A5070769211- Privacy-Preserving Technologies in Data
- Cryptography and Data Security
- Network Security and Intrusion Detection
- Blockchain Technology Applications and Security
- Spam and Phishing Detection
- Internet Traffic Analysis and Secure E-voting
- Privacy, Security, and Data Protection
- Adversarial Robustness in Machine Learning
- Software-Defined Networks and 5G
- Caching and Content Delivery
- UAV Applications and Optimization
- Misinformation and Its Impacts
- Mobile Crowdsensing and Crowdsourcing
- Advanced Malware Detection Techniques
- User Authentication and Security Systems
- Stochastic Gradient Optimization Techniques
- IoT and Edge/Fog Computing
- Advanced Image Processing Techniques
- Advanced Wireless Network Optimization
- Image Enhancement Techniques
- Complex Network Analysis Techniques
- Recommender Systems and Techniques
- Cloud Data Security Solutions
- Patient Dignity and Privacy
- Multimodal Machine Learning Applications
Minzu University of China
2023-2025
Beijing Institute of Technology
2018-2022
State Key Laboratory of Cryptology
2019
Machine learning (ML) techniques have been widely used in many smart city sectors, where a huge amount of data is gathered from various (IoT) devices. As typical ML model, support vector machine (SVM) enables efficient classification and thereby finds its applications real-world scenarios, such as disease diagnosis anomaly detection. Training an SVM classifier usually requires collection labeled IoT multiple entities, raising great concerns about privacy. Most the existing solutions rely on...
Existing distributed denial-of-service attack detection in software defined networks (SDNs) typically perform a single domain. In reality, abnormal traffic usually affects multiple network domains. Thus, cross-domain has been proposed to improve performance. However, when participating detection, the domain of each SDN needs provide large amount real data, from which private information may be leaked. multiparty privacy protection schemes often achieve guarantees by sacrificing accuracy or...
Machine learning (ML) techniques are expected to be used for specific applications in Vehicular Social Networks (VSNs). Support vector machine (SVM) is one of the typical ML methods and widely its high efficiency. Due limitation data sources, collected by different entities usually contain attributes that quite different. However, some real-world scenarios, when training an SVM classifier, many face same problem they lacking with adequate attributes. Thus multiple required share combine a...
Federated learning (FL) protects training data in clients by collaboratively local machine models of for a global model, instead directly feeding the to server. However, existing studies show that FL is vulnerable various attacks, resulting leakage or interfering with model training. Specifically, an adversary can analyze gradients and infer clients' data, poison generate inaccurate model. It extremely challenging guarantee strong privacy protection while ensuring robustness None achieve...
With the rapid advances in information and communication technologies, Internet of Things (IoT) has become large complex, bearing tremendous amounts data running devices various scenarios. Leveraging artificial intelligence (AI) IoT can achieve superior extraction, analytics, decision making, which resulted revolutionized AI (AIoT). AIoT alleviate pressure storage, computation, communication. Despite promising features brought by combining technologies into infrastructure, systems still face...
Heterogeneous networks, as a critical component of modern communication technology, have experienced rapid development in recent years [...]
Real-time forwarding prediction for predicting online contents' popularity is beneficial to various social applications enhancing interactive behaviors. Cascade graphs, formed by propagation, play a vital role in real-time prediction. Existing cascade graph modeling methods are inadequate embed graphs that have hub structures and deep paths, or they fail handle the short-term outbreak of amount. To this end, we propose novel method includes an effective approach embedding variation sensitive...
Empowered by promising artificial intelligence, the traditional Internet of Things is evolving into Artificial Intelligence (AIoT), which an important enabling technology for Industry 4.0. Collaborative learning a key AIoT to build machine (ML) models on distributed datasets. However, there are two critical concerns collaborative AIoT: privacy leakage sensitive data and dishonest computation. Specifically, contains information users, cannot be openly shared model learning. Furthermore,...
In developing smart city, the growing popularity of machine learning (ML) that appreciates high-quality training data sets generated from diverse Internet-of-Things (IoT) devices raises natural questions about privacy guarantees can be provided in such settings. Privacy-preserving ML an aggregation scenario enables a model demander to securely train models with sensitive IoT gathered devices. The existing solutions are generally server aided, cannot deal collusion threat between servers or...
Sentiment Quantification aims to detect the overall sentiment polarity of users from a set reviews corresponding target. Existing methods equally treat and aggregate individual reviews' judge polarity. However, confidence each review is not equal in quantification where perturbation arising high- low-confidence may degrade accuracy Quantification. Specifically, fake with deceptive sentiments are low confidence, which perturbs prediction. Whereas, some generated by responsible high...
Semantic communication has emerged as a promising technique that can integrate the meaning and context of information into process. It goes beyond raw data transmission enables sender receiver to transmit recover semantic messages through their respective knowledge bases. However, it is challenging achieve secure due complicated security privacy threats in communication. In this article, we first introduce background architecture Then, sort out multiple potential existing each step Following...
Community detection has been a subject of extensive research due to its broad applications across social media, computer science, biology, and complex systems. Modularity stands out as predominant metric guiding community detection, with numerous algorithms aimed at maximizing modularity. However, modularity encounters resolution limit problem when identifying small structures. To tackle this challenge, paper presents novel approach by defining structure information from the perspective...
In the realms of Internet Things (IoT) and artificial intelligence (AI) security, ensuring integrity quality visual data becomes paramount, especially under low-light conditions, where image enhancement emerges as a crucial technology. However, current methods for enhancing images conditions still face some challenging issues, including inability to effectively handle uneven illumination distribution, suboptimal denoising performance, insufficient correlation among branch network. Addressing...
In this paper, we propose a system framework of Pub/Sub-based multimodal data transmission to mobile terminals. For various sensors, classify the in different topics, but transmit them uniform channel, with mechanisms. case, terminals receive via mechanisms, push notification and request-response connection. The proposed makes more efficient flexible.