MiNT: Multi-Network Training for Transfer Learning on Temporal Graphs
This work addresses the challenge of transfer learning for temporal graphs, offering a foundational step towards Temporal Graph Foundation Models, though it appears incremental in extending pre-training to multiple networks.
The paper tackles the problem of learning from multiple temporal networks to improve transferability to unseen networks, achieving state-of-the-art results in zero-shot inference with a novel pre-training approach on up to 64 networks.
Temporal Graph Learning (TGL) has become a robust framework for discovering patterns in dynamic networks and predicting future interactions. While existing research has largely concentrated on learning from individual networks, this study explores the potential of learning from multiple temporal networks and its ability to transfer to unobserved networks. To achieve this, we introduce Temporal Multi-network Training MiNT, a novel pre-training approach that learns from multiple temporal networks. With a novel collection of 84 temporal transaction networks, we pre-train TGL models on up to 64 networks and assess their transferability to 20 unseen networks. Remarkably, MiNT achieves state-of-the-art results in zero-shot inference, surpassing models individually trained on each network. Our findings further demonstrate that increasing the number of pre-training networks significantly improves transfer performance. This work lays the groundwork for developing Temporal Graph Foundation Models, highlighting the significant potential of multi-network pre-training in TGL.