Identification of dynamic networks community by fusing deep learning and evolutionary clustering

Area of Science:

• Network analysis
• Social computing
• Machine learning

Background:

• Community detection is crucial for understanding dynamic networks.
• Existing shallow models struggle with complex, non-linear network structures.
• Accurate community detection in dynamic networks has significant implications.

Purpose of the Study:

• To propose a novel method for detecting evolving communities in dynamic networks.
• To overcome limitations of shallow models in capturing complex network structures.
• To enhance the accuracy and robustness of dynamic community detection.

Main Methods:

• Developed a novel dynamic community detection method by fusing Deep Learning and Evolutionary Clustering (DLEC).
• Utilized a matrix construction strategy to reveal community structures.
• Employed a multi-layer deep autoencoder for latent deep representation extraction.
• Incorporated graph regularization for community evolution smoothness.
• Applied K-means clustering in a low-dimensional space.

Main Results:

• The proposed DLEC algorithm effectively detects high-quality communities.
• DLEC demonstrates superior performance on both synthetic and real-world networks.
• The method successfully captures complex, non-linear structures in dynamic networks.

Conclusions:

• The fusion of deep learning and evolutionary clustering offers a powerful approach for dynamic community detection.
• DLEC provides an accurate and robust solution for identifying evolving communities.
• This framework advances the field of network analysis and social computing.