torch_geometric.nn.conv.PDNConv

class PDNConv(in_channels: int, out_channels: int, edge_dim: int, hidden_channels: int, add_self_loops: bool = True, normalize: bool = True, bias: bool = True, **kwargs)[source]

Bases: MessagePassing

The pathfinder discovery network convolutional operator from the “Pathfinder Discovery Networks for Neural Message Passing” paper.

\[\mathbf{x}^{\prime}_i = \sum_{j \in \mathcal{N}(i) \cup \{i\}}f_{\Theta}(\textbf{e}_{(j,i)}) \cdot f_{\Omega}(\mathbf{x}_{j})\]

where \(z_{i,j}\) denotes the edge feature vector from source node \(j\) to target node \(i\), and \(\mathbf{x}_{j}\) denotes the node feature vector of node \(j\).

Parameters:
  • in_channels (int) – Size of each input sample.

  • out_channels (int) – Size of each output sample.

  • edge_dim (int) – Edge feature dimensionality.

  • hidden_channels (int) – Hidden edge feature dimensionality.

  • add_self_loops (bool, optional) – If set to False, will not add self-loops to the input graph. (default: True)

  • normalize (bool, optional) – Whether to add self-loops and compute symmetric normalization coefficients on the fly. (default: True)

  • bias (bool, optional) – If set to False, the layer will not learn an additive bias. (default: True)

  • **kwargs (optional) – Additional arguments of torch_geometric.nn.conv.MessagePassing.

Shapes:
  • input: node features \((|\mathcal{V}|, F_{in})\), edge indices \((2, |\mathcal{E}|)\), edge features \((|\mathcal{E}|, D)\) (optional)

  • output: node features \((|\mathcal{V}|, F_{out})\)

forward(x: Tensor, edge_index: Union[Tensor, SparseTensor], edge_attr: Optional[Tensor] = None) Tensor[source]

Runs the forward pass of the module.

reset_parameters()[source]

Resets all learnable parameters of the module.