Module ssl 

Graph Self-Supervised Learning (SSL) methods.

Provides contrastive learning and masked autoencoder approaches for learning graph representations without labels.

Sub-module	Method	Reference
contrastive	GraphCL, SimGRACE, NT-Xent loss	You et al. 2020; Xia 2022
masked_autoencoder	GraphMAE with SCE loss	Hou et al. 2022

Re-exports

pub use contrastive::augment_edges;

pub use contrastive::augment_features;

pub use contrastive::nt_xent_loss;

pub use contrastive::simgrace_perturb;

pub use contrastive::GraphClConfig;

pub use contrastive::ProjectionHead;

pub use masked_autoencoder::GraphMae;

pub use masked_autoencoder::GraphMaeConfig;

pub use pretrain::infonce_loss;

pub use pretrain::AttrReconConfig;

pub use pretrain::AttrReconConfig as AttributeReconConfig;

pub use pretrain::AttributeReconstructionObjective;

pub use pretrain::GraphContextConfig;

pub use pretrain::GraphContextPretrainer;

pub use pretrain::NodeMaskingConfig;

pub use pretrain::NodeMaskingPretrainer;

Modules

contrastive

Graph Contrastive Learning: GraphCL (You et al. 2020) and SimGRACE (Xia 2022).

masked_autoencoder

Graph Masked Autoencoder (GraphMAE, Hou et al. 2022).

pretrain

Graph pre-training strategies: node masking, graph-context contrastive, attribute reconstruction. Graph pre-training strategies for self-supervised learning.