This paper studies a novel problem of out-of-distribution graph models merging, which aims to construct a generalized model from multiple graph models pre-trained on different domains with distribution discrepancy. This problem is challenging because of the difficulty in learning domain-invariant knowledge implicitly in model parameters and consolidating expertise from potentially heterogeneous GNN backbones. In this work, we propose a graph generation strategy that instantiates the mixture distribution of multiple domains. Then, we merge and fine-tune the pre-trained graph models via a MoE module and a masking mechanism for generalized adaptation. Our framework is architecture-agnostic and can operate without any source/target domain data. Both theoretical analysis and experimental results demonstrate the effectiveness of our approach in addressing the model generalization problem.
@article{arxiv.2506.03674,
title = {Out-of-Distribution Graph Models Merging},
author = {Yidi Wang and Ziyue Qiao and Jiawei Gu and Xubin Zheng and Pengyang Wang and Xiaobing Pei and Xiao Luo},
journal= {arXiv preprint arXiv:2506.03674},
year = {2026}
}