Class GraphConv

java.lang.Object
ai.beestream.jitnn.nn.Module
ai.beestream.jitnn.nn.GraphConv
All Implemented Interfaces:
AutoCloseable

public final class GraphConv extends Module
Graph convolution (Kipf and Welling, 2017), PyTorch Geometric's GCNConv:  (x W^T) + b, with  the normalised adjacency SparseMatrix.gcn(int, int[], int[]) of the graph, given with each call since one layer serves many graphs. x [nodes, in] gives [nodes, out].
  • Constructor Details

    • GraphConv

      public GraphConv(int in, int out, long seed)
      Graph convolution from in to out features: a linear map without bias initialised from seed, and a bias of zeros.
  • Method Details