A5041563693- Cloud Data Security Solutions
- Cryptography and Data Security
- Privacy-Preserving Technologies in Data
- Topic Modeling
- Recommender Systems and Techniques
- Cloud Computing and Resource Management
- Software System Performance and Reliability
- Advanced Graph Neural Networks
- Natural Language Processing Techniques
- Security and Verification in Computing
- Network Security and Intrusion Detection
- IoT and Edge/Fog Computing
- Context-Aware Activity Recognition Systems
- Speech and dialogue systems
- Service-Oriented Architecture and Web Services
- Software Engineering Research
- Caching and Content Delivery
- Anomaly Detection Techniques and Applications
- Image Retrieval and Classification Techniques
- Advanced Data Storage Technologies
- Advanced Malware Detection Techniques
- Software Testing and Debugging Techniques
- Access Control and Trust
- Complexity and Algorithms in Graphs
- Face and Expression Recognition
Peking University
2016-2025
Software (Spain)
2013-2025
Institute of Software
2022-2024
Chinese Academy of Geological Sciences
2012-2023
Tencent (China)
2023
Ministry of Natural Resources
2023
Tsinghua University
2019
Lenovo (China)
2014
Jiangsu Normal University
2008
Geomechanica (Canada)
2004
Deep neural networks (DNNs) have progressed rapidly during the past decade and been deployed in various real-world applications. Meanwhile, DNN models shown to be vulnerable security privacy attacks. One such attack that has attracted a great deal of attention recently is backdoor attack. Specifically, adversary poisons target model's training set mislead any input with an added secret trigger class. Previous attacks predominantly focus on computer vision (CV) applications, as image...
Generative dialogue systems tend to produce generic responses, which often leads boring conversations. For alleviating this issue, Recent studies proposed retrieve and introduce knowledge facts from graphs. While paradigm works a certain extent, it usually retrieves only based on the entity word itself, without considering specific context. Thus, introduction of context-irrelevant can impact quality generations. To end, paper proposes novel commonsense knowledge-aware generation model,...
Entity alignment which aims at linking entities with the same meaning from different knowledge graphs (KGs) is a vital step for fusion. Existing research focused on learning embeddings of by utilizing structural information KGs entity alignment. These methods can aggregate neighboring nodes but may also bring noise neighbors. Most recently, several researchers attempted to compare in pairs enhance However, they ignored relations between are important neighborhood matching. In addition,...
Recommendation based on heterogeneous information network(HIN) is attracting more and attention due to its ability emulate collaborative filtering, content-based context-aware recommendation combinations of any these semantics. Random walk methods are usually used mine the paths, weigh compute closeness or relevance between two nodes in a HIN. A key for success how properly set weights links In existing methods, mostly heuristically. this paper, we propose Bayesian Personalized Ranking(BPR)...
Social recommendation, which leverages social connections to construct Recommender Systems (RS), plays an important role in alleviating information overload. Recently, Graph Neural Networks (GNNs) have received increasing attention due their great capacity for graph data. Since data RS is essentially the structure of graphs, GNN-based flourishing. However, existing works lack in-depth thinking recommendations. These methods contain implicit assumptions that are not well analyzed practice. To...
Most of heterogeneous information network (HIN) based recommendation models are on the user and item modeling with meta-paths. However, they always model users items in isolation under each meta-path, which may lead to extraction misled. In addition, only consider structural features HINs when during exploring HINs, useful for lost irreversibly. To address these problems, we propose a HIN unified embedding recommendation, called HueRec. We assume there exist some common characteristics...
Insufficient semantic understanding of dialogue always leads to the appearance generic responses, in generative systems. Recently, high-quality knowledge bases have been introduced enhance understanding, as well reduce prevalence boring responses. Although such knowledge-aware approaches shown tremendous potential, they utilize a black-box fashion. As result, generation process is somewhat uncontrollable, and it also not interpretable. In this paper, we introduce topic fact-based commonsense...
The identification of stressfulness under certain driving condition is an important issue for safety, security and health. Sensors systems have been placed or implemented as wearable devices drivers. Features are extracted from the data collected combined to predict symptoms. challenge select feature set most relevant stress. In this paper, we propose a selection method based on performance diversity between two features. sets selected then using combinatorial fusion. We also compare our...
In modern society, more and people are suffering from stress. The accumulation of stress will result in poor health condition to people. Effectively detecting the human being time provides a helpful way for better manage their Much work has been done on recognizing level by extracting features bio-signals acquired physiological sensors. However, little focused feature selection. this paper, we propose selection method based Principal Component Analysis (PCA). After selected, effectiveness...
Traditional conversational systems can only access the given query during response generation, leading to meaningless responses. To this end, researchers proposed enhance dialogue generation by integrating external knowledge. Although such methods have achieved remarkable gains, use of single-source knowledge often makes existing knowledge-enhanced degenerate into traditional models in real scenarios because insufficient coverage improve applicability methods, we propose two novel frameworks...
Matching suitable jobs provided by employers with qualified candidates is a crucial task for online recruitment. Typically, and have specific expectations in recruitment market, leading them to prefer similar candidates, respectively. Metric learning provides promising way capture the similarity propagation between jobs. However, existing metric technologies rely on symmetric distance measures, which fail model asymmetric relationships of bilateral users (i.e., employers) two-way selective...
Federated learning has become a promising solution for collaboration among medical institutions. However, data owned by each institution would be highly heterogeneous and the distribution is always non-independent identical (non-IID), resulting in client drift unsatisfactory performance. Despite existing federated methods attempting to solve non-IID problems, they still show marginal advantages but rely on frequent communication which incur high costs privacy concerns. In this paper, we...
With the increasing complexity of modern software systems, it is essential yet hard to detect anomalies and diagnose problems precisely. Existing log-based anomaly detection approaches rely on a few key assumptions system logs perform well in some experimental systems. However, real-world industrial systems are often with poor logging quality, which noisy violate existing approaches. This makes these inefficient. paper first conducts comprehensive study three large-scale Through study, we...
The access control mechanisms of existing cloud systems, mainly OpenStack, fail to provide two key factors: i) centralized mediation and ii) flexible policy customization. This situation prevents administrators end customers from enhancing their security. Furthermore, a variety clouds have implemented systems policies in separated ways. might confuse the whose businesses are built on multiple clouds, as they take efforts accommodate for different platforms. OpenStack Security Modules (OSM)...
Over the last few years, cloud computing industry has witnessed wider adoption of container-based technologies. And it is obvious to see that Docker become de facto standard approaches. However, security mechanism far from satisfaction owing its rapid development without adequate concerns. This paper primarily identifies several possible covert channels against Docker, which causes critical results like information leak between one container and another (or even host). Furthermore, we also...