
Research Article
Dynamic Network Change Detection via Dynamic Network Representation Learning
@INPROCEEDINGS{10.1007/978-3-030-41114-5_48, author={Hao Feng and Yan Liu and Ziqiao Zhou and Jing Chen}, title={Dynamic Network Change Detection via Dynamic Network Representation Learning}, proceedings={Communications and Networking. 14th EAI International Conference, ChinaCom 2019, Shanghai, China, November 29 -- December 1, 2019, Proceedings, Part I}, proceedings_a={CHINACOM}, year={2020}, month={2}, keywords={Network representation learning Social network Egonet}, doi={10.1007/978-3-030-41114-5_48} }
- Hao Feng
Yan Liu
Ziqiao Zhou
Jing Chen
Year: 2020
Dynamic Network Change Detection via Dynamic Network Representation Learning
CHINACOM
Springer
DOI: 10.1007/978-3-030-41114-5_48
Abstract
The structure of the network in the real world is very complex, as the dynamic network structure evolves in time dimension, how to detect network changes accurately and further locate abnormal nodes is a research hotspot. Most current feature learning methods are difficult to capture a variety of network connectivity patterns, and have a high time complexity. In order to overcome this limitation, we introduce the network embedding method into the field of network change detection, we find that node-based egonet can better reflect the connectivity patterns of the node, so a dynamic network embedding model Egonet2Vec is proposed, which is based on extracting the connectivity patterns of the node-based egonets. After the dynamic network representation learning, we use a dynamic network change detection strategy to detect network change time points and locate abnormal nodes. We apply our method to real dynamic network datasets to demonstrate the validity of this method.