arXiv subjects

Linyuan Lü

Publications and source records attributed to Linyuan Lü.

At least 37 records · Page 2Linked to original sources

arXiv2025

Weighted cycle-based identification of influential node groups in complex networks

Identifying influential node groups in complex networks is crucial for optimizing information dissemination, epidemic control, and viral marketing. However, traditional centrality-based methods often focus on individual nodes, resulting in overlapping influence zones and diminished collective effectiveness. To overcome these limitations, we propose Weighted Cycle (WCycle), a novel indicator that incorporates basic cycle structures and node behavior traits (edge weights) to comprehensively assess node importance. WCycle effectively identifies spatially dispersed and structurally diverse key node group, thereby reducing influence redundancy and enhancing network-wide propagation. Extensive experiments on six real-world networks demonstrate WCycle's superior performance compared to five benchmark methods across multiple evaluation dimensions, including influence propagation efficiency, structural differentiation, and cost-effectiveness. The findings highlight WCycle's robustness and scalability, establishing it as a promising tool for complex network analysis and practical applications requiring effective influence maximization.

cs.SI↗

arXiv2024

Uncovering key predictors of high-growth firms via explainable machine learning

Predicting high-growth firms has attracted increasing interest from the technological forecasting and machine learning communities. Most existing studies primarily utilize financial data for these predictions. However, research suggests that a firm's research and development activities and its network position within technological ecosystems may also serve as valuable predictors. To unpack the relative importance of diverse features, this paper analyzes financial and patent data from 5,071 firms, extracting three categories of features: financial features, technological features of granted patents, and network-based features derived from firms' connections to their primary technologies. By utilizing ensemble learning algorithms, we demonstrate that incorporating financial features with either technological, network-based features, or both, leads to more accurate high-growth firm predictions compared to using financial features alone. To delve deeper into the matter, we evaluate the predictive power of each individual feature within their respective categories using explainable artificial intelligence methods. Among non-financial features, the maximum economic value of a firm's granted patents and the number of patents related to a firms' primary technologies stand out for their importance. Furthermore, firm size is positively associated with high-growth probability up to a certain threshold size, after which the association plateaus. Conversely, the maximum economic value of a firm's granted patents is positively linked to high-growth probability only after a threshold value is exceeded. These findings elucidate the complex predictive role of various features in forecasting high-growth firms and could inform technological resource allocation as well as investment decisions.

physics.soc-ph↗

arXiv2024

A universal optimization framework based on cycle ranking for influence maximization in complex networks

Influence maximization aims to identify a set of influential individuals, referred to as influencers, as information sources to maximize the spread of information within networks, constituting a vital combinatorial optimization problem with extensive practical applications and sustained interdisciplinary interest. Diverse approaches have been devised to efficiently address this issue, one of which involves selecting the influencers from a given centrality ranking. In this paper, we propose a novel optimization framework based on ranking basic cycles in networks, capable of selecting the influencers from diverse centrality measures. The experimental results demonstrate that, compared to directly selecting the top-k nodes from centrality sequences and other state-of-the-art optimization approaches, the new framework can expand the dissemination range by 1.5 to 3 times. Counterintuitively, it exhibits minimal hub property, with the average distance between influencers being only one-third of alternative approaches, regardless of the centrality metrics or network types. Our study not only paves the way for novel strategies in influence maximization but also underscores the unique potential of underappreciated cycle structures.

cs.SI↗

arXiv2024

A Fast Maximum Clique Algorithm Based on Network Decomposition for Large Sparse Networks

Finding maximum cliques in large networks is a challenging combinatorial problem with many real-world applications. We present a fast algorithm to achieve the exact solution for the maximum clique problem in large sparse networks based on efficient graph decomposition. A bunch of effective techniques is being used to greatly prune the graph and a novel concept called Complete-Upper-Bound-Induced Subgraph (CUBIS) is proposed to ensure that the structures with the potential to form the maximum clique are retained in the process of graph decomposition. Our algorithm first pre-prunes peripheral nodes, subsequently, one or two small-scale CUBISs are constructed guided by the core number and current maximum clique size. Bron-Kerbosch search is performed on each CUBIS to find the maximum clique. Experiments on 50 empirical networks with a scale of up to 20 million show the CUBIS scales are largely independent of the original network scale. This enables an approximately linear runtime, making our algorithm amenable for large networks. Our work provides a new framework for effectively solving maximum clique problems on massive sparse graphs, which not only makes the graph scale no longer the bottleneck but also shows some light on solving other clique-related problems.

cs.SI↗

arXiv2024

Local dominance unveils clusters in networks

Clusters or communities can provide a coarse-grained description of complex systems at multiple scales, but their detection remains challenging in practice. Community detection methods often define communities as dense subgraphs, or subgraphs with few connections in-between, via concepts such as the cut, conductance, or modularity. Here we consider another perspective built on the notion of local dominance, where low-degree nodes are assigned to the basin of influence of high-degree nodes, and design an efficient algorithm based on local information. Local dominance gives rises to community centers, and uncovers local hierarchies in the network. Community centers have a larger degree than their neighbors and are sufficiently distant from other centers. The strength of our framework is demonstrated on synthesized and empirical networks with ground-truth community labels. The notion of local dominance and the associated asymmetric relations between nodes are not restricted to community detection, and can be utilised in clustering problems, as we illustrate on networks derived from vector data.

physics.soc-ph↗

arXiv2024

Beyond network centrality: Individual-level behavioral traits for predicting information superspreaders in social media

Understanding the heterogeneous role of individuals in large-scale information spreading is essential to manage online behavior as well as its potential offline consequences. To this end, most existing studies from diverse research domains focus on the disproportionate role played by highly-connected ``hub" individuals. However, we demonstrate here that information superspreaders in online social media are best understood and predicted by simultaneously considering two individual-level behavioral traits: influence and susceptibility. Specifically, we derive a nonlinear network-based algorithm to quantify individuals' influence and susceptibility from multiple spreading event data. By applying the algorithm to large-scale data from Twitter and Weibo, we demonstrate that individuals' estimated influence and susceptibility scores enable predictions of future superspreaders above and beyond network centrality, and reveal new insights on the network position of the superspreaders.

cs.SI↗

arXiv2024

The role of susceptible individuals in spreading dynamics

Exploring the internal mechanism of information spreading is critical for understanding and controlling the process. Traditional spreading models often assume individuals play the same role in the spreading process. In reality, however, individuals' diverse characteristics contribute differently to the spreading performance, leading to a heterogeneous infection rate across the system. To investigate network spreading dynamics under heterogeneous infection rates, we integrate two individual-level features -- influence (i.e., the ability to influence neighbors) and susceptibility (i.e., the extent to be influenced by neighbors) -- into the independent cascade model. Our findings reveal significant differences in spreading performance under heterogeneous and constant infection rates, with traditional structural centrality metrics proving more effective in the latter scenario. Additionally, we take the constant and heterogeneous infection rates into a state-of-the-art maximization algorithm, the well-known TIM algorithm, and find the seeds selected by heterogeneous infection rates are more dispersed compared to those under constant rates. Lastly, we find that both individuals' influence and susceptibility are vital to the spreading performance. Strikingly, susceptible individuals are particularly important to spreading when information is disseminated by social celebrities. By integrating influence and susceptibility into the spreading model, we gain a more profound understanding of the underlying mechanisms driving information spreading.

cs.SI↗

arXiv2024

Higher-order Graph Convolutional Network with Flower-Petals Laplacians on Simplicial Complexes

Despite the recent successes of vanilla Graph Neural Networks (GNNs) on various tasks, their foundation on pairwise networks inherently limits their capacity to discern latent higher-order interactions in complex systems. To bridge this capability gap, we propose a novel approach exploiting the rich mathematical theory of simplicial complexes (SCs) - a robust tool for modeling higher-order interactions. Current SC-based GNNs are burdened by high complexity and rigidity, and quantifying higher-order interaction strengths remains challenging. Innovatively, we present a higher-order Flower-Petals (FP) model, incorporating FP Laplacians into SCs. Further, we introduce a Higher-order Graph Convolutional Network (HiGCN) grounded in FP Laplacians, capable of discerning intrinsic features across varying topological scales. By employing learnable graph filters, a parameter group within each FP Laplacian domain, we can identify diverse patterns where the filters' weights serve as a quantifiable measure of higher-order interaction strengths. The theoretical underpinnings of HiGCN's advanced expressiveness are rigorously demonstrated. Additionally, our empirical investigations reveal that the proposed model accomplishes state-of-the-art performance on a range of graph tasks and provides a scalable and flexible solution to explore higher-order interactions in graphs. Codes and datasets are available at https://github.com/Yiminghh/HiGCN.

cs.LG↗

arXiv2024

Network higher-order structure dismantling

Diverse higher-order structures, foundational for supporting a network's "meta-functions", play a vital role in structure, functionality, and the emergence of complex dynamics. Nevertheless, the problem of dismantling them has been consistently overlooked. In this paper, we introduce the concept of dismantling higher-order structures, with the objective of disrupting not only network connectivity but also eradicating all higher-order structures in each branch, thereby ensuring thorough functional paralysis. Given the diversity and unknown specifics of higher-order structures, identifying and targeting them individually is not practical or even feasible. Fortunately, their close association with k-cores arises from their internal high connectivity. Thus, we transform higher-order structure measurement into measurements on k-cores with corresponding orders. Furthermore, we propose the Belief Propagation-guided High-order Dismantling (BPDH) algorithm, minimizing dismantling costs while achieving maximal disruption to connectivity and higher-order structures, ultimately converting the network into a forest. BPDH exhibits the explosive vulnerability of network higher-order structures, counterintuitively showcasing decreasing dismantling costs with increasing structural complexity. Our findings offer a novel approach for dismantling malignant networks, emphasizing the substantial challenges inherent in safeguarding against such malicious attacks.

physics.soc-ph↗

arXiv2023

Identifying vital nodes through augmented random walks on higher-order networks

Empirical networks possess considerable heterogeneity of node connections, resulting in a small portion of nodes playing crucial roles in network structure and function. Yet, how to characterize nodes' influence and identify vital nodes is by far still unclear in the study of networks with higher-order interactions. In this paper, we introduce a multi-order graph obtained by incorporating the higher-order bipartite graph and the classical pairwise graph, and propose a Higher-order Augmented Random Walk (HoRW) model through random walking on it. This representation preserves as much information about the higher-interacting network as possible. The results indicate that the proposed method effectively addresses the localization problem of certain classical centralities. In contrast to random walks along pairwise interactions only, performing more walks along higher-order interactions assists in not only identifying the most important nodes but also distinguishing nodes that ranked in the middle and bottom. Our method outperforms classical centralities in identifying vital nodes and can scale to various tasks in networks, including information spread maximization and network dismantling problems. The proposed higher-order representation and the random walk model provide novel insights and potent tools for studying higher-order mechanisms and functionality.

cs.SI↗

arXiv2023

Cooperative Network Learning for Large-Scale and Decentralized Graphs

Graph research, the systematic study of interconnected data points represented as graphs, plays a vital role in capturing intricate relationships within networked systems. However, in the real world, as graphs scale up, concerns about data security among different data-owning agencies arise, hindering information sharing and, ultimately, the utilization of graph data. Therefore, establishing a mutual trust mechanism among graph agencies is crucial for unlocking the full potential of graphs. Here, we introduce a Cooperative Network Learning (CNL) framework to ensure secure graph computing for various graph tasks. Essentially, this CNL framework unifies the local and global perspectives of GNN computing with distributed data for an agency by virtually connecting all participating agencies as a global graph without a fixed central coordinator. Inter-agency computing is protected by various technologies inherent in our framework, including homomorphic encryption and secure transmission. Moreover, each agency has a fair right to design or employ various graph learning models from its local or global perspective. Thus, CNL can collaboratively train GNN models based on decentralized graphs inferred from local and global graphs. Experiments on contagion dynamics prediction and traditional graph tasks (i.e., node classification and link prediction) demonstrate that our CNL architecture outperforms state-of-the-art GNNs developed at individual sites, revealing that CNL can provide a reliable, fair, secure, privacy-preserving, and global perspective to build effective and personalized models for network applications. We hope this framework will address privacy concerns in graph-related research and integrate decentralized graph data structures to benefit the network research community in cooperation and innovation.

cs.LG↗

arXiv2023

Unveiling Explosive Vulnerability of Networks through Edge Collective Behavior

Edges, binding together nodes within networks, have the potential to induce dramatic transitions when specific collective failure behaviors emerge. These changes, initially unfolding covertly and then erupting abruptly, pose substantial, unforeseeable threats to networked systems, and are termed explosive vulnerability. Thus, identifying influential edges capable of triggering such drastic transitions, while minimizing cost, is of utmost significance. Here, we address this challenge by introducing edge collective influence (ECI), which builds upon the optimal percolation theory applied to line graphs. ECI embodies features of both optimal and explosive percolation, involving minimized removal costs and explosive dismantling tactic. Furthermore, we introduce two improved versions of ECI, namely IECI and IECIR, tailored for objectives of hidden and fast dismantling, respectively, with their superior performance validated in both synthetic and empirical networks. Finally, we present a dual competitive percolation (DCP) model, whose reverse process replicates the explosive dismantling process and the trajectory of the cost function of ECI, elucidating the microscopic mechanisms enabling ECI's optimization. ECI and the DCP model demonstrate the profound connection between optimal and explosive percolation. This work significantly deepens our comprehension of percolation and provides valuable insights into the explosive vulnerabilities arising from edge collective behaviors.

physics.soc-ph↗

arXiv2023

Stock network inference: A framework for market analysis from topology perspective

From a complex network perspective, investigating the stock market holds paramount significance as it enables the systematic revelation of topological features inherent in the market. This approach is crucial in exploring market interconnectivity, systemic risks, portfolio management, and structural evolution. However, prevailing methodologies for constructing networks based on stock data rely on threshold filtering, often needing help to uncover intricate underlying associations among stocks. To address this, we introduce the Stock Network Inference Framework (SNIF), which leverages a self-encoding mechanism. Specifically, the Stock Network Inference Encoder (SNIE) facilitates network construction, while the Movement Prediction Decoder (MPD) enhances movement forecasting. This integrated process culminates in the inference of a stock network, exhibiting remarkable performance across applications such as market structure analysis, stock movement prediction, portfolio construction, and community evolution analysis. Our approach streamlines the automatic construction of stock networks, liberating the process from threshold dependencies and eliminating the need for additional financial indicators. Incorporating Graph Convolutional Network (GCN) and Long Short-Term Memory (LSTM) models within the SNIF framework, we effectively unearth deep-seated associations among stocks, augmenting the toolset available for comprehensive financial market research. This integration empowers our methodology to automatically construct stock networks without threshold dependencies or reliance on additional economic indicators.

physics.soc-ph↗

arXiv2023

A generalized simplicial model and its application

Higher-order structures, consisting of more than two individuals, provide a new perspective to reveal the missed non-trivial characteristics under pairwise networks. Prior works have researched various higher-order networks, but research for evaluating the effects of higher-order structures on network functions is still scarce. In this paper, we propose a framework to quantify the effects of higher-order structures (e.g., 2-simplex) and vital functions of complex networks by comparing the original network with its simplicial model. We provide a simplicial model that can regulate the quantity of 2-simplices and simultaneously fix the degree sequence. Although the algorithm is proposed to control the quantity of 2-simplices, results indicate it can also indirectly control simplexes more than 2-order. Experiments on spreading dynamics, pinning control, network robustness, and community detection have shown that regulating the quantity of 2-simplices changes network performance significantly. In conclusion, the proposed framework is a general and effective tool for linking higher-order structures with network functions. It can be regarded as a reference object in other applications and can deepen our understanding of the correlation between micro-level network structures and global network functions.

physics.soc-ph↗

arXiv2023

Influential Simplices Mining via Simplicial Convolutional Network

Simplicial complexes have recently been in the limelight of higher-order network analysis, where a minority of simplices play crucial roles in structures and functions due to network heterogeneity. We find a significant inconsistency between identifying influential nodes and simplices. Therefore, it remains elusive how to characterize simplices' influence and identify influential simplices, despite the relative maturity of research on influential nodes (0-simplices) identification. Meanwhile, graph neural networks (GNNs) are potent tools that can exploit network topology and node features simultaneously, but they struggle to tackle higher-order tasks. In this paper, we propose a higher-order graph learning model, named influential simplices mining neural network (ISMnet), to identify vital h-simplices in simplicial complexes. It can tackle higher-order tasks by leveraging novel higher-order presentations: hierarchical bipartite graphs and higher-order hierarchical (HoH) Laplacians, where targeted simplices are grouped into a hub set and can interact with other simplices. Furthermore, ISMnet employs learnable graph convolutional operators in each HoH Laplacian domain to capture interactions among simplices, and it can identify influential simplices of arbitrary order by changing the hub set. Empirical results demonstrate that ISMnet significantly outperforms existing methods in ranking 0-simplices (nodes) and 2-simplices. In general, this novel framework excels in identifying influential simplices and promises to serve as a potent tool in higher-order network analysis.

cs.SI↗

arXiv2023

Higher-order link prediction via local information

Link prediction has been widely studied as an important research direction. Higher-order link prediction has gained especially significant attention since higher-order networks provide a more accurate description of real-world complex systems. However, higher-order networks contain more complex information than traditional pairwise networks, making the prediction of higher-order links a formidable challenging task. Recently, researchers have discovered that local features have advantages over long-range features in higher-order link prediction. Therefore, it is necessary to develop more efficient and concise higher-order link prediction algorithms based on local features. In this paper, we proposed two similarity metrics via local information, simplicial decomposition weight (SDW) and closed ratio weight (CRW), to predict possible future higher-order interactions (simplices) in simplicial networks. These two algorithms capture local higher-order information at two aspects: simplex decomposition and cliques' state (closed or open). We tested their performance in eight empirical simplicial networks, and the results show that our proposed metrics outperform other benchmarks in predicting third-order and fourth-order interactions (simplices) in most cases. In addition, we explore the robustness of the proposed algorithms, and the results suggest that the performance of these novel algorithms is advanced under different sizes of training sets.

cs.SI↗

arXiv2022

Complete cavity map of the C. elegans connectome

Network structure or topology is the basis for understanding complex systems. Recently, higher-order structures have been considered as a new research direction that can provide new perspectives and phenomena. However, most existing studies have focused on simple 1-cycles and triangles, and few have explored the higher-order and non-trivial cycles. The current study focused on the cavity, which is a non-trivial cycle. We proposed a method to compute cavities with different orders based on pruning in the boundary matrix and calculated all cavities of the neural network of Caenorhabditis elegans (\emph{C. elegans}) neural network. This study reports for the first time a complete cavity map of C. elegans neural network, developing a new method for mining higher-order structures that can be applied by researchers in neuroscience, network science and other interdisciplinary fields to explore higher-order structural markers of complex systems.

q-bio.NC↗

arXiv2022

Expression might be enough: representing pressure and demand for reinforcement learning based traffic signal control

Many studies confirmed that a proper traffic state representation is more important than complex algorithms for the classical traffic signal control (TSC) problem. In this paper, we (1) present a novel, flexible and efficient method, namely advanced max pressure (Advanced-MP), taking both running and queuing vehicles into consideration to decide whether to change current signal phase; (2) inventively design the traffic movement representation with the efficient pressure and effective running vehicles from Advanced-MP, namely advanced traffic state (ATS); and (3) develop a reinforcement learning (RL) based algorithm template, called Advanced-XLight, by combining ATS with the latest RL approaches, and generate two RL algorithms, namely "Advanced-MPLight" and "Advanced-CoLight" from Advanced-XLight. Comprehensive experiments on multiple real-world datasets show that: (1) the Advanced-MP outperforms baseline methods, and it is also efficient and reliable for deployment; and (2) Advanced-MPLight and Advanced-CoLight can achieve the state-of-the-art.

cs.AI↗