A Transfer Framework for Enhancing Temporal Graph Learning in Data-Scarce Settings
This work addresses the challenge of limited training data for TGNNs in domains like social platforms and recommendation systems, offering a novel transfer method that is incremental but provides strong gains in low-data settings.
The paper tackles the problem of data scarcity in Temporal Graph Neural Networks (TGNNs) for predicting future connections in time-evolving graphs, introducing a transfer approach that disentangles node representations to enable effective knowledge transfer, resulting in performance improvements of up to 56% over non-transfer baselines and 36% over existing transfer strategies.
Dynamic interactions between entities are prevalent in domains like social platforms, financial systems, healthcare, and e-commerce. These interactions can be effectively represented as time-evolving graphs, where predicting future connections is a key task in applications such as recommendation systems. Temporal Graph Neural Networks (TGNNs) have achieved strong results for such predictive tasks but typically require extensive training data, which is often limited in real-world scenarios. One approach to mitigating data scarcity is leveraging pre-trained models from related datasets. However, direct knowledge transfer between TGNNs is challenging due to their reliance on node-specific memory structures, making them inherently difficult to adapt across datasets. To address this, we introduce a novel transfer approach that disentangles node representations from their associated features through a structured bipartite encoding mechanism. This decoupling enables more effective transfer of memory components and other learned inductive patterns from one dataset to another. Empirical evaluations on real-world benchmarks demonstrate that our method significantly enhances TGNN performance in low-data regimes, outperforming non-transfer baselines by up to 56\% and surpassing existing transfer strategies by 36\%