Cao Keyan

Dr. Cao Keyan


Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2020 in IEEE Access
Verified
Journal Article
Open access
CKCao Keyan
Cao Keyan
YLYefan Liu
Yefan Liu
GMGongjie Meng
Gongjie Meng
QSQimeng Sun
Qimeng Sun
With the rapid development of the Internet of Everything (IoE), the number of smart devices connected to the Internet is increasing, resulting in large-scale data, which has caused problems such as bandwidth load, slow response speed, poor security, and poor privacy in traditional cloud computing models. Traditional cloud computing is no longer sufficient to support the diverse needs of today’s intelligent society for data processing, so edge computing technologies have emerged. It is a new computing paradigm for performing calculations at the edge of the network. Unlike cloud computing, it emphasizes closer to the user and closer to the source of the data. At the edge of the network, it is lightweight for local, small-scale data storage and processing. This article mainly reviews the related research and results of edge computing. First, it summarizes the concept of edge computing and compares it with cloud computing. Then summarize the architecture of edge computing, keyword technology, security and privacy protection, and finally summarize the applications of edge computing.

10 March 2020 in IEEE Access
Verified
Journal Article
Open access
CKCao Keyan
Cao Keyan
JGJingjing Guo
Jingjing Guo
GMGongjie Meng
Gongjie Meng
HLHaoli Liu
Haoli Liu
YLYefan Liu
Yefan Liu
GLGui Li
Gui Li
With the advancement of the Internet of Everything era and the popularity of mobile devices, Location-based Social Networks (LBSN) have penetrated people’s lives. People can take advantage of portable edge terminal devices and use the geographic information in LBSN to arrange or adjust their travel plans. However, due to the explosive growth of current Internet applications and users, it has brought greater pressure and operation and maintenance costs to cloud storage. It is a key research direction based on location recommendation to accurately obtain the places of interest of users and push them to clients in such a large amount of original data. In order to better process the data generated by edge devices, this paper firstly uses the Rank-FBPR matrix decomposition framework based on social network to analyze the user’s personal preference function on the edge server. Then interact with the geographic information stored in the Cloud to cluster the POIs. And embeds the geographic information into the framework to get the candidate points of interest. Finally, the scores of candidate points of interest are predicted using the personal preference function and power law distribution, then a sorted list of points of interest is generated in descending order of scores, and the list is recommended to the target user. This algorithm effectively integrates the time information and geographic information of users’ check-in in the LBSN, and proposes a POIs recommendation algorithm that comprehensively considers edge devices and Cloud. The experiments verify the effectiveness of framework from both cold start and non-cold start. The experimental results on the Foursquare and the Yelp datasets show that Rank-FBPR has higher recommendation accuracy and recall than other comparison models, and can adapt to cold start problems.

13 July 2023 in Journal of X-Ray Science and Technology
Verified
Journal Article
Closed access
PDF file
CKCao Keyan
Cao Keyan
HTHangbo Tao
Hangbo Tao
ZWZhiqiong Wang
Zhiqiong Wang
XJXi Jin
Xi Jin
BACKGROUND: Accurate classification of benign and malignant pulmonary nodules using chest computed tomography (CT) images is important for early diagnosis and treatment of lung cancer. In terms of natural image classification, the ViT-based model has greater advantages in extracting global features than the traditional CNN model. However, due to the small image dataset and low image resolution, it is difficult to directly apply the ViT-based model to pulmonary nodule classification. OBJECTIVE: To propose and test a new ViT-based MSM-ViT model aiming to achieve good performance in classifying pulmonary nodules. METHODS: In this study, CNN structure was used in the task of classifying pulmonary nodules to compensate for the poor generalization of ViT structure and the difficulty in extracting multi-scale features. First, sub-pixel fusion was designed to improve the ability of the model to extract tiny features. Second, multi-scale local features were extracted by combining dilated convolution with ordinary convolution. Finally, MobileViT module was used to extract global features and predict them at the spatial level. RESULTS: CT images involving 442 benign nodules and 406 malignant nodules were extracted from LIDC-IDRI data set to verify model performance, which yielded the best accuracy of 94.04% and AUC value of 0.9636 after 10 cross-validations. CONCLUSION: The proposed new model can effectively extract multi-scale local and global features. The new model performance is also comparable to the most advanced models that use 3D volume data training, but its occupation of video memory (training resources) is less than 1/10 of the conventional 3D models.

2 November 2023 in Frontiers in Physiology
Verified
Journal Article
Open access
CKCao Keyan
Cao Keyan
MWMingrui Wang
Mingrui Wang
Human activity recognition (HAR) has recently become a popular research field in the wearable sensor technology scene. By analyzing the human behavior data, some disease risks or potential health issues can be detected, and patients’ rehabilitation progress can be evaluated. With the excellent performance of Transformer in natural language processing and visual tasks, researchers have begun to focus on its application in time series. The Transformer model models long-term dependencies between sequences through self-attention mechanisms, capturing contextual information over extended periods. In this paper, we propose a hybrid model based on the channel attention mechanism and Transformer model to improve the feature representation ability of sensor-based HAR tasks. Extensive experiments were conducted on three public HAR datasets, and the results show that our network achieved accuracies of 98.10%, 97.21%, and 98.82% on the HARTH, PAMAP2, and UCI-HAR datasets, respectively, The overall performance is at the level of the most advanced methods.

1 January 2016 in Neurocomputing
Verified
Journal Article
Closed access
CKCao Keyan
Cao Keyan
GWGuoren Wang
Guoren Wang
DHDonghong Han
Donghong Han
MBMei Bai
Mei Bai
SLShuoru Li
Shuoru Li

Verified
Journal Article
Open access
PDF file
HZHaochen Zou
Haochen Zou
CKCao Keyan
Cao Keyan
CJChong Jiang
Chong Jiang
Urban road traffic spatio-temporal characters reflect how citizens move and how goods are transported, which is crucial for trip planning, traffic management, and urban design. Video surveillance camera plays an important role in intelligent transport systems (ITS) for recognizing license plate numbers. This paper proposes a spatio-temporal visualization method to discover urban road vehicle density, city-wide regional vehicle density, and hot routes using license plate number data recorded by video surveillance cameras. To improve the accuracy of the visualization effect, during data analysis and processing, this paper utilized Internet crawler technology and adopted an outlier detection algorithm based on the Dixon detection method. In the design of the visualization map, this paper established an urban road vehicle traffic index to intuitively and quantitatively reveal the traffic operation situation of the area. To verify the feasibility of the method, an experiment in Guiyang on data from road video surveillance camera system was conducted. Multiple urban traffic spatial and temporal characters are recognized concisely and efficiently from three visualization maps. The results show the satisfactory performance of the proposed framework in terms of visual analysis, which will facilitate traffic management and operation.