Abstract:Graph contrastive learning (GCL) is a promising direction toward alleviating the label dependence, poor generalization and weak robustness of graph neural networks, learning representations with invariance, and discriminability by solving pretasks. The pretasks are mainly built on mutual information estimation, which requires data augmentation to construct positive samples with similar semantics to learn invariant signals and negative samples with dissimilar semantics to empower representation discriminability. However, an appropriate data augmentation configuration depends heavily on lots of empirical trials such as choosing the compositions of data augmentation techniques and the corresponding hyperparameter settings. We propose an augmentation-free GCL method, invariant-discriminative GCL (iGCL), that does not intrinsically require negative samples. iGCL designs the invariant-discriminative loss (ID loss) to learn invariant and discriminative representations. On the one hand, ID loss learns invariant signals by directly minimizing the mean square error (MSE) between the target samples and positive samples in the representation space. On the other hand, ID loss ensures that the representations are discriminative by an orthonormal constraint forcing the different dimensions of representations to be independent of each other. This prevents representations from collapsing to a point or subspace. Our theoretical analysis explains the effectiveness of ID loss from the perspectives of the redundancy reduction criterion, canonical correlation analysis (CCA), and information bottleneck (IB) principle. The experimental results demonstrate that iGCL outperforms all baselines on five node classification benchmark datasets. iGCL also shows superior performance for different label ratios and is capable of resisting graph attacks, which indicates that iGCL has excellent generalization and robustness. The source code is available at https://github.com/lehaifeng/ T-GCN/tree/master/iGCL.

Augmentation-Free Graph Contrastive Learning of Invariant-Discriminative Representations

Debiased Graph Contrastive Learning.

SimGRACE: A Simple Framework for Graph Contrastive Learning Without Data Augmentation

G-Censor: Graph Contrastive Learning with Task-Oriented Counterfactual Views

Towards Equivariant Graph Contrastive Learning via Cross-Graph Augmentation

Augmentation-Free Graph Contrastive Learning with Performance Guarantee

Full-Attention Driven Graph Contrastive Learning: with Effective Mutual Information Insight

Community-Invariant Graph Contrastive Learning

Adversarial Graph Augmentation to Improve Graph Contrastive Learning

Graph Contrastive Learning with Augmentations

CGCL: Collaborative Graph Contrastive Learning without Handcrafted Graph Data Augmentations

Graph Contrastive Learning with Reinforced Augmentation

Graph Contrastive Learning with Adaptive Augmentation

Graph Contrastive Learning with Personalized Augmentation

Graph Contrastive Learning with Generative Adversarial Network

Unbiased and augmentation-free self-supervised graph representation learning

Contrastive Graph Representation Learning with Adversarial Cross-view Reconstruction and Information Bottleneck

Prototypical Graph Contrastive Learning

Neighbor Contrastive Learning on Learnable Graph Augmentation

Graph Contrastive Learning with Reinforcement Augmentation

Let Invariant Rationale Discovery Inspire Graph Contrastive Learning