EC-BED-NETS: A Novel Deep Learning Framework for Recognizing Dominant Nodes in Multifaceted and Social Networks

Area of Science:

• Network Science
• Machine Learning
• Artificial Intelligence

Background:

• Identifying influential nodes is crucial in multifaceted and social networks.
• Traditional centrality measures struggle with the nonlinear relationships of functional importance.
• Existing methods often fail to accurately capture the complex structural and functional roles of nodes.

Purpose of the Study:

• To propose a novel hybrid boosted ensemble Long Short-Term Memory (LSTM) framework for accurate identification of influential nodes.
• To overcome the limitations of traditional centrality measures in detecting functional importance.
• To develop a robust method for classifying and ranking nodes based on their network influence.

Main Methods:

• Utilizing enhanced centrality methods to create feature vectors reflecting node positions.
• Employing a boosted deep learning framework, specifically LSTM, for node classification.
• Categorizing nodes based on constructed feature vectors and network measurements.

Main Results:

• The proposed hybrid boosted ensemble LSTM framework achieved a classification accuracy of 95.5%.
• The framework demonstrated superior performance compared to traditional centrality measurements.
• The model outperformed other existing machine learning and deep learning approaches.

Conclusions:

• The developed LSTM framework effectively identifies influential nodes in complex networks.
• This approach offers a significant advancement in network analysis and node influence detection.
• The study highlights the potential of deep learning, particularly LSTM, in solving complex network problems.