S2GAE: Self-Supervised Graph Autoencoders are Generalizable Learners with Graph Masking

Keyphrases

• Graph Autoencoder 100%
• Link Prediction 42%
• Classification Task 28%
• Node Graph 28%
• Self-supervised Learning 28%
• Learning Framework 14%
• Node Representation 14%
• Node Classification 14%
• Node Level 14%
• Benchmark Dataset 14%
• Publicly Available 14%
• Input Structure 14%
• Prediction Task 14%
• Contrastive Learning 14%
• Downstream Task 14%
• Graph Classification 14%
• Pre-training Model 14%
• Supervised Learning Method 14%
• Encoder-decoder 14%
• Edge-level 14%
• Link Classification 14%
• Missing Edge 14%
• Popular 14%
• Masking Strategy 14%
• Universal Representations 14%
• Ground-truth Labels 14%

Computer Science

• Autoencoder 100%
• Link Prediction 42%
• Self-Supervised Learning 28%
• Classification Task 28%
• Training Model 14%
• Learning Framework 14%
• Building-Blocks 14%
• Contrastive Learning 14%
• Learning Approach 14%
• Generalizability 14%