gammagl.models.GPRGNNModel

class GPRGNNModel(feature_dim, hidden_dim, num_class, drop_rate, K, Init, alpha, dprate, Gamma=None)[source]

Graph Convolutional Network proposed in “Adaptive Universal Generalized PageRank Graph Neural Network” paper.

Parameters:

  • feature_dim – input feature dim.

  • hidden_dim – hidden feature dim.

  • num_class – output feature dim.

  • drop_rate – drop rate for dropout layer.

  • K – K steps to propagate.

  • Init – initialization method.

  • alpha – help to initial the value of gamma.

  • dprate – special dropout for GPR-GNN (before starting GPR propagation) to dropout some nodes.

  • Gamma – Only used when Init == “WS”, self-defined initial value of gamma.

reset_parameters()[source]
forward(x, edge_index, edge_weight, num_nodes)[source]

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.