Document Type
Thesis
Date of Award
5-31-2025
Degree Name
Master of Science in Data Science - (M.S.)
Department
Data Science
First Advisor
Mengjia Xu
Second Advisor
Ioannis Koutis
Third Advisor
Hai Nhat Phan
Fourth Advisor
Shuai Zhang
Abstract
Dynamic graph embedding is a key technique for modeling temporal dependencies in evolving networks. While transformer-based models perform well, their quadratic complexity limits scalability on long graph sequences. This thesis compares transformer approaches with the Mamba architecture-a linear-complexity state-space model—for temporal graph embedding.
Two frameworks are proposed: DG-Mamba and GDG-Mamba. DG-Mamba uses standard GCN-based spatial encoding, while GDG-Mamba incorporates domain-aware edge features using Graph Isomorphism Network with Edge Convolution (GraphGINE). Experiments on UCI, Reality Mining, Slashdot, Bitcoin-OTC, and SBM datasets show that Mamba-based models match or exceed transformer performance, especially on graphs with high temporal variability.
The thesis also applies GDG-Mamba to PM2.5 forecasting using the KnowAir dataset. Replacing the GRU encoder in PM2.5-GNN with a Mamba-based temporal module improves the model's ability to capture long-range dependencies in pollutant transport.
These results highlight the scalability and effectiveness of Mamba-based architectures for dynamic graph modeling and environmental forecasting.
Recommended Citation
Pandey, Ashish, "A novel framework for dynamic graph representation learning with mamba" (2025). Theses. 3170.
https://digitalcommons.njit.edu/theses/3170
