Abstract:
Due to the time-evolving nature of many real-world networks, learning low-dimensional topological representations in dynamic environments is gaining attention. Dynamic Network Embedding (DNE) aims to efficiently update node embeddings while preserving network topology at each time step.
Most DNE methods only capture topological changes at or around the most affected nodes and update node embeddings. This approximation can improve efficiency but cannot preserve the global topology of a dynamic network at each time step because it does not consider the inactive sub-networks that receive accumulated topological changes propagated via high-order proximity.
We propose a new node selecting strategy to diversely select representative nodes over a network, coordinated with a new incremental learning paradigm of Skip-Gram-based embedding approach.
GloDyNE can outperform state-of-the-art DNE methods in three downstream tasks with a small fraction of nodes selected, according to extensive experiments. GloDyNE excels in graph reconstruction, demonstrating its global topology preservation.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
