arXiv subjects

Chuang Liu

Publications and source records attributed to Chuang Liu.

At least 73 records · Page 4Linked to original sources

arXiv2022

Fault Diagnosis for Power Electronics Converters based on Deep Feedforward Network and Wavelet Compression

A fault diagnosis method for power electronics converters based on deep feedforward network and wavelet compression is proposed in this paper. The transient historical data after wavelet compression are used to realize the training of fault diagnosis classifier. Firstly, the correlation analysis of the voltage or current data running in various fault states is performed to remove the redundant features and the sampling point. Secondly, the wavelet transform is used to remove the redundant data of the features, and then the training sample data is greatly compressed. The deep feedforward network is trained by the low frequency component of the features, while the training speed is greatly accelerated. The average accuracy of fault diagnosis classifier can reach over 97%. Finally, the fault diagnosis classifier is tested, and final diagnosis result is determined by multiple-groups transient data, by which the reliability of diagnosis results is improved. The experimental result proves that the classifier has strong generalization ability and can accurately locate the open-circuit faults in IGBTs.

eess.SP↗

arXiv2022

Routing and QoS Policy Optimization in SD-WAN

In modern SD-WAN networks, a global controller continuously optimizes application and user intents by selecting the proper routing policies for each application. Nevertheless, the competition between flows can still occur at each overlay link and it may degrade Quality of Service (QoS). To mitigate performance degradations in case of congestion, QoS policies can also be dynamically optimized to limit the rate of low-priority flows and share the bandwidth among all flows in a fair manner. This demonstration presents a comprehensive control plane architecture to coordinate path selection and rate allocation, in order to meet application requirements while optimizing a global objective (e.g., low congestion, best quality, and minimum cost).

cs.NI↗

arXiv2022

Review for AI-based Open-Circuit Faults Diagnosis Methods in Power Electronics Converters

Power electronics converters have been widely used in aerospace system, DC transmission, distributed energy, smart grid and so forth, and the reliability of power electronics converters has been a hotspot in academia and industry. It is of great significance to carry out power electronics converters open-circuit faults monitoring and intelligent fault diagnosis to avoid secondary faults, reduce time and cost of operation and maintenance, and improve the reliability of power electronics system. Firstly, the faults features of power electronic converters are analyzed and summarized. Secondly, some AI-based fault diagnosis methods and application examples in power electronics converters are reviewed, and a fault diagnosis method based on the combination of random forests and transient fault features is proposed for three-phase power electronics converters. Finally, the future research challenges and directions of AI-based fault diagnosis methods are pointed out.

eess.SY↗

arXiv2022

Vital node identification in hypergraphs via gravity model

Hypergraphs that can depict interactions beyond pairwise edges have emerged as an appropriate representation for modeling polyadic relations in complex systems. With the recent surge of interest in researching hypergraphs, the centrality problem has attracted abundant attention due to the challenge of how to utilize the higher-order structure for the definition of centrality metrics. In this paper, we propose a new centrality method (HGC) on the basis of the gravity model as well as a semi-local HGC (LHGC) which can achieve a balance between accuracy and computational complexity. Meanwhile, two comprehensive evaluation metrics, i.e., a complex contagion model in hypergraphs that mimics the group influence during the spreading process and network s-efficiency based on the higher-order distance between nodes, are first proposed to evaluate the effectiveness of our methods. The results show that our methods can filter out nodes that have fast spreading ability and are vital in terms of hypergraph connectivity.

physics.soc-ph↗

arXiv2022

Comprehensive Graph Gradual Pruning for Sparse Training in Graph Neural Networks

Graph Neural Networks (GNNs) tend to suffer from high computation costs due to the exponentially increasing scale of graph data and the number of model parameters, which restricts their utility in practical applications. To this end, some recent works focus on sparsifying GNNs with the lottery ticket hypothesis (LTH) to reduce inference costs while maintaining performance levels. However, the LTH-based methods suffer from two major drawbacks: 1) they require exhaustive and iterative training of dense models, resulting in an extremely large training computation cost, and 2) they only trim graph structures and model parameters but ignore the node feature dimension, where significant redundancy exists. To overcome the above limitations, we propose a comprehensive graph gradual pruning framework termed CGP. This is achieved by designing a during-training graph pruning paradigm to dynamically prune GNNs within one training process. Unlike LTH-based methods, the proposed CGP approach requires no re-training, which significantly reduces the computation costs. Furthermore, we design a co-sparsifying strategy to comprehensively trim all three core elements of GNNs: graph structures, node features, and model parameters. Meanwhile, aiming at refining the pruning operation, we introduce a regrowth process into our CGP framework, in order to re-establish the pruned but important connections. The proposed CGP is evaluated by using a node classification task across 6 GNN architectures, including shallow models (GCN and GAT), shallow-but-deep-propagation models (SGC and APPNP), and deep models (GCNII and ResGCN), on a total of 14 real-world graph datasets, including large-scale graph datasets from the challenging Open Graph Benchmark. Experiments reveal that our proposed strategy greatly improves both training and inference efficiency while matching or even exceeding the accuracy of existing methods.

cs.LG↗

arXiv2022

Using EBGAN for Anomaly Intrusion Detection

As an active network security protection scheme, intrusion detection system (IDS) undertakes the important responsibility of detecting network attacks in the form of malicious network traffic. Intrusion detection technology is an important part of IDS. At present, many scholars have carried out extensive research on intrusion detection technology. However, developing an efficient intrusion detection method for massive network traffic data is still difficult. Since Generative Adversarial Networks (GANs) have powerful modeling capabilities for complex high-dimensional data, they provide new ideas for addressing this problem. In this paper, we put forward an EBGAN-based intrusion detection method, IDS-EBGAN, that classifies network records as normal traffic or malicious traffic. The generator in IDS-EBGAN is responsible for converting the original malicious network traffic in the training set into adversarial malicious examples. This is because we want to use adversarial learning to improve the ability of discriminator to detect malicious traffic. At the same time, the discriminator adopts Autoencoder model. During testing, IDS-EBGAN uses reconstruction error of discriminator to classify traffic records.

cs.CR↗

arXiv2022

Influence Maximization in Hypergraphs

Influence maximization in complex networks, i.e., maximizing the size of influenced nodes via selecting K seed nodes for a given spreading process, has attracted great attention in recent years. However, the influence maximization problem in hypergraphs, in which the hyperedges are leveraged to represent the interactions among more than two nodes, is still an open question. In this paper, we propose an adaptive degree-based heuristic algorithm, i.e., Heuristic Degree Discount (HDD), which iteratively selects nodes with low influence overlap as seeds, to solve the influence maximization problem in hypergraphs. We further extend algorithms from ordinary networks as baselines and compare the performance of the proposed algorithm and baselines on both real data and synthetic hypergraphs. Results show that HDD outperforms the baselines in terms of both effectiveness and efficiency. Moreover, the experiments on synthetic hypergraphs indicate that HDD shows high performance, especially in hypergraphs with heterogeneous degree distribution.

cs.SI↗

arXiv2022

Multi-objective Pointer Network for Combinatorial Optimization

Multi-objective combinatorial optimization problems (MOCOPs), one type of complex optimization problems, widely exist in various real applications. Although meta-heuristics have been successfully applied to address MOCOPs, the calculation time is often much longer. Recently, a number of deep reinforcement learning (DRL) methods have been proposed to generate approximate optimal solutions to the combinatorial optimization problems. However, the existing studies on DRL have seldom focused on MOCOPs. This study proposes a single-model deep reinforcement learning framework, called multi-objective Pointer Network (MOPN), where the input structure of PN is effectively improved so that the single PN is capable of solving MOCOPs. In addition, two training strategies, based on representative model and transfer learning, respectively, are proposed to further enhance the performance of MOPN in different application scenarios. Moreover, compared to classical meta-heuristics, MOPN only consumes much less time on forward propagation to obtain the Pareto front. Meanwhile, MOPN is insensitive to problem scale, meaning that a trained MOPN is able to address MOCOPs with different scales. To verify the performance of MOPN, extensive experiments are conducted on three multi-objective traveling salesman problems, in comparison with one state-of-the-art model DRL-MOA and three classical multi-objective meta-heuristics. Experimental results demonstrate that the proposed model outperforms all the comparative methods with only 20\% to 40\% training time of DRL-MOA.

cs.LG↗

arXiv2022

Toward Structural Controllability and Predictability in Directed Networks

The lack of studying the complex organization of directed network usually limits to the understanding of underlying relationship between network structures and functions. Structural controllability and structural predictability, two seemingly unrelated subjects, are revealed in this paper to be both highly dependent on the critical links previously thought to only be able to influence the number of driver nodes in controllable directed networks. Here, we show that critical links can not only contribute to structural controllability, but they can also have a significant impact on the structural predictability of networks, suggesting the universal pattern of structural reciprocity in directed networks. In addition, it is shown that the fraction and location of critical links have a strong influence on the performance of prediction algorithms. Moreover, these empirical results are interpreted by introducing the link centrality based on corresponding line graphs. This work bridges the gap between the two independent research fields, and it provides indications of developing advanced control strategies and prediction algorithms from a microscopic perspective.

physics.soc-ph↗

arXiv2021

Subtask-dominated Transfer Learning for Long-tail Person Search

Person search unifies person detection and person re-identification (Re-ID) to locate query persons from the panoramic gallery images. One major challenge comes from the imbalanced long-tail person identity distributions, which prevents the one-step person search model from learning discriminative person features for the final re-identification. However, it is under-explored how to solve the heavy imbalanced identity distributions for the one-step person search. Techniques designed for the long-tail classification task, for example, image-level re-sampling strategies, are hard to be effectively applied to the one-step person search which jointly solves person detection and Re-ID subtasks with a detection-based multi-task framework. To tackle this problem, we propose a Subtask-dominated Transfer Learning (STL) method. The STL method solves the long-tail problem in the pretraining stage of the dominated Re-ID subtask and improves the one-step person search by transfer learning of the pretrained model. We further design a Multi-level RoI Fusion Pooling layer to enhance the discrimination ability of person features for the one-step person search. Extensive experiments on CUHK-SYSU and PRW datasets demonstrate the superiority and effectiveness of the proposed method.

cs.CV↗

arXiv2021

Quantification of network structural dissimilarities based on graph embedding

Identifying and quantifying structural dissimilarities between complex networks is a fundamental and challenging problem in network science. Previous network comparison methods are based on the structural features, such as the length of shortest path, degree and graphlet, which may only contain part of the topological information. Therefore, we propose an efficient network comparison method based on network embedding, i.e., \textit{DeepWalk}, which considers the global structural information. In detail, we calculate the distance between nodes through the vector extracted by \textit{DeepWalk} and quantify the network dissimilarity by spectral entropy based Jensen-Shannon divergences of the distribution of the node distances. Experiments on both synthetic and empirical data show that our method outperforms the baseline methods and can distinguish networks perfectly by only using the global embedding based distance distribution. In addition, we show that our method can capture network properties, e.g., average shortest path length and link density. Moreover, the experiments of modularity further implies the functionality of our method.

cs.SI↗

arXiv2021

Making Person Search Enjoy the Merits of Person Re-identification

Person search is an extended task of person re-identification (Re-ID). However, most existing one-step person search works have not studied how to employ existing advanced Re-ID models to boost the one-step person search performance due to the integration of person detection and Re-ID. To address this issue, we propose a faster and stronger one-step person search framework, the Teacher-guided Disentangling Networks (TDN), to make the one-step person search enjoy the merits of the existing Re-ID researches. The proposed TDN can significantly boost the person search performance by transferring the advanced person Re-ID knowledge to the person search model. In the proposed TDN, for better knowledge transfer from the Re-ID teacher model to the one-step person search model, we design a strong one-step person search base framework by partially disentangling the two subtasks. Besides, we propose a Knowledge Transfer Bridge module to bridge the scale gap caused by different input formats between the Re-ID model and one-step person search model. During testing, we further propose the Ranking with Context Persons strategy to exploit the context information in panoramic images for better retrieval. Experiments on two public person search datasets demonstrate the favorable performance of the proposed method.

cs.CV↗

arXiv2021

Measuring and utilizing temporal network dissimilarity

Quantifying the structural and functional differences of temporal networks is a fundamental and challenging problem in the era of big data. This work proposes a temporal dissimilarity measure for temporal network comparison based on the fastest arrival distance distribution and spectral entropy based Jensen-Shannon divergence. Experimental results on both synthetic and empirical temporal networks show that the proposed measure could discriminate diverse temporal networks with different structures by capturing various topological and temporal properties. Moreover, the proposed measure can discern the functional distinctions and is found effective applications in temporal network classification and spreadability discrimination.

cs.SI↗

arXiv2021

The Aging Effect in Evolving Scientific Citation Networks

The study of citation networks is of interest to the scientific community. However, the underlying mechanism driving individual citation behavior remains imperfectly understood, despite the recent proliferation of quantitative research methods. Traditional network models normally use graph theory to consider articles as nodes and citations as pairwise relationships between them. In this paper, we propose an alternative evolutionary model based on hypergraph theory in which one hyperedge can have an arbitrary number of nodes, combined with an aging effect to reflect the temporal dynamics of scientific citation behavior. Both theoretical approximate solution and simulation analysis of the model are developed and validated using two benchmark datasets from different disciplines, i.e. publications of the American Physical Society (APS) and the Digital Bibliography & Library Project (DBLP). Further analysis indicates that the attraction of early publications will decay exponentially. Moreover, the experimental results show that the aging effect indeed has a significant influence on the description of collective citation patterns. Shedding light on the complex dynamics driving these mechanisms facilitates the understanding of the laws governing scientific evolution and the quantitative evaluation of scientific outputs.

physics.soc-ph↗

arXiv2021

Effective Model Integration Algorithm for Improving Link and Sign Prediction in Complex Networks

Link and sign prediction in complex networks bring great help to decision-making and recommender systems, such as in predicting potential relationships or relative status levels. Many previous studies focused on designing the special algorithms to perform either link prediction or sign prediction. In this work, we propose an effective model integration algorithm consisting of network embedding, network feature engineering, and an integrated classifier, which can perform the link and sign prediction in the same framework. Network embedding can accurately represent the characteristics of topological structures and cooperate with the powerful network feature engineering and integrated classifier can achieve better prediction. Experiments on several datasets show that the proposed model can achieve state-of-the-art or competitive performance for both link and sign prediction in spite of its generality. Interestingly, we find that using only very low network embedding dimension can generate high prediction performance, which can significantly reduce the computational overhead during training and prediction. This study offers a powerful methodology for multi-task prediction in complex networks.

physics.soc-ph↗

arXiv2021

Exploring the effect of social media and spatial characteristics during the COVID-19 pandemic in China

The declaration of COVID-19 as a pandemic has largely amplified the spread of related information on social media, such as Twitter, Facebook, and WeChat.Unlike the previous studies which focused on how to detect the misinformation or fake news related toCOVID-19, we investigate how the disease and information co-evolve in the population. We focus onCOVID-19and its information during the period when the disease was widely spread in China, i.e., from January 25th to March 24th, 2020. We first explore how the disease and information co-evolve via the spatial analysis of the two spreading processes. We visualize the geo-location of both disease and information at the province level and find that disease is more geo-localized compared to information. We find a high correlation between the disease and information data, and also people care about the spread only when it comes to their neighborhood. Regard to the content of the information, we find that positive messages are more negatively correlated with the disease compared to negative and neutral messages. Additionally, we introduce machine learning algorithms, i.e., linear regression and random forest, to further predict the number of infected using different disease spatial related and information-related characteristics. We obtain that the disease spatial related characteristics of nearby cities can help to improve the prediction accuracy. Meanwhile, information-related characteristics can also help to improve the prediction performance, but with a delay, i.e., the improvement comes from using, for instance, the number of messages 10 days ago, for disease prediction. The methodology proposed in this paper may shed light on new clues of emerging infections

physics.soc-ph↗

arXiv2020

Information Spreading Dynamics on Adaptive Social Networks

There is currently growing interest in modeling the information diffusion on social networks across multi-disciplines. The majority of the corresponding research has focused on information diffusion independently, ignoring the network evolution in the diffusion process. Therefore, it is more reasonable to describe the real diffusion systems by the co-evolution between network topologies and information states. In this work, we propose a mechanism considering the coevolution between information states and network topology simultaneously, in which the information diffusion was executed as an SIS process and network topology evolved based on the adaptive assumption. The theoretical analyses based on the Markov approach were very consistent with simulation. Both simulation results and theoretical analyses indicated that the adaptive process, in which informed individuals would rewire the links between the informed neighbors to a random non-neighbor node, can enhance information diffusion (leading to much broader spreading). In addition, we obtained that two threshold values exist for the information diffusion on adaptive networks, i.e., if the information propagation probability is less than the first threshold, information cannot diffuse and dies out immediately; if the propagation probability is between the first and second threshold, information will spread to a finite range and die out gradually; and if the propagation probability is larger than the second threshold, information will diffuse to a certain size of population in the network. These results may shed some light on understanding the co-evolution between information diffusion and network topology.

physics.soc-ph↗

arXiv2016

Identifying the Academic Rising Stars

Predicting the fast-rising young researchers (Academic Rising Stars) in the future provides useful guidance to the research community, e.g., offering competitive candidates to university for young faculty hiring as they are expected to have success academic careers. In this work, given a set of young researchers who have published the first first-author paper recently, we solve the problem of how to effectively predict the top k% researchers who achieve the highest citation increment in Δt years. We explore a series of factors that can drive an author to be fast-rising and design a novel impact increment ranking learning (IIRL) algorithm that leverages those factors to predict the academic rising stars. Experimental results on the large ArnetMiner dataset with over 1.7 million authors demonstrate the effectiveness of IIRL. Specifically, it outperforms all given benchmark methods, with over 8% average improvement. Further analysis demonstrates that the prediction models for different research topics follow the similar pattern. We also find that temporal features are the best indicators for rising stars prediction, while venue features are less relevant.

cs.DL↗