Zheng Liu

Nanjing University of Posts and Telecommunications · School of Computer Science

Multiplex network embedding is an effective technique to jointly learn the low-dimensional representations of nodes across network layers. However, the number of edges among layers may vary significantly. This data imbalance will lead to performance degradation especially on the sparse layer due to learning bias and the adverse effects of irrelevan...

Researchers are solving the challenges of spatial-temporal prediction by combining Federated Learning (FL) and graph models with respect to the constrain of privacy and security. In order to make better use of the power of graph model, some researchs also combine split learning (SL). However, there are still several issues left unattended: 1) Clien...

Directed graph is able to model asymmetric relationships between nodes and research on directed graph embedding is of great significance in downstream graph analysis and inference. Learning source and target embeddings of nodes separately to preserve edge asymmetry has become the dominant approach, but also poses challenge for learning representati...

Virtualization technologies provide solutions for cloud computing. Virtual resource scheduling is a crucial task in data centers, and the power consumption of virtual resources is a critical foundation of virtualization scheduling. Containers are the smallest unit of virtual resource scheduling and migration. Although many practical models for esti...

Multiplex network embedding is an effective technique to jointly learn the low-dimensional representations of nodes across network layers. However, the number of edges among layers may vary significantly. This data imbalance will lead to performance degradation especially on the sparse layer due to learning bias and the adverse effects of irrelevan...

Directed graphs model asymmetric relationships between nodes and research on directed graph embedding is of great significance in downstream graph analysis and inference. Learning source and target embedding of nodes separately to preserve edge asymmetry has become the dominant approach, but also poses challenge for learning representations of low...

Researchers are solving the challenges of spatial-temporal prediction by combining Federated Learning (FL) and graph models with respect to the constrain of privacy and security. However, there are still several issues left unattended: 1) Clients might not be able to access the server during inference phase; 2) The graph of clients designed manuall...

The power forecasting has a guiding effect on power-aware scheduling strategies to reduce unnecessary power consumption in data centers. Many metrics related to power consumption can be collected in physical servers, such as the status of CPU, memory, and other components. However, most existing methods empirically exploit a small number of metrics...

Airdrop is a crucial concept in tokenomics. Startups of decentralized applications (DApps) reward early supporters by airdropping newly issued tokens up to a certain amount as a free giveaway. This naturally induces greedy hackers, called Sybils, to create multiple accounts for more shares. Most airdrops have prerequisites for qualification, in whi...

Patents bring technology companies commercial values in modern business operations. However, companies have to bear the high cost of handling patent applications or infringement cases. A common yet expensive task among these jobs is to analyze relevant patent literature. Lengthy and technically complicated patents require a large number of human ef...

Due to the rapid increase in the number and scale of data centers, the information and communication technology (ICT) equipment in data centers consumes an enormous amount of power. A power prediction model is therefore essential for decision‐making optimization and power management of ICT equipment. However, it is difficult to predict the power co...

Graph convolutional networks (GCNs) have recently drawn extensive attention due to their superior learning performance on graph data. Through graph convolution, topological structure and node attributes can be simultaneously aggregated in a local neighborhood. In heterogeneous information networks (HINs), the diversity of node and edge types poses...

Many applications in financial investment management, environmental pollution reduction, energy resource scheduling, and consumer sales promotion need to forecast future values in multivariate tie series. Despite extensive research efforts laid on the task for decades from the perspective of prediction models, this paper explores boost the predicti...

Forecasting future values is a core task in many applications dealing with multivariate time series data. In pollution monitoring, for example, forecasting future PM\(_{\boldsymbol{2.5}}\) values in air is very common, which is a crucial indicator of the air quality index (AQI). These values in time series are sometimes affiliated with category inf...

For time series forecasting, the weight distribution among multivariables and the long-short-term time dependence are always very important and challenging. Traditional machine forecasting can’t automatically select the effective features of multivariable input and can’t capture the time dependence of sequences. The key to solve this problem is to...

Deriving a successful document representation is the critical challenge in many downstream tasks in NLP, especially when documents are very short. It is challenging to handle the sparsity and the noise problems confronting short texts. Some approaches employ latent topic models, based on global word co-occurrence, to obtain topic distribution as th...

Modern cloud computing relies heavily on data centers, which usually host tens of thousands of servers. Predicting the power consumption accurately in data center operations is crucial for energy optimization. In this paper, we formulate the power consumption prediction at both the fine-grained and coarse-grained level. We carefully discuss the des...

Recently, the Graph Convolutional Networks (GCNs) have achieved state-of-the-art performance in many graph data related tasks. However, traditional GCNs may generate redundant information in the message passing phase. In order to solve this problem, we propose a novel graph convolution named Push-and-Pull Convolution (PPC), which follows the messag...

Anomaly analysis plays a significant role in building a secure and reliable system. Raw system logs contain important system information, such as execution paths and execution time. People often use system logs for fault diagnosis and root cause localization. However, due to the complexity of raw system logs, these tasks can be arduous and ineffect...

Named Entity Recognition (NER) is a basic task in Natural Language Processing (NLP). Recently, the sequence-to-sequence (seq2seq) model has been widely used in NLP task. Different from the general NLP task, 60% sentences in the NER task do not contain entities. Traditional seq2seq method cannot address this issue effectively. To solve the aforement...

Log data is typically the only available data source recording system health information. Event extraction converts unstructured log messages into structured event signatures. Existing methods, whether batch or streaming methods, require true event signatures to guide parameter selection. This paper presents a streaming event extraction method that...

Understanding contents in social networks by inferring high-quality latent topics from short texts is a significant task in social analysis, which is challenging because social network contents are usually extremely short, noisy and full of informal vocabularies. Due to the lack of sufficient word co-occurrence instances, well-known topic modeling...

Patent retrieval primarily focuses on searching relevant legal documents with respect to a given query. Depending on the purposes of specific retrieval tasks, processes of patent retrieval may differ significantly. Given a patent application, it is challenging to determine its patentability, i.e., to decide whether a similar invention has been publ...

Extracting keyphrases from documents helps to reduce the document information and further assist in information retrieval. In this paper, we construct a multi-relational graph by considering heterogeneous latent word relations (the co-occurrence and the semantic) in a document. Then we optimize the random walks on the multi-relational graph to dete...

Extracting keyphrases from documents for providing a quick and insightful summarization is an interesting and important task, on which lots of research efforts have been laid. Most of the existing methods could be categorized as co-occurrence based, statistic-based, or semantics-based. The co-occurrence based methods do not take various word relati...

In large scale and complex IT service environments, a problematic incident is logged as a ticket and contains the ticket summary (sys-tem status and problem description). The system administrators log the step-wise resolution description when such tickets are resolved. The repeating service events are most likely resolved by inferring similar histo...

Many systems, such as distributed operating systems, complex networks, and high throughput web-based applications, are continuously generating large volume of event logs. These logs contain useful information to help system administrators to understand the system running status and to pinpoint the system failures. Generally, due to the scale and co...

The advent of Big Data era drives data analysts from different domains to use data mining techniques for data analysis. However, performing data analysis in a specific domain is not trivial; it often requires complex task configuration, onerous integration of algorithms, and efficient execution in distributed environments. Few efforts have been pai...

Modern forms of computing systems are becoming progressively more complex, with an increasing number of heterogeneous hardware and software components. As a result, it is quite challenging to manage these complex systems and meet the requirements in manageability, dependability, and performance that are demanded by enterprise customers. This survey...

Structural relationships between objects are used to model as graphs in many applications. In this paper, we study the problem of identifying relevant subgraphs in large networks. Relevant subgraphs in large networks contain network elements which are maintained by network administrators. We formalize the problem and propose a framework consisting...

Frequent subgraph mining has been an important research problem in the literature. However, the huge number of discovered frequent subgraphs becomes the bottleneck for exploring and understanding the generated patterns. In this paper, we propose to summarize frequent subgraphs with an independence probabilistic model, with the goal to restore the f...

We study a problem of detecting priming events based on a time series index and an evolving document stream. We define a priming event as an event which triggers abnormal movements of the time series index, i.e., the Iraq war with respect to the president approval index of President Bush. Existing solutions either focus on organizing coherent keywo...

Graph patterns are able to represent the complex structural relations among objects in many applications in various domains. The objective of graph summarization is to obtain a concise representation of a single large graph, which is interpretable and suitable for analysis. A good summary can reveal the hidden relationships between nodes in a graph...

Graph summarization is to obtain a concise representation of a large graph, which is suitable for visualization and analysis. The main idea is to construct a super-graph by grouping similar nodes together. In this paper, we propose a new information-preserving approach for graph summarization, which consists of two parts: a super-graph and a list o...

Discovery of evolving regions in large graphs is an important issue because it is the basis of many applications such as spam websites detection in the Web, community lifecycle exploration in social networks, and so forth. In this paper, we aim to study a new problem, which explores the evolution process between two historic snapshots of an evolvin...

Evolving graphs are used to model the relationship variations between objects in many application domains such as social networks, sensor networks, and telecommunication. In this paper, we study a new problem of discovering burst areas that exhibit dramatic changes during some periods in evolving graphs. We focus on finding the top-k results in a s...

Graphs are popularly used to model structural relation- ships between objects. In many application domains such as social networks, sensor networks and telecommunication, graphs evolve over time. In this paper, we study a new prob- lem of discovering the subgraphs that exhibit significant changes in evolving graphs. This problem is challenging sinc...

In many real world applications, decisions are usually made by collecting and judging information from multiple different data sources. Let us take the stock market as an example. We never make our decision based on just one single piece of advice, but always rely on a collection of information, such as the stock price movements, exchange volumes,...

In this paper, we propose a new model, called co-movement model, for constructing financial portfolios by analyzing and mining the co-movement patterns among multiple time series. Unlike the existing approaches where the portfolios’ expected risks are computed based on the co-variances among the assets in the portfolios, we model their risks by con...

We study the problem of detecting the shape anomalies in this paper. Our shape anomaly detection algorithm is performed on the one-dimensional representation (time series) of shapes, whose similarity is modeled by a generalized segmental hidden Markov model (HMM) under a scaling, translation and rotation invariant manner. Experimental results show...

Driven by many real applications, in this paper we study the problem of similarity search with implicit object features; that is, the features of each object are not pre-computed/evaluated. As the ex- isting similarity search techniques are not applicable, a novel and ecient algorithm is developed in this paper to approach the problem. The R-tree b...

Finding motifs in time-series is proposed to make clustering of time-series subsequences meaningful, because most existing algorithms of clustering time-series subsequences are reported meaningless in recent studies. The existing motif finding algorithms emphasize the efficiency at the expense of quality, in terms of the number of time-series subse...