magnet.graph_utils.loss_heterogeneous_domains

magnet.graph_utils.loss_heterogeneous_domains(Y: Tensor, graph: Data) → Tensor

Compute loss function for heterogeneous domains.

Parameters:

• y (torch.Tensor) – Evaluation output of the Neural Network: tensor of shape (num_nodes, 2) whose values are the probabilities of belonging to one of two sets assigned by the GNN to node of the graph.

• graph (Data) – Graph on which the GNN was evaluated.

Returns:

The value of the loss function.

Return type:

torch.Tensor

Notes

This loss function is the sum of the expected normalized cut and a term that penalizes the presence of very different physical groups in the same set of nodes, with suitable weights.