def process_sparse(graph_list, node_feat, edge_feat):
graph_sizes = [graph_list[i].num_nodes for i in range(len(graph_list))]
node_degs = [
torch.Tensor(graph_list[i].degs) + 1 for i in range(len(graph_list))
]
node_degs = torch.cat(node_degs).unsqueeze(1)
n2n_sp, e2n_sp, subg_sp = GNNLIB.PrepareSparseMatrices(graph_list)
n2n_sp = n2n_sp.cuda()
e2n_sp = e2n_sp.cuda()
subg_sp = subg_sp.cuda()
node_degs = node_degs.cuda()
node_feat = Variable(node_feat)
if edge_feat is not None:
edge_feat = Variable(edge_feat)
if torch.cuda.is_available() and isinstance(node_feat,
torch.cuda.FloatTensor):
edge_feat = edge_feat.cuda()
n2n_sp = Variable(n2n_sp)
e2n_sp = Variable(e2n_sp)
subg_sp = Variable(subg_sp)
node_degs = Variable(node_degs)
if edge_feat is not None:
input_edge_linear = edge_feat
e2npool_input = gnn_spmm(e2n_sp, input_edge_linear)
node_feat = torch.cat([node_feat, e2npool_input], 1)
return graph_sizes, n2n_sp, e2n_sp, subg_sp, node_degs
def forward(self, graph_list, node_feat, edge_feat):
graph_sizes = [graph_list[i].num_nodes for i in range(len(graph_list))]
node_degs = [
torch.Tensor(graph_list[i].degs) + 1
for i in range(len(graph_list))
]
node_degs = torch.cat(node_degs).unsqueeze(1)
n2n_sp, e2n_sp, subg_sp = GNNLIB.PrepareSparseMatrices(graph_list)
if torch.cuda.is_available() and isinstance(node_feat,
torch.cuda.FloatTensor):
n2n_sp = n2n_sp.cuda()
e2n_sp = e2n_sp.cuda()
subg_sp = subg_sp.cuda()
node_degs = node_degs.cuda()
node_feat = Variable(node_feat)
if edge_feat is not None:
edge_feat = Variable(edge_feat)
if torch.cuda.is_available() and isinstance(
node_feat, torch.cuda.FloatTensor):
edge_feat = edge_feat.cuda()
n2n_sp = Variable(n2n_sp)
e2n_sp = Variable(e2n_sp)
subg_sp = Variable(subg_sp)
node_degs = Variable(node_degs)
h = self.sortpooling_embedding(node_feat, edge_feat, n2n_sp, e2n_sp,
subg_sp, graph_sizes, node_degs)
return h
def forward(self, graph_list, node_feat, edge_feat):
graph_sizes = [graph_list[i].num_nodes for i in range(len(graph_list))]
# print('node_feat:', node_feat)
dv = node_feat.device
# print('dv:', dv)
n2n_sp, e2n_sp, subg_sp = GNNLIB.PrepareSparseMatrices(graph_list)
n2n_sp, e2n_sp, subg_sp = n2n_sp.to(dv), e2n_sp.to(dv), subg_sp.to(dv)
node_degs = 0
node_feat = Variable(node_feat)
if edge_feat is not None:
edge_feat = Variable(edge_feat)
n2n_sp = Variable(n2n_sp)
e2n_sp = Variable(e2n_sp)
subg_sp = Variable(subg_sp)
h = self.attention_gcn(node_feat, edge_feat, n2n_sp, e2n_sp, node_degs)
h = self.attention_pooling(h, subg_sp, graph_sizes)
# reg_term = self.get_reg(self.reg)
# return h, (reg_term/subg_sp.size()[0])
return h