pygmtools.neural_solvers

Neural network-based graph matching solvers. It is recommended to integrate these networks as modules into your existing deep learning pipeline (either supervised, unsupervised or reinforcement learning).

Functions

cie	The CIE (Channel Independent Embedding) graph matching neural network model for processing two individual graphs (KB-QAP).
genn_astar	The GENN-A* (Graph Edit Neural Network A*) solver for graph matching (and graph edit distance) based on the fusion of traditional A* and Neural Network.
ipca_gm	The IPCA-GM (Iterative Permutation loss and Cross-graph Affinity Graph Matching) neural network model for processing two individual graphs (KB-QAP).
ngm	The NGM (Neural Graph Matching) model for processing the affinity matrix (the most general form of Lawler's QAP).
pca_gm	The PCA-GM (Permutation loss and Cross-graph Affinity Graph Matching) neural network model for processing two individual graphs (KB-QAP).