preference vector. Then, unsupervised‐deep auto encoder based density‐based clustering algorithm is proposed to cluster the user preference vector, which groups the visited web page details into k clusters. The web pages are recommended based on the highest score calculated by the page ranking method. Finally, the topmost visited web pages are recommended to web users based on the ranking score. “All the news”, MSNBC, and Weblog datasets are taken for evaluation. The proposed method is evaluated with respect to standard measures like F1 measure, accuracy, recall, precision, and root mean square error. Compared with other existing methods, the proposed technique achieved the highest accuracy, 97.99% for All the news datasets, 93.03% for the MSNBC database, and 90.07% for the Weblog database.

Lifelong Generative Modelling Using Dynamic Expansion Graph Model

Variational Autoencoders (VAEs) suffer from degenerated performance, when learning several successive tasks. This is caused by catastrophic forgetting. In order to address the knowledge loss, VAEs are using either Generative Replay (GR) mechanisms or Expanding Network Architectures (ENA). In this paper we study the forgetting behaviour of VAEs using a joint GR and ENA methodology, by deriving an upper bound on the negative marginal log-likelihood. This theoretical analysis provides new insights into how VAEs forget the previously learnt knowledge during lifelong learning. The analysis indicates the best performance achieved when considering model mixtures, under the ENA framework, where there are no restrictions on the number of components. However, an ENA-based approach may require an excessive number of parameters. This motivates us to propose a novel Dynamic Expansion Graph Model (DEGM). DEGM expands its architecture, according to the novelty associated with each new database, when compared to the information already learnt by the network from previous tasks. DEGM training optimizes knowledge structuring, characterizing the joint probabilistic representations corresponding to the past and more recently learned tasks. We demonstrate that DEGM guarantees optimal performance for each task while also minimizing the required number of parameters.

A Dynamic Graph Model of Strategy Learning for Predicting Human Behavior in Repeated Games

Abstract We present a model that explains the process of strategy learning by the players in repeated normal-form games. The proposed model is based on a directed weighted graph, which we define and call as the game’s dynamic graph. This graph is used as a framework by a learning algorithm that predicts which actions will be chosen by the players during the game and how the players are acting based on their gained experiences and behavioral characteristics. We evaluate the model’s performance by applying it to some human-subject datasets and measure the rate of correctly predicted actions. The results show that our model obtains a better average hit-rate compared to that of respective models. We also measure the model’s descriptive power (its ability to describe human behavior in the self-play mode) to show that our model, in contrast to the other behavioral models, is able to describe the alternation strategy in the Battle of the sexes game and the cooperating strategy in the Prisoners’ dilemma game.

A dynamic ancestral graph model and GPU-based simulation of a community based on metagenomic sampling.

In this paper, we present an ancestral graph model of the evolution of a guild in an ecological community. The model is based on a metagenomic sampling design in that a random sample is taken at the community, as opposed the taxon, level and species are discovered by genetic sequencing. The specific implementation of the model envisions an ecological guild that was founded by colonization at some point in the past that then potentially undergoes diversification by natural selection. Within the graph, species emerge and evolve through the diversification process and their densities in the graph are dynamic and governed by both ecological drift and random genetic drift, as well as differential viability. We employ the 3% sequence divergence rule at a marker locus to identify operational taxonomic units. We then explore approaches to see whether there are indirect signals of the diversification process, including population genetic and ecological approaches. In terms of population genetics, we study the joint site frequency spectrum of OTUs, as well its associated statistics. In terms of ecology, we study the species (or OTU) abundance distribution. For both, we observe deviations from neutrality, which indicates that there may be signals of diversifying selection in metagenomic studies under certain conditions. The model is available as a GPU-based computer program in C/C++ and using OpenCL, with the long-term goal of adding functionality iteratively to model large-scale eco-evolutionary processes for metagenomic data.

DGInet: Dynamic graph and interaction-aware convolutional network for vehicle trajectory prediction

This paper investigates vehicle trajectory prediction problems in real traffic scenarios by fully harnessing the spatio-temporal dependencies between multiple vehicles. The existing GCN-based trajectory predictions are often considered in a single traffic scene without time attributes, complete interaction information, dynamic graph-based model, etc. Time and interaction aware models are more challenging than the existing ones. Despite very well does the graph-based model describe the relationship between driving vehicles, the critical problem in the traffic scene is how to deeply explore the spatio-temporal characteristics of dynamic graphs. Therefore, a novel dynamic graph and interaction-aware neural network model called as DGInet is proposed by combining a semi-global graph mechanism and an M-product based graph convolutional network, which are built into novel dual-network architecture in the entire model. The DGInet is built by exploiting the dynamic interaction in depth between driving vehicles in urban traffic scenarios, and then realized by utilizing semi-global graph convolution operations on the input data cell to capture the basic spatial interaction features of the driving scene. Meanwhile, the dynamic graph is further extracted by a novel M-product approach, in which the embedding of the model is then established along with the embedding of the semi-global network to perform the final embedding. Extensive experiments have been conducted on the two public datasets, named NGSIM and Apollo respectively, to show that our approach outperforms the existing ones with better performance and less computing time. Besides the real-world Shenzhen traffic dataset, China, is also developed to verify the effectiveness of our approach.

A Parallel Algorithm Template for Updating Single-Source Shortest Paths in Large-Scale Dynamic Networks

The Single Source Shortest Path (SSSP) problem is a classic graph theory problem that arises frequently in various practical scenarios; hence, many parallel algorithms have been developed to solve it. However, these algorithms operate on static graphs, whereas many real-world problems are best modeled as dynamic networks, where the structure of the network changes with time. This gap between the dynamic graph modeling and the assumed static graph model in the conventional SSSP algorithms motivates this work. We present a novel parallel algorithmic framework for updating the SSSP in large-scale dynamic networks and implement it on the shared-memory and GPU platforms. The basic idea is to identify the portion of the network affected by the changes and update the information in a rooted tree data structure that stores the edges of the network that are most relevant to the analysis. Extensive experimental evaluations on real-world and synthetic networks demonstrate that our proposed parallel updating algorithm is scalable and, in most cases, requires significantly less execution time than the state-of-the-art recomputing-from-scratch algorithms.

Handling of resource allocation in flying ad hoc network through dynamic graph modeling

Handling of resource allocation in flying ad hoc network through dynamic graph modeling

Linear actuator–based knee exoskeleton for stand–sit–stand motions: a bond graph approach

People with knee disorders often find it difficult to perform common mobility tasks, such as stand–sit–stand motions. High knee torque is required to complete such transitions, as the chances of toppling increase during these motions. Most of the existing conventional approaches, such as wheelchairs and crutches, have failed to provide complete independence to the users. Conversely, contemporary systems like lower body exoskeletons which are bulky, complex, and expensive do not specifically target the knee joint instead of assisting other joints. Hence, there is a need to aid the knee joint using a robotic knee exoskeleton capable of accurately providing the desired knee torque. In the present work, to assist the user in performing the stand–sit–stand motions, an electromyography sensor-based four-bar knee exoskeleton actuated by a linear actuator is proposed. The modeling of the complete exoskeleton is developed using bond graph technique, as the components exist in different energy domains and it is possible to frame a dynamic bond graph model using only kinematic equations. The prototype is fabricated, and experiments are carried out on an artificial limb to prove the efficacy of the design of the current knee exoskeleton. The assistive torque developed by the actuator at the knee joint of the exoskeleton is found to be suitable to assist the wearer. As a result, little effort is required by the wearer for performing the stand–sit–stand motions. The rotation of the thigh link of the developed exoskeleton was found to be suitable for performing the stand–sit–stand activity.

Relational Reasoning Over Spatial-Temporal Graphs for Video Summarization.

In this paper, we propose a dynamic graph modeling approach to learn spatial-temporal representations for video summarization. Most existing video summarization methods extract image-level features with ImageNet pre-trained deep models. Differently, our method exploits object-level and relation-level information to capture spatial-temporal dependencies. Specifically, our method builds spatial graphs on the detected object proposals. Then, we construct a temporal graph by using the aggregated representations of spatial graphs. Afterward, we perform relational reasoning over spatial and temporal graphs with graph convolutional networks and extract spatial-temporal representations for importance score prediction and key shot selection. To eliminate relation clutters caused by densely connected nodes, we further design a self-attention edge pooling module, which disregards meaningless relations of graphs. We conduct extensive experiments on two popular benchmarks, including the SumMe and TVSum datasets. Experimental results demonstrate that the proposed method achieves superior performance against state-of-the-art video summarization methods.

Analysis of the Spatio-Temporal Dynamics of COVID-19 in Massachusetts via Spectral Graph Wavelet Theory

The rapid spread of COVID-19 disease has had a significant impact on the world. In this paper, we study COVID-19 data interpretation and visualization using open-data sources for 351 cities and towns in Massachusetts from December 6, 2020 to September 25, 2021. Because cities are embedded in rather complex transportation networks, we construct the spatio-temporal dynamic graph model, in which the graph attention neural network is utilized as a deep learning method to learn the pandemic transition probability among major cities in Massachusetts. Using the spectral graph wavelet transform (SGWT), we process the COVID-19 data on the dynamic graph, which enables us to design effective tools to analyze and detect spatio-temporal patterns in the pandemic spreading. We design a new node classification method, which effectively identifies the anomaly cities based on spectral graph wavelet coefficients. It can assist administrations or public health organizations in monitoring the spread of the pandemic and developing preventive measures. Unlike most work focusing on the evolution of confirmed cases over time, we focus on the spatio-temporal patterns of pandemic evolution among cities. Through the data analysis and visualization, a better understanding of the epidemiological development at the city level is obtained and can be helpful with city-specific surveillance.

Dynamic graph in a symbolic data framework: An account of the causal relation using COVID-19 reports and some reflections on the financial world

This article aims to evaluate a complex relation structure represented by a graph, considering a high dimensional dataset in the Symbolic Data Analysis domain. We consider COVID-19 pandemic dynamic data regarding the first semester of 2020 associated with the daily infection rate in 214 countries with remarkable trends from the financial market; thus, the empirical causality. This work is innovative as we developed a dynamic graphical model for interval data based on center-range representation, which can shrink the parametric high-dimensional time series space and uncovers causal relations. Symbolic Data Analysis provides tools to reduce data dimension through the fusion of multivariate time series in data classes, which allows considering complex information through symbolic interval multivalued variables. Additionally, the Multiregression Dynamic Model (MDM) approach estimates a Directed Acyclic Graph (DAG) which distinguishes structural changes and irregular patterns by modeling the joint learning of multivariate time series, that is, allowing heterogeneous pattern collections and simultaneously estimating relationships across series, now as symbolic interval data. Time-varying parameter estimates of allowed us to translate the influence (internal and external) of these structures dynamically, during the first months of 2020, on the interconnectedness of global regions and the spread of coronavirus worldwide. Then, descriptions of the internal variation of the regions are obtained, after the first months of the semester, reflecting the lockdown (that is, the virus transmission occurs in a generalized way worldwide, then reduced, but concentrated within the regions and not more between them). Finally, an association was sought on the impact of the disclosure (news) of COVID-19 and empirical impacts with performances of the main indices of the global financial market, in which an association between these phenomena was noticeable.

Dynamic Graph Optimization and Performance Evaluation for Delay-Tolerant Aeronautical Ad Hoc Network

With the growing interest in providing internet access and cellular connectivity in the commercial aircraft, the compatibility between Air-to-Air (A2A) communications and current Air-to-Ground (A2G) macro-cellular communications is necessary to form an Aeronautical Ad hoc Network (AANet). Due to the typical features of airborne communications, such as high mobility, long transmission range and three-dimensional macro-cells, AANet is affected by intermittent links, which results in variant delayed data transmissions. To this end, we exploit a dynamic graph approach to model such an AANet with rapidly changing topology under a realistic airborne scenario. Targeting the problem of intermittent transmissions, by employing A2A, an AANet can bear three transmission modes: direct transmission, connected relay transmission and opportunistic transmission. To investigate the potential gain of A2A transmissions in terms of the end-to-end data flow transmission delay and the data traffic amount, we formulate an integer non-linear programming problem that minimizes the average end-to-end delay from Internet Gateways (IGWs) on shore to the aircraft receivers through multiple options of the transmission modes. Through solving this minimization problem based on the realistic dynamic graph model, the tradeoffs between the end-to-end transmission delay and other key network factors related to AANet formation and resource allocation can be obtained without much compromise on the transmission delay. The results show that the average end-to-end delay of the AANet is comparable to the direct transmission delay when having a tradeoff with the data traffic amount, buffer size and the spectrum sharing.

Dynamic graph and polynomial chaos based models for contact tracing data analysis and optimal testing prescription

In this study, we address three important challenges related to disease transmissions such as the COVID-19 pandemic, namely, (a) providing an early warning to likely exposed individuals, (b) identifying individuals who are asymptomatic, and (c) prescription of optimal testing when testing capacity is limited. First, we present a dynamic-graph based SEIR epidemiological model in order to describe the dynamics of the disease propagation. Our model considers a dynamic graph/network that accounts for the interactions between individuals over time, such as the ones obtained by manual or automated contact tracing, and uses a diffusion–reaction mechanism to describe the state dynamics. This dynamic graph model helps identify likely exposed/infected individuals to whom we can provide early warnings, even before they display any symptoms and/or are asymptomatic. Moreover, when the testing capacity is limited compared to the population size, reliable estimation of individual’s health state and disease transmissibility using epidemiological models is extremely challenging. Thus, estimation of state uncertainty is paramount for both eminent risk assessment, as well as for closing the tracing-testing loop by optimal testing prescription. Therefore, we propose the use of arbitrary Polynomial Chaos Expansion, a popular technique used for uncertainty quantification, to represent the states, and quantify the uncertainties in the dynamic model. This design enables us to assign uncertainty of the state of each individual, and consequently optimize the testing as to reduce the overall uncertainty given a constrained testing budget. These tools can also be used to optimize vaccine distribution to curb the disease spread when limited vaccines are available. We present a few simulation results that illustrate the performance of the proposed framework, and estimate the impact of incomplete contact tracing data.

A method for root cause diagnosis with picture fuzzy sets based dynamic uncertain causality graph

Root cause diagnosis is of great significance to make efficient decisions in industrial production processes. It is a procedure of fusing knowledge, such as empirical knowledge, process knowledge, and mechanism knowledge. However, it is insufficient and low reliability of cause analysis methods by using crisp values or fuzzy numbers to represent uncertain knowledge. Therefore, a dynamic uncertain causality graph model (DUCG) based on picture fuzzy set (PFS) is proposed to address the problem of uncertain knowledge representation and reasoning. It combines the PFS with DUCG model to express expert doubtful ideas in a complex system. Then, a new PFS operator is introduced to characterize the importance of factors and connections among various information. Moreover, an enhanced knowledge reasoning algorithm is developed based on the PFS operators to resolve causal inference problems. Finally, a numerical example illustrates the effectiveness of the method, and the results show that the proposed model is more reliable and flexible than the existing models.