- Network Security and Intrusion Detection
- Data Management and Algorithms
- Data Mining Algorithms and Applications
- Software-Defined Networks and 5G
- Anomaly Detection Techniques and Applications
- Adversarial Robustness in Machine Learning
- Rough Sets and Fuzzy Logic
- Advanced Malware Detection Techniques
- Cloud Computing and Resource Management
- IoT and Edge/Fog Computing
- Software System Performance and Reliability
- Internet Traffic Analysis and Secure E-voting
- Privacy, Security, and Data Protection
- Image and Video Quality Assessment
- Cold Fusion and Nuclear Reactions
- Impact of AI and Big Data on Business and Society
- Planetary Science and Exploration
- Insurance and Financial Risk Management
- Data-Driven Disease Surveillance
- Indoor and Outdoor Localization Technologies
- Distributed and Parallel Computing Systems
- Magnetic Properties of Alloys
- Privacy-Preserving Technologies in Data
- Digital Rights Management and Security
- Advanced Database Systems and Queries
Nankai University
2023-2024
Tianjin University
2005-2023
Dalian Maritime University
2016-2017
A new broad of video services that support live streaming has become tremendously popular in recent years. Compared with traditional video-on-demand (VOD) services, much higher requirements on Quality-of-Experience (QoE), including low rebuffering, high definition, latency and bitrate oscillations. While previous adaptive algorithms (ABR) solely optimize for ensuring QoE VOD, a larger decision space, making the optimization problem more difficult to solve. We propose Deeplive, which...
Event detection by discovering frequent itemsets is very popular in sensor network communities. However, the recorded data often a probability rather than determined value really productive environment as sensed affected noise. In this paper, we study to detect events finding patterns over probabilistic under Possible World Semantics. This technically challenging records can generate an exponential number of possible worlds. Although several efficient algorithms are proposed literature, it...
Detecting anomaly in images is challenging due to the high dimension nature of image data. While previous learning-based detection approaches can detect a particular type precisely, they often fail detecting multiple types abnormal samples simultaneously.We identify two specific anomalies that be precisely detected by either compress-based or reconstruction-based approaches, named global and local anomaly. We then propose Glad, an detector both them at same time. Glad adopts joint approach...
Domain generation algorithms (DGAs) can be categorized into three types: <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">zero-knowledge</i> , xmlns:xlink="http://www.w3.org/1999/xlink">partial-knowledge</i> and xmlns:xlink="http://www.w3.org/1999/xlink">full-knowledge</i> . While prior research merely focused on types, we characterize their anti-detection ability practicality find that DGAs present low against...
Security vendors can take down botnets by detecting the malicious domain names crafted attackers. However, adversarial <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Domain Generation Algorithms</i> ( xmlns:xlink="http://www.w3.org/1999/xlink">DGAs</i> ) greatly challenge existing detection schemes, in particular, DGAs actively compromise arbitrarily specified systems crafting names. To resist DGAs, we propose a game theory-based defending...
Nowadays, various algorithms are widely used in the field of economy and trade, economic trade management laws also need to introduce scientific effective data models for optimization. In this paper, support vector machine algorithm logistic regression analyze process actual case bank loan user data, a hybrid model is established. This study first introduces basic definitions contents algorithm, then constructs by randomly dividing using calculate results, inputting them into algorithm. The...
Frequent itemsets discovery is popular in database communities recently. Because real data often affected by noise, this paper, we study to find frequent over probabilistic under the Possible World Semantics. It challenging because there may be exponential number of possible worlds for database. Although several efficient algorithms are proposed literature, it hard mine large uncertain due high time consuming. To address issue, propose an algorithm itemsets. A pruning strategy also presented...
To reduce the cost of online cloud services, service providers often employ elastic approach that allows tenants to "scale out" or up" their applications at runtime. However, traditional virtual machine-based cannot meet fine-grained fluctuating demand due its slow startup time and high reconfiguration cost. address this challenge, we present ElaX, an manager minimizes resource provisioning for containerized services while guaranteeing tail latency requirement. ElaX designs a workload-aware...
Accurate and quick failure localization is critical for automatic network troubleshooting. While it particularly difficult to solve the problem in traditional due uncertain routing, Software Defined Networking (SDN) enables deterministic routing packet transmission through traffic engineering algorithm centralized controller.
Detecting abnormal service performance is significant for Internet-based management and operation. Recent advances in anomaly detection methods prefer unsupervised learning algorithms since they can work without manually labelled data. However, existing converge into suboptimal solutions due to their heuristic-based objectives. Moreover, frequently rely on the strong assumption that noise follows a Gaussian distribution, accuracy also highly sensitive threshold settings. To detect anomalies...
Research on underwater sensor networks is growing more and important in marine areas, among which localization has recently gained widely research attention. However, the existing algorithms still can not achieve both reliability efficiency three dimensional space. To address this issue, we present a novel network approach based distance transform-based skeleton extraction. In our design, success rate of keeps steady high most cases. At same time, extracted robust to boundary noise....
Embedding-as-a-Service (EaaS) has emerged as a successful business pattern but faces significant challenges related to various forms of copyright infringement, including API misuse and different attacks. Various studies have proposed backdoor-based watermarking schemes protect the EaaS services. In this paper, we reveal that previous possess semantic-independent characteristics propose Semantic Perturbation Attack (SPA). Our theoretical experimental analyses demonstrate nature makes current...
In modern data center networks (DCNs), failures of network devices always occur and it is difficult to localize these failures. Our key observation that latency can reflect profile status. We use this information resolve issues like failure localization. paper, we present NetCruiser, a system able by learning from data. It both measure collect monitor the status whole pinpoint which switch or router encounters failure. And design structure handle With construction structure, build machine...
Association rules and frequent patterns discovery is always a hot topic in database communities. As real data often affected by noise, this paper, we study to find generate association over probabilistic under the Possible World Semantics. This technically challenging, since can have an exponential number of possible worlds. Although several efficient algorithms are proposed literature, there still large space for improvement due redundancy property data. To address issue, employ approximate...
With the invention of big data era, releasing is becoming a hot topic in database community. Meanwhile, privacy also raises attention users. As far as protection models that have been proposed, differential model widely utilized because its many advantages over other models. However, for private multi-dimensional sets, existing algorithms are publishing usually with low availability. The reason noise released rapidly grown increasing dimensions. In view this issue, we propose based on...
Domain generation algorithms (DGAs) can be categorized into three types: zero-knowledge, partial-knowledge, and full-knowledge. While prior research merely focused on zero-knowledge full-knowledge types, we characterize their anti-detection ability practicality find that DGAs present low against detectors, suffer from due to the strong assumption they are fully detector-aware. Given these observations, propose PKDGA, a partial knowledge-based domain algorithm with high practicality. PKDGA...
Using the part exchange-coupling model, interaction between magnetically soft and hard grains in Nd2Fe14B/α-Fe nanocomposite its effect on effective anisotropy of material have been investigated. By testing six different functions, series curves constant versus mean grain size were obtained. The calculation results show that composite increases slightly with decreasing size, reaches maximum at about 23nm, then decreases abruptly. calculated resemble experimental coercivity nanocomposite.
This paper describes a micro-target semiautomatic assembly system used in ICF. In the two halves of cylindrical gold hohlraum (1.6 mm long by 0.8 diameter) and microsphere capsule (about 0.2 are assembled. CCD cameras applied to monitor process closed loop assembly.