heterogeneous-graph-neural-networks

HACH-MVC is an end-to-end framework that constructs view-specific graphs, iteratively refines them through a heterogeneous GNN, and employs a learnable dual-fusion mechanism to derive a consensus graph and feature representation followed by a novel contrastive learning and bi-stage clustering methods.

A comprehensive bridge impact analysis system combining heterogeneous graph neural networks (HGNN) for closure-impact prediction with graph autoencoders (GAE/VGAE/HetVGAE) for unsupervised bridge similarity learning. This extends the system with heterogeneous graph variational autoencoders for metapath-based bridge classification.