ganl

gnl but on more abstract graphs, should probably not be used directly unless you unstand what the difference is

Arguments:

  • A: an Adjacency Matrix
  • X: a list of Feature vectors
  • gs: the node number
  • a: the abstraction factor
  • param: how many parameters for each feature vector
  • iterations=1: run each graph update step iteration times (one layer)
  • alin=[]: alinearity used, defined in glm
  • usei=False: use inverse graph update layers

Returns:

  • X: the updated feature object