Reproduce “What's Behind the Mask: Understanding Masked Graph Modeling for Graph Autoencoders” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jintang Li, Ruofan Wu, Wangbin Sun, Liang Chen, Sheng Tian, Liang Zhu, Changhua Meng, Zibin Zheng and 1 more
This paper introduces Masked Graph Autoencoder (MaskGAE), a self-supervised learning framework that applies masked graph modeling to enhance graph autoencoders (GAEs). It provides theoretical justification for how masking reduces information redundancy in GAEs and demonstrates MaskGAE's state-of-the-art performance in link prediction and node classification across various datasets.
Key insights
- Minimizing GAE loss is asymptotically equivalent to maximizing mutual information between k-hop subgraphs of adjacent nodes, implicitly performing contrastive learning.
- Masking edges reduces "task-irrelevant information" or redundancy between k-hop subgraph views, which improves the generalizability of learned representations.
- Path-wise masking effectively breaks local connections, forcing the model to infer higher-order proximity and capture more robust structural information.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.