Visible to the public Biblio

Filters: Author is Chen, Hongwei  [Clear All Filters]
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z 
Xu, Hui, Zhang, Wei, Gao, Man, Chen, Hongwei.  2020.  Clustering Analysis for Big Data in Network Security Domain Using a Spark-Based Method. 2020 IEEE 5th International Symposium on Smart and Wireless Systems within the Conferences on Intelligent Data Acquisition and Advanced Computing Systems (IDAACS-SWS). :1—4.
Considering the problem of network security under the background of big data, the clustering analysis algorithms can be utilized to improve the correctness of network intrusion detection models for security management. As a kind of iterative clustering analysis algorithm, K-means algorithm is not only simple but also efficient, so it is widely used. However, the traditional K-means algorithm cannot well solve the network security problem when facing big data due to its high complexity and limited processing ability. In this case, this paper proposes to optimize the traditional K-means algorithm based on the Spark platform and deploy the optimized clustering analysis algorithm in the distributed architecture, so as to improve the efficiency of clustering algorithm for network intrusion detection in big data environment. The experimental result shows that, compared with the traditional K-means algorithm, the efficiency of the optimized K-means algorithm using a Spark-based method is significantly improved in the running time.
Zhang, Rui, Chen, Hongwei.  2019.  Intrusion Detection of Industrial Control System Based on Stacked Auto-Encoder. 2019 Chinese Automation Congress (CAC). :5638—5643.

With the deep integration of industrial control systems and Internet technologies, how to effectively detect whether industrial control systems are threatened by intrusion is a difficult problem in industrial security research. Aiming at the difficulty of high dimensionality and non-linearity of industrial control system network data, the stacked auto-encoder is used to extract the network data features, and the multi-classification support vector machine is used for classification. The research results show that the accuracy of the intrusion detection model reaches 95.8%.