---
title: "Mitigating Dynamic Graph Distribution Shifts via Spectral Augmentation"
authors:
  - "Qianyu Song"
  - "Chao Li"
  - "Zhongying Zhao"
  - "Hua Duan"
  - "Qingtian Zeng"
doi: "10.24963/ijcai.2026/335"
url: "https://ijcai.org/proceedings/2026/335"
pdf: "https://ijcai.org/proceedings/2026/0335.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "3015-3023"
track: "Main Track"
keywords:
  - "Data Mining: Mining spatial and/or temporal data"
  - "Data Mining: Mining graphs"
---

# Mitigating Dynamic Graph Distribution Shifts via Spectral Augmentation

Qianyu Song, Chao Li, Zhongying Zhao, Hua Duan, Qingtian Zeng

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 3015-3023. DOI: [10.24963/ijcai.2026/335](https://doi.org/10.24963/ijcai.2026/335)

## Abstract

Dynamic Graph Neural Networks (DyGNNs) are facing challenges with distribution shifts between training and test data that are similar but not identical. Existing DyGNNs for out-of-distribution scenarios primarily focus on discovering invariant patterns in the spatial domain, overlooking its impact on the structural properties in the spectral domain. In this paper, we propose a spectral-based graph augmentation framework designed to investigate and improve generalization behavior in dynamic graphs under distribution shifts. Specifically, we augment the input graph spectra into a mixture of shift components by maximizing the variance of spectral distance and propose an efficient approximation to reduce the computational cost brought by eigen-decomposition. Building on this, we develop a multi-encoder architecture in which each encoder targets a specific spectral shift component to generate referential representations. Our method adopts a new learning objective to encourage the model to rely on robust spectral properties for better generalization under distribution shifts. Extensive experiments on both real-world and synthetic datasets demonstrate that our proposed method significantly outperforms state-of-the-art approaches in node classification and link prediction tasks. Source codes are available at https://github.com/SSQiana/DSPA.