A5051411972- Privacy-Preserving Technologies in Data
- Rough Sets and Fuzzy Logic
- Indoor and Outdoor Localization Technologies
- Radio Wave Propagation Studies
- Precipitation Measurement and Analysis
- Evaluation Methods in Various Fields
- Cooperative Communication and Network Coding
- Cryptography and Data Security
- Software Testing and Debugging Techniques
- Opportunistic and Delay-Tolerant Networks
- Millimeter-Wave Propagation and Modeling
- Evaluation and Optimization Models
- Neural Networks and Applications
- Electromagnetic Simulation and Numerical Methods
- Industrial Technology and Control Systems
- Central Venous Catheters and Hemodialysis
- Cardiac Arrest and Resuscitation
- Advanced Computational Techniques and Applications
- Caching and Content Delivery
- Sarcoma Diagnosis and Treatment
- Civil and Geotechnical Engineering Research
- RFID technology advancements
- Wireless Communication Security Techniques
- Mobile Crowdsensing and Crowdsourcing
- Subtitles and Audiovisual Media
Sun Yat-sen University
2019-2023
Naval University of Engineering
2006-2020
Guangdong Polytechnic Normal University
2019-2020
Chinese PLA General Hospital
2016
Second Xiangya Hospital of Central South University
2016
Central South University
2016
Mount Sinai Beth Israel
2015
Affiliated Hospital of Guizhou Medical University
2015
Guiyang Medical University
2015
Lanzhou University of Finance and Economics
2014
Intraosseous access is a rapid and effective route of fluid drug administration. Its use has been proven in emergency medicine, pediatrics, the military. We aimed to assess its performance utilization against landmark-guided central venous catheter placement during inpatient medical emergencies.Prospective observational study.Eight hundred fifty-six-bed urban teaching hospital.Adult inpatients requiring emergencies.Intraosseous device training was added standard beginning February 2012. were...
Deep learning-based Wi-Fi indoor fingerprint localization, which requires a large received signal strength (RSS) dataset for training, has been widely studied. Federated learning (FL) is recently introduced into localization in order to address the problem of data sharing without privacy disclosure. However, under serious heterogeneity FL, averaged model will perform worse individual client. In this letter, multi-level federated graph and self-attention based personalized method proposed...
Fingerprint-based indoor localization has drawn increasing attention with the development of deep learning. Nevertheless, it faces challenges from frequent data collection and corresponding exposure privacy. Federated Learning (FL) is introduced into recently for overcoming these challenges. However, most current FL-based studies only focus on static distributions ignore fact that users have their preferred requirements. In this paper, two typical scenarios are considered. Clients various...
AIM:To explore Chinese physicians' perceptions towards fecal microbiota transplantation (FMT) and to provide information an assessment of FMT development in China. METHODS:A self-administered questionnaire was developed according the practice guidelines distributed physicians hospitals via Internet Research Electronic Data Capture (REDcap) software electronic mails assess their attitudes toward knowledge FMT.The included a brief introduction that followed by 20 questions.The participants...
This letter presents a higher order finite-difference (FD) and Padé approximations method for the three-dimensional (3-D) parabolic equation (PE) to predict radio-wave propagation. uses fourth-order FD approximation of differential operator in transverse direction propagation direction. The (4FDHP) is then derived. Leontovich impedance boundary 4FDHP second-order (2FDHP) are also important problem angle different 3D-PE investigated. Simulated results show that proposed achieves larger...
In this letter, we propose a high-accuracy low-cost hybrid finite-difference time-domain (FDTD)/alternating-direction-implicit parabolic equation (ADI-PE) method for ultralarge-scale electromagnetic simulation on distributed computing platform. The offers fast and precise 3-D deterministic radio-wave propagation predictions in difficult-terrain scenario with specific detailed structures both near field far field. Since the typical parallel ADI-PE suffers from high-frequency data...
By caching the most popular content into mobile devices, users can retrieve directly from nearby devices through Device to (D2D) communications, which significantly reduce backhaul traffic and improve network performance. Most existing D2D cache placement strategies are proactive approaches, cannot deal with problem of timely updating. In this paper, we propose a coding-based socially-aware strategy, takes geographical proximity social relationships consideration. First, physical high...
Deep learning-assisted indoor fingerprint localization based on frequent data collection is motivating renewed interest via crowdsourcing. Uploading raw training may cause exposure of privacy in Federated Learning (FL) thus introduced into since its advantages protection. Nevertheless, most current FL-based studies do not consider deploying systems real wireless environments. Furthermore, transmission latency and outages caused by unreliable networks are ignored. In addition, centralized...
Complete text of publication follows. Despite the seemingly ever growing power computers, inversions full 3D DC resistivity data are still challenging and time-consuming. Since inversion processes based on forward modeling, speed up modeling is always an interesting topic. As iterative solver for discretized large linear equation systems, Multigrid (MG) methods known their high convergence rate which independent to number grid nodes. However, when applied this attractive property may be...
By taking advantage of Deep Learning (DL), Received signal strength (RSS) fingerprint-based indoor localization has attracted more attention. Training DL models require an immense amount RSS samples, and crowdsourcing been another way to collect data by users. Recently, some researchers apply Federated (FL) in instead for privacy protection. However, practical issues are still not considered.For further guaranteeing deploying systems real time, this paper proposes OPFL, a privacy-preserved...
Aimed at the limitation of feedforward and feedback ANN, shortcoming that diagnostic characteristic parameters are considered separately in conventional fault forecast method for machinery equipment, multivariable gray model, MGM(l,n), RBF network introduced into prediction, which allows to be described from viewpoint systems. It predicts future considering past current information, then is used predict online. The prediction example indicates model has good precision. offers an effective...
A new approach to fault diagnosis is presented, with a combination of D-S theory evidence and AHP. The method for constructing assignment function basic probability based on AHP 5-unit preference scale proposed in detail its comparison different from the pairwise standard In addition, weight each sensor taken into account process combination. An illustrative example given expatiate application diagnosis. After theses pieces multi sensors are combined, reliability diagnostic result improved...
Device to (D2D) communication is a key technology in 5th generation wireless systems increase capacity and spectral efficiency. Applying caching into D2D networks, the device can retrieve content from other devices by establishing links. In this way, backhaul traffic be significantly reduced. However, most of existing schemes are proactive caching, which cannot satisfy requirement real-time updating. paper, we propose an Indian Buffet Process based strategy (IBPSC). Firstly, construct...
The main problems of traffic at present are low design and over design, this paper puts forward equilibrium theory road capacity based on the then corresponding model for purpose guiding improving road, takes Zhouzhuang Road in Wuhan as example, conducts test different organization optimization scheme. Meanwhile, will evaluate with VISSIN, demonstrates feasibility effectiveness model.
Dempster-Shafer theory is an effective method for uncertain inference. It can combine evidences from different sources and therefore it has found successful application in many domains. However, the traditional may be invalid if conflicts between are severe. Priority factors introduced to reallocate conflict probabilities so as improve evidence combination method. The improved increases probability of true propositions. Thus, result becomes more reasonable than methods. An illustrative...
At present, the technique of network fault diagnosis has been a very hot research domain. The scholars from both domestic and abroad have put forward many approaches, but which some disadvantages in dealing with uncertain problems. This paper proposes rough set-support vector machine algorithm after studying set support theories. In order to reduce dimensions classification space, first diminishes attributes faults by means theory (RST), thus improves effect (SVM).
The computational complexity of reasoning within the Dempster-Shafer theory evidence is one major points criticism this formalism has to face. Various approximation algorithms have been suggested that aim at overcoming difficulty. This paper presents an improved practical algorithm through reducing number focal elements in belief function involved. In proposed algorithm, all every piece are classified into dereliction and remainder, basic probability assignments those derelictions reassigned...
To improve dynamic updating of privacy protected data release caused by multidimensional sensitivity attribute differences in relational data, we propose a method for protection based on the differences. By adopting multi-sensitive bucketization technology (MSB), this performs quantitative classification sensitive difference and recorded value, provides basic operation unit, thereby realizes among data. The experiment confirms that can secure efficiency while ensuring quality release.