Fig 8 - uploaded by Kai Wu
Content may be subject to copyright.
Source publication
The problem of reconstructing nonlinear and complex dynamical systems from available data or time series is prominent in many fields including engineering, physical, computer, biological, and social sciences. Many methods have been proposed to address this problem and their performance is satisfactory. However, none of them can reconstruct network...
Contexts in source publication
Context 1
... find that when NS=20, m=10, these parameters have little effect on the performance of Online-NR and Online-NR can fully reconstruct the structure of the EG model. The experimental results about 1 are shown in Fig. 8. Moreover, the trend of 2 and is similar to that of 1. We also find that these parameters play an important role in the performance of Online-NR when the length of profit sequences is limited. However, due to the different characteristics of datasets, it is difficult to learn similar patterns of model selection. The proper ...
Context 2
... find that when NS=20, m=10, these parameters have little effect on the performance of Online-NR and Online-NR can fully reconstruct the structure of the EG model. The experimental results about 1 are shown in Fig. 8. Moreover, the trend of 2 and is similar to that of 1. We also find that these parameters play an important role in the performance of Online-NR when the length of profit sequences is limited. However, due to the different characteristics of datasets, it is difficult to learn similar patterns of model selection. The proper ...
Similar publications
One of the most fundamental characteristic of a complex system is its size (or volume), which, in many modelling, is represented by the number of its individual components. Complex systems under investigation nowadays are typically large and/or time-varying, rendering their identification challenging. We propose here an accurate and efficient metho...
Citations
... To analyze these spontaneous synchronization behaviors, the interactions between individuals are first simplified into a coupled complex network, and then studied using the knowledge of network synchronization. However, this is actually difficult to achieve in practice [3,4]. Realistic interactions are often invisible, which may lead to incorrect network topologies. ...
The synchronization of complex networks is believed to be closely related to network topology. How to study synchronization of real network without clear topology and parameters? In this Letter, a synchronized stability boundary equation is derived that is system- and disturbance-independent and applicable to arbitrarily coupled grids. The results imply that spontaneous synchronization in a network may be network independent. These conclusions provide new research paths for network synchronization and analyze the synchronization stability of grids in a unified way.
... The distinction between FOBOS and RDA lies in their computation methods, with the former calculating the current gradient and L1 regularization terms, while the latter employs cumulative gradients and L1 regularization terms; furthermore, there are variations in the restrictions imposed on W. Despite achieving increased sparsity, RDA exhibits a slight decrease in accuracy. Therefore, as its name suggests, the FTRL algorithm incorporates and balances the advantages of the previous algorithms while considering both the accuracy of FOBOS and the sparsity of RDA [34,35]. Its feature-weight-updating formula can be expressed as follows: ...
During industrial production activities, industrial products serve as critical resources whose performance is subject to various external factors and usage conditions. To ensure uninterrupted production processes and to guarantee the safety of the production personnel, a real-time analysis of the industrial product quality and subsequent decision making are essential. Conventional detection methods have inherent limitations in meeting the real-time demands of processing large volumes of data and achieving high response speeds. For instance, the regular inspection and maintenance of cars can be time-consuming and labor-intensive if performed manually. Furthermore, monitoring the damage situation of bearings in real time through a manual inspection may lead to delays and may hinder production efficiency. Therefore, this paper presents online machine-learning-based methods to address these two practical problems and simulates them on various datasets to meet the requirements of efficiency and speed. Prior to being fed into the network for training, the data undergo identity parsing to transform them into easily identifiable streaming data. The training process demonstrates that online machine learning ensures timely model updates as small batches of data are sent to the network. The test results indicate that the online learning method exhibits highly stable and effective performance, optimizing the training process.
... The sparsity constraints mean that the coefficient vector ξ j is sparse with only a few nonzero entries, each representing a subsistent item in the function library. Subsequently, we exploit the follow-the-regularized-leader (FTRL)-proximal [37][38][39] style methodology to optimize the outcome by considering solely one instance each time. ...
Discovering the governing equations of evolving systems from available observations is essential and challenging. In this paper, we consider a different scenario: discovering governing equations from streaming data. Current methods struggle to discover governing differential equations with considering measurements as a whole, leading to failure to handle this task. We propose an online modeling method capable of handling samples one by one sequentially by modeling streaming data instead of processing the entire data set. The proposed method performs well in discovering ordinary differential equations and partial differential equations from streaming data. Evolving systems are changing over time, which invariably changes with system status. Thus finding the exact change points is critical. The measurement generated from a changed system is distributed dissimilarly to before; hence the difference can be identified by the proposed method. Our proposal is competitive in identifying the change points and discovering governing differential equations in three hybrid systems and two switching linear systems.
... In Fig. 1, in the upper-left corner of the network, the solid lines represent known network links, the dotted lines represent hidden network links, and the network topology corresponds to the matrix in the upper-right corner. The reconstruction process of debt network topology among enterprises involves determining the complete network topology matrix according to the element-missing matrix of the debt network topology, and determining the implied network links to determine the accurate enterprise debt network topology 16,17 . When the number n of enterprises is sufficiently large, the network structure will show complex network properties, such as self-organization, self-similarity and small world. ...
This study aims at the problem of reconstruction the unknown links in debt networks among enterprises. We use the topological matrix of the enterprise debt network as the object of reconstruction and use the time series data of accounts receivable and payable as input and output information in the debt network to establish an underdetermined linear system about the topological matrix of the debt network. We establish an iteratively reweighted least-squares algorithm, which is an algorithm in compressed sensing. This algorithm uses reweighted ℓ2-minimization to approximate ℓ1-norm of the target vectors. We solve the ℓ1-minimization problem of the underdetermined linear system using the iteratively reweighted least-squares algorithm and obtain the reconstructed topological matrix of the debt network. Simulation experiments show that the topology matrix reconstruction method of enterprise debt networks based on compressed sensing can reconstruct over 70% of the unknown network links, and the error is controlled within 2%.
... Current methods require a user-defined parameter to control the trade-off between the natural sparsity of interaction topologies and measurement error [27,48,49], which has a great effect on the performance. Here, we develop a special threshold method for K-similarity according to its right-skewed distribution of random results. ...
Interactions and dynamics are critical mechanisms for multi-agent systems to achieve complex intelligence through the cooperation of simple agents. Yet, inferring interactions of the multi-agent system is still a common and open problem. A new method named K-similarity is designed to measure the global relative similarities for inferring the interactions among multiple agents in this paper. K-similarity is defined to be a synthetic measure of relative similarity on each observation snapshot where regular distances are nonlinearly mapped into a network. Therefore, K-similarity contains the global relative similarity information, and the interaction topology can be inferred from the similarity matrix. It has the potential to transform into distance strictly and detect multi-scale information with various K strategies. Therefore, K-similarity can be flexibly applied to various synchronized dynamical systems with fixed, switching, and time-varying topologies. In the experiments, K-similarity outperforms four benchmark methods in accuracy in most scenarios on both simulated and real datasets, and shows strong stability towards outliers. Furthermore, according to the property of K-similarity we develop a Gaussian Mixture Model (GMM)-based threshold to select probable interactions. Our method contributes to not only similarity measurement in multi-agent systems, but also other global similarity measurement problems.
... Pan et al. [24] considered satellite node constraints, such as visibility, connectable time, and connectivity, and developed satellite network topology links and nodes model, and an improved space information network topology reconstruction algorithm based on the artificial bee colony algorithm. Wu et al. [25] developed an online network reconfiguration method based on streaming data. In the field of distribution network reconfiguration, Wang [26] proposed a reconstruction solution with the best overall performance over a long-term operating period based on a comprehensive analysis of operation scenarios, which reduced network loss of traditional single-operation scenarios. ...
Network reconfiguration is an important means of improving network invulnerability. However, most existing network reconfiguration methods fail to consider node importance, edge importance, and hierarchical characteristics, and the local and global information of command and control (C2) networks are difficult to satisfy comprehensively. Therefore, this study designed a hierarchy-entropy-based method for reconfiguring C2 networks. By combining hierarchical and operational link entropy, the probability of inter-node edge reconfiguration based on hierarchy entropy is proposed. Additionally, methods for calculating the node level-up, cross-level, and swap degrees, and a portfolio reconfiguration strategy are proposed. Finally, to validate the proposed method, a case study was simulated, and the repair probability, adjustable parameters, and reconfiguration effects of the different reconfiguration methods and modes were determined. The comparison results demonstrate that the proposed algorithm improves the reconfiguration effect and reduces the reconfiguration cost.
... Solving the above equation has been proven to be an NP-hard problem [13]. So far, a lot of approaches have been proposed to solve small-scale NRPs, where the regularization term in (2) is replaced by X 1 in most cases [14], [15]. Napoletani and Sauer [16] proposed a constrained optimization technique based on chaotic time-series analysis to reconstruct nonlinear dynamical networks, which appeared to be limited to small-scale networks with sparse connections. ...
Currently, the problem of uncovering complex network structure and dynamics from time series is prominent in many fields. Despite the recent progress in this area, reconstructing large-scale networks from limited data remains a tough problem. Existing works treat connections of nodes as continuous values, leaving a challenge of setting a proper cut-off value to distinguish whether the connections exist or not. Besides, their performances on large-scale networks are far from satisfactory. Considering the reconstruction error and sparsity as two objectives, this paper proposes a subspace learning based evolutionary multiobjective network reconstruction algorithm, termed as SLEMO-NR, to solve the aforementioned problems. In the evolutionary process, we assume that binary-coded individuals obey the Bernoulli distribution and can use the probability and natural parameter as the alternative representations. Moreover, our approach utilizes the logistic principal component analysis (LPCA) to learn a subspace containing the features of network structure. The offspring solutions are generated in the learned subspace and then can be mapped back to the original space via LPCA. Benefitting from the alternative representations, a preference-based local search operator is proposed to concentrate on finding solutions approximate to the true sparsity. The experimental results on synthetic networks and six real-world networks demonstrate that, due to the well-learned network structure subspace and the preference-based strategy, our approach is effective in reconstructing large-scale networks compared to six existing methods.
... Solving the above equation has been proven to be an NP-hard problem [13]. So far, a lot of approaches have been proposed to solve small-scale NRPs, where the regularization term in (2) is replaced by X 1 in most cases [14], [15]. Napoletani and Sauer [16] proposed a constrained optimization technique based on chaotic time-series analysis to reconstruct nonlinear dynamical networks, which appeared to be limited to small-scale networks with sparse connections. ...
Currently, the problem of uncovering complex network structure and dynamics from time series is prominent in many fields. Despite the recent progress in this area, reconstructing large-scale networks from limited data remains a tough problem. Existing works treat connections of nodes as continuous values, leaving a challenge of setting a proper cut-off value to distinguish whether the connections exist or not. Besides, their performances on large-scale networks are far from satisfactory. Considering the reconstruction error and sparsity as two objectives, this paper proposes a subspace learning based evolutionary multiobjective network reconstruction algorithm, termed as SLEMO-NR, to solve the aforementioned problems. In the evolutionary process, we assume that binary-coded individuals obey the Bernoulli distribution and can use the probability and natural parameter as the alternative representations. Moreover, our approach utilizes the logistic principal component analysis (LPCA) to learn a subspace containing the features of network structure. The offspring solutions are generated in the learned subspace and then can be mapped back to the original space via LPCA. Benefitting from the alternative representations, a preference-based local search operator is proposed to concentrate on finding solutions approximate to the true sparsity. The experimental results on synthetic networks and six real-world networks demonstrate that, due to the well-learned network structure subspace and the preference-based strategy, our approach is effective in reconstructing large-scale networks compared to six existing methods.
... [13]- [20], and online learning [21]. Current methods for detecting communities when the network structure is unobservable typically involve a complicated process that is highly sensitive to specific design decisions and parameter selection. ...
This paper focuses on inferring network structure and community structure from the dynamics of the nonlinear and complex dynamical systems, which is prominent in many fields. Many methods have been proposed to solely address these two problems, but none of them consider explicit shareable knowledge across these two tasks. Inspired by the fact that a more precise network structure may promote the accuracy of community discovery and the better communities may promote the performance of network reconstruction (NR), this paper develops an evolutionary multitasking framework to make full use of explicit shareable knowledge among these two tasks to improve their performance; we refer to this framework as EMTNRCD. In EMTNRCD, we first establish these two tasks as a multitasking NR and community detection (CD) problem where one mission is to reconstruct network structure from dynamics and the other is to discover communities from dynamics. In the process of EMTNRCD, the NR task explicitly transfers several better network structures for the CD task and the CD task explicitly transfers a better community structure to assist the NR task, which improves the reconstruction accuracy of the NR task and the community division quality of the CD task. Moreover, to transfer knowledge from the study of the NR task to the CD task, EMTNRCD models the study of CD from dynamics as the problem of finding communities in the dynamic network and then decides whether to conduct knowledge transfer across tasks. This paper also designs a test suite for multitasking NR and CD problems (MTNRCDPs) to verify the performance of EMTNRCD. The experimental results have demonstrated that joint NR with CD has a synergistic effect.
... So far, a lot of approaches have been proposed to solve small-scale NRPs, where the regularization term in (2) is replaced by X 1 in most cases [14], [15]. Napoletani et al. [16] proposed a constrained optimization technique based on chaotic time-series analysis to reconstruct nonlinear dynamical networks, which appeared to be limited to small-scale networks with sparse connections. ...
Currently, the problem of uncovering complex network structure and dynamics from time series is prominent in many fields. Despite the recent progress in this area, reconstructing
large-scale networks from limited data remains a tough problem. Existing works treat connections of nodes as continuous values, leaving a challenge of setting a proper cut-off value to distinguish whether the connections exist or not. Besides, their performances on large-scale networks are far from satisfactory. Considering the reconstruction error and sparsity as two objectives, this paper proposes a subspace learning based evolutionary
multiobjective network reconstruction algorithm, termed as SLEMO-NR, to solve the aforementioned problems. In the evolutionary process, we assume that binary-coded individuals obey the Bernoulli distribution and can use the probability and natural parameter as the alternative representations. Moreover, our approach utilizes the logistic principal component analysis (LPCA) to learn a subspace containing the features of network structure. The offspring solutions are generated in the learned subspace and then can be mapped back to the original space via LPCA. Benefitting from the alternative representations, a preference-based local search operator is proposed to concentrate on finding solutions approximate to the true sparsity. The experimental results on synthetic networks and six real-world networks demonstrate that, due to the well-learned network structure subspace and the preference-based strategy, our approach is effective in reconstructing large-scale networks compared to six existing methods.
