Chengli Zhao

• National University of Defense Technology

Differential equations are widely used to describe complex dynamical systems with evolving parameters in nature and engineering. Effectively learning a family of maps from the parameter function to the system dynamics is of great significance. In this study, we propose a novel learning framework of symbolic continuous-depth neural networks, termed...

Differential equations are widely used to describe complex dynamical systems with evolving parameters in nature and engineering. Effectively learning a family of maps from the parameter function to the system dynamics is of great significance. In this study, we propose a novel learning framework of symbolic continuous-depth neural networks, termed...

Modeling complex systems using standard neural ordinary differential equations (NODEs) often faces some essential challenges, including high computational costs and susceptibility to local optima. To address these challenges, we propose a simulation-free framework, called Fourier NODEs (FNODEs), that effectively trains NODEs by directly matching th...

Modeling complex systems using standard neu-ral ordinary differential equations (NODEs) often faces some essential challenges, including high computational costs and susceptibility to local optima. To address these challenges, we propose a simulation-free framework, called Fourier NODEs (FNODEs), that effectively trains NODEs by directly matching t...

In many fields, accurate prediction of cascade outbreaks during their early stages of propagation is of paramount importance. Based on percolation theory, we propose a global propagation probability algorithm that effectively estimates the probability of information spreading from source nodes to the giant component. Building on this, we further in...

Recently, machine learning methods, including reservoir computing (RC), have been tremendously successful in predicting complex dynamics in many fields. However, a present challenge lies in pushing for the limit of prediction accuracy while maintaining the low complexity of the model. Here, we design a data-driven, model-free framework named higher...

Revealing the character of journals based on citation data remains an interesting issue nowadays. It aims to establish a reasonable journal evaluation system and provides suitable journals for scholars to submit to. As for traditional methods, based on first-order citation networks, they are poor at describing the multivariate sequential interactio...

Event detection is one of the most important areas of complex network research. It aims to identify abnormal points in time corresponding to social events. Traditional methods of event detection, based on first-order network models, are poor at describing the multivariate sequential interactions of components in complex systems and at accurately id...

Human mobility data are indispensable in modeling large-scale epidemics, especially in predicting the spatial spread of diseases and in evaluating spatial heterogeneity intervention strategies. However, statistical data that can accurately describe large-scale population migration are often difficult to obtain. We propose an algorithm model based o...

The social network is closely related to people’s lives. And social events are the products of the human subjective initiative during the evolution of networks. Therefore, there is a close correlation between social events and network evolution. This paper studies the characteristics of network evolution corresponding to social events from the pers...

Identification of multiple influential spreaders on complex networks is of great significance, which can help us speed up information diffusion and prevent disease from spreading to some extent. The traditional top-k strategy to solve an influence maximization problem based on node centrality is unsuitable for selecting several spreaders simultaneo...

Anomaly detection is a classic problem on complex networks. An anomaly detection method based on network projection is proposed in this study on networks with fundamental bipartite connection relationships and repeated interactions, such as the Internet and computer networks. First of all, two network partition algorithms are advanced to discover t...

With the increasing network threat, network anomaly detection has become a very challenging and indispensable task. In this article, we propose an anomaly detection algorithm through modeling computer network as temporal network. The active subnetwork is extracted from the original computer communication network and then projected to an undirected...

Based on percolation theory and the independent cascade model, this paper considers the selection of the optimal propagation source when the propagation probability is greater than the percolation threshold. First, based on the percolation characteristics of real networks, this paper presents an iterative algorithm of linear complexity to solve the...

Source localization is a typical inverse problem in complex networks, which is widely used in disease outbreak, rumor propagation and pollutants spread. In this paper, we propose that, based on network topology and the times at which the diffusion reached partial nodes, it is easy to identify the source. The results show that in six different netwo...

Simultaneous outbreaks of contagion are a great threat against human life, resulting in great panic in society. It is urgent for us to find an efficient multiple sources localization method with the aim of studying its pathogenic mechanism and minimizing its harm. However, our ability to locate multiple sources is strictly limited by incomplete inf...

Locating the source that undergoes a diffusion-like process is a fundamental and challenging problem in complex network, which can help inhibit the outbreak of epidemics among humans, suppress the spread of rumors on the Internet, prevent cascading failures of power grids, etc. However, our ability to accurately locate the diffusion source is stric...

Epidemic source localization is one of the most meaningful areas of research in complex networks, which helps solve the problem of infectious disease spread. Limited by incomplete information of nodes and inevitable randomness of the spread process, locating the epidemic source becomes a little difficult. In this paper, we propose an efficient algo...

With the deep understanding of the time-varying characteristics of real systems, research studies focusing on the temporal network spring up like mushrooms. Community detection is an accompanying and meaningful problem in the temporal network, but the analysis of this problem is still in its developing stage. In this paper, we proposed a temporal s...

We thoroughly study the robustness of partially interdependent networks when suffering attack combinations of random, targeted, and localized attacks. We compare analytically and numerically the robustness of partially interdependent networks with a broad range of parameters including coupling strength, attack strength, and network type. We observe...

Identification of multiple influential seed spreaders is one of the most promising domain in the research area of complex networks, which is also called the influence maximization problem in sociology domains. In order to evaluate the spreading ability of nodes in networks, many centrality-based indices have been proposed and achieved relatively go...

Multiplex networks are a special class of multilayered networks in which a fixed set of nodes is connected by different types of links. The core organization, the residual graph from recursively removing dead-end nodes and their nearest neighbor, plays a significant role in a wide range of typical problems. However previous study about core structu...

The study of identifying important nodes in networks has a wide application in different fields. However, the current researches are mostly based on static or aggregated networks. Recently, the increasing attention to networks with time-varying structure promotes the study of node centrality in temporal networks. In this paper, we define a supra-ev...

How to effectively identify a set of influential spreaders in complex networks is of great theoretical and practical value, which can help to inhibit the rapid spread of epidemics, promote the sales of products by word-of-mouth advertising, and so on. A naive strategy is to select the top ranked nodes as identified by some centrality indices, and o...

The topological structure of many real networks changes with time. Thus, locating the sources of a temporal network is a creative and challenging problem, as the enormous size of many real networks makes it unfeasible to observe the state of all nodes. In this paper, we propose an algorithm to solve this problem, named the backward temporal diffusi...

It is a crucial and fundamental issue to identify a small subset of influential spreaders that can control the spreading process in networks. In previous studies, a degree-based heuristic called DegreeDiscount has been shown to effectively identify multiple influential spreaders and has severed as a benchmark method. However, the basic assumption o...

Gnutella. Data of Gnutella peer-to-peer network. (RAR)

Cond-mat. Data of Cond-mat collaboration network. (RAR)

Epinions. Data of Epinions social network. (RAR)

Enron. Data of Enron email communication network. (RAR)

With the rapid development of social networks, how to effectively identify a small group of nodes to maximize their spreading influence becomes a crucial topic. Traditional centrality-based methods are often very simple but not so effective compared to other complex methods. In this paper, we propose a heuristic method to select spreaders sequentia...

Link prediction plays a significant role in explaining the evolution of networks. However it is still a challenging problem that has been addressed only with topological information in recent years. Based on the belief that network nodes with a great number of common neighbors are more likely to be connected, many similarity indices have achieved c...

Recent years, the studies of link prediction have been overwhelmingly emphasizing on undirected networks. Compared with it, how to identify missing and spurious interactions in directed networks has received less attention and still is not well understood. In this paper, we make use of classical link prediction indices for undirected networks, adap...

Link prediction in directed network is attracting growing interest among many network scientists. Compared with predicting the existence of a link, determining its direction is more complicated. In this paper, we propose an efficient solution named Local Directed Path to predict link direction. By adding an extra ground node to the network, we solv...

In practice, complex systems often change over time, and the temporal characteristics of a complex network make their behavior difficult to predict. Traditional link prediction methods based on structural similarity are good for mining underlying information from static networks, but do not always capture the temporal relevance of dynamic networks....

Mining how network evolves is a crucial topic in extracting underlying information from networks. Among all the mechanisms, Common Neighbor (CN) and Preferential Attachment (PA) are basic and efficient. Recently, a new framework named Local-Community-Paradigm (LCP) provides a self-organized mechanism about the evolution of networks. Via using commu...

Plenty of algorithms for link prediction have been proposed to extract missing information, identify spurious interactions, reconstruct networks, and so on. Stochastic block models are one of the most accurate methods among all of them. However, this algorithm is designed only for simple graphs and ignores the variation in node degree which is typi...

A kind of descriptor for depicting the Point Cloud Structure is Carlsson's 1D-Barcode. This paper puts forward a brand new Point Cloud Descriptor: 2D-Barcode, which promotes the Barcode's dimension from 1 to 2 and results in containing much more structure information and visible. By proposing a new method of constructing the relationships among the...

By making use of the principles of systems science, the scientific community can explain many complicated matters of the world and shed new light on unsettled problems. Each real science has its own particular methodology for not only qualitative but also quantitative analyses, so it is important to understand the organic whole of systems research...

Purpose The purpose of this paper is to research integrating web text resources and mine its emergence. Design/methodology/approach With the understanding of characteristics of internet resources, this paper will focus on solving the problem of text resource aggregation in open environment and its emergence showed during aggregation over time. The...

In this paper, we formally define and study the event graph model based on set theory and multi-relations theory, and discuss the methods of modeling event and event relations in detail. The event graph model is mainly designed to extract the potential events and the relationships between events from massive text streams, and further discover the t...

Massive text stream is a very important source of information. There are rich graph structures contained in massive texts, which can be used as an effective method to mine trends embodied in the contents in text streams. This paper formally defines the event graph model based on systems science theory, and discusses its properties. This model aims...

In this paper, we propose a general framework for content mining, which combines statistical model and network structure to leverages the power of both statistical topic models and network method. This method is a novel view to both text oriented method and network oriented method. The proposed framework is general, it can be applied to any text co...