$\texttt{SynC}$: Synergistic Boosting of Structure and Representation for Deep Graph Clustering

TL;DR

SynC is a deep graph clustering framework that synergistically enhances structure and representation learning, improving performance especially on low homophily graphs, with shared weights and a structure fine-tuning strategy.

Contribution

The paper introduces SynC, a novel deep graph clustering method that jointly optimizes structure and representation with shared weights and a fine-tuning strategy for better generalization.

Findings

SynC outperforms existing methods on benchmark datasets.

Shared weights enable synergistic boosting of structure and representation.

The structure fine-tuning improves generalization on low homophily graphs.

Abstract

Employing graph neural networks (GNNs) for graph clustering has shown promising results in deep graph clustering. However, existing methods disregard the reciprocal relationship between representation learning and structure augmentation: the more homogeneous the graph, the more cohesive the node representations; the more cohesive the node representations, the more reliable the structure augmentation becomes. Moreover, the generalization ability of existing GNN-based models on the low homophily graph is relatively poor. To this end, we propose a graph clustering framework named Synergistic Deep Graph Clustering Network (SynC). SynC employs a Transform Input Graph Auto-Encoder (TIGAE) to obtain high-quality embeddings via mitigating the representations collapse issue of GAE for guiding structure augmentation. Then, we re-capture neighborhood representations on the refined graph to obtain clustering-friendly embeddings and conduct self-supervised clustering. Notably, these two stages share weights, resulting in synergistic boosting while significantly reducing the number of model parameters. Additionally, we introduce a structure fine-tuning strategy to improve the model's generalization on the low homophily graph. Extensive experiments on benchmark datasets demonstrate the superiority of SynC. The code is released at GitHub.