A5024464918- Network Security and Intrusion Detection
- Anomaly Detection Techniques and Applications
- Software System Performance and Reliability
- Internet Traffic Analysis and Secure E-voting
- Traffic Prediction and Management Techniques
- Transportation Planning and Optimization
- Mobile Crowdsensing and Crowdsourcing
- Human Mobility and Location-Based Analysis
- Advanced Malware Detection Techniques
- Indoor and Outdoor Localization Technologies
- Caching and Content Delivery
- Data Visualization and Analytics
- Privacy-Preserving Technologies in Data
- Cloud Computing and Resource Management
- Complex Network Analysis Techniques
- Blockchain Technology Applications and Security
- Power System Reliability and Maintenance
- Time Series Analysis and Forecasting
- Cryptography and Data Security
- Computational Physics and Python Applications
- Software-Defined Networks and 5G
- Sparse and Compressive Sensing Techniques
- Age of Information Optimization
- Software Engineering Research
- Big Data and Digital Economy
Computer Network Information Center
2024
Chinese Academy of Sciences
2020-2024
Institute of Computing Technology
2020-2024
Purple Mountain Laboratories
2020-2023
University of Chinese Academy of Sciences
2021
Hunan University
2018-2019
State University of New York
2019
Graph convolutional neural networks (GCNN) have become an increasingly active field of research. It models the spatial dependencies nodes in a graph with pre-defined Laplacian matrix based on node distances. However, many application scenarios, change over time, and use fixed cannot capture change. To track among traffic data, we propose dynamic spatio-temporal GCNN for accurate forecasting. The core our deep learning framework is finding estimator. enable timely low complexity, creatively...
The prediction of short-term volatile traffic becomes increasingly critical for efficient engineering in intelligent transportation systems. Accurate forecast results can assist management and pedestrian route selection, which will help alleviate the huge congestion problem system. This paper presents a novel hybrid DTMGP model to accurately volume passenger flows multi-step ahead with comprehensive consideration factors from temporal, origin-destination spatial, frequency self-similarity...
The attacks, faults, and severe communication/system conditions in Mobile Crowd Sensing (MCS) make false data detection a critical problem. Observing the intrinsic low dimensionality of general monitoring sparsity data, can be performed based on separation normal anomalies. Although existing algorithm Direct Robust Matrix Factorization (DRMF) is proven to effective, requiring iteratively performing Singular Value Decomposition (SVD) for low-rank matrix approximation would result...
Traffic sensor networks are widely applied in smart cities to monitor traffic real-time and record huge volumes of data. Exploiting such data forecast future conditions have the potential enhance decision-making capabilities intelligent transportation systems, which attracts widespread attention from both industries academia. Among them, network-wide prediction based on graph convolutional neural networks(GCN) has become mainstream. It models spatial dependencies sensors a with pre-defined...
With the proliferation of smartphones, a novel sensing paradigm called Mobile Crowd Sensing (MCS) has emerged very recently. However, attacks and faults in MCS cause serious false data problem. Observing intrinsic low dimensionality general monitoring sparsity data, detection can be performed based on separation normal anomalies. Although existing algorithm Direct Robust Matrix Factorization (DRMF) is proven to effective, requiring iteratively performing Singular Value Decomposition (SVD)...
Massive key performance indicators (KPIs) are monitored as multivariate time series data (MTS) to ensure the reliability of software applications and service system. Accurately detecting abnormality MTS is very critical for subsequent fault elimination. The scarcity anomalies manual labeling has led development various self-supervised anomaly detection (AD) methods, which optimize an overall objective/loss encompassing all metrics' regression objectives/losses. However, our empirical study...
A large volume of logs provides a reliable data source for online services. For instance, 5G, Big 5G or Beyong cloud core network and its User Plane Functions require analysis built on massive log storage. The storage cost remains problem the industry. Compressing before archives most popular way to reduce it. However, structured logs, which usually present in tabular format, unstructured build variable templates, have use different compression strategies. Using parsers can eliminate this...
Power system is becoming larger and more connectible, closely relating to human life social development. However, with the increasing data volume complexity, danger of frequency collapse has always exist. To tackle this challenge, paper proposes a parallel maximum flow based complex network approach identify critical lines. First, power modeled as graph edges (transmission lines, transformers, etc.) nodes (buses, substations, etc.)based on theory. Then an improved proposed for topology...
Similarity (distance) measurement among network features (e.g. IP address, MAC port number, and protocol, etc.) based on logs is a critical step for data mining in intrusion detection, anomaly prediction, log analysis. A practical approach necessarily accurate, fast, incremental due to the dynamic environment. However, existing solutions fail satisfy these demands simultaneously. Therefore, we propose novel unsupervised feature embedding approach: Network Vector (NeVe). It learns similarity...
By deploying virtualized network elements (hosts, switches, routers, links, etc.) on clusters of commodity machines, distributed emulations (DNE) closely mimic the behaviors systems and provide real-time interactions analysis for service management. However, DNE encounters scalability challenges when faced with large topologies. These can be boiled down to assignment problem: which physical machine each element should assigned so that largest possible topology emulated? In this paper, we...
Massive key performance indicators (KPIs) are monitored as multivariate time series data (MTS) to ensure the reliability of software applications and service system. Accurately detecting abnormality MTS is very critical for subsequent fault elimination. The scarcity anomalies manual labeling has led development various self-supervised anomaly detection (AD) methods, which optimize an overall objective/loss encompassing all metrics' regression objectives/losses. However, our empirical study...
Time series analysis can explore the internal relations among data, so as to further realize prediction, pattern discovery and anomaly detection. However, these analyses largely rely on labeled which needs identify particular patterns from entire time data. Because labeling is costly domain specific, how automatically label raw data with predefined becomes a valuable problem. In this paper, we first discuss particularities that learned practical applications of task in terms inconsistent...