MiNT: Multi-Network Training for Transfer Learning on Temporal Graphs

TL;DR

This paper introduces MiNT, a multi-network pre-training method for temporal graph learning that enhances transferability and zero-shot inference across unseen dynamic networks.

Contribution

MiNT is the first approach to pre-train on multiple temporal networks, significantly improving transfer learning in temporal graph models.

Findings

MiNT outperforms models trained on individual networks in zero-shot inference.

Increasing pre-training networks improves transfer performance.

Pre-training on 64 networks enhances transfer to 20 unseen networks.

Abstract

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.