def test_gc_lstm_layer():
"""
Testing the GCLSTM Layer.
"""
number_of_nodes = 100
edge_per_node = 10
in_channels = 64
out_channels = 16
K = 2
X, edge_index = create_mock_data(number_of_nodes, edge_per_node,
in_channels)
edge_weight = create_mock_edge_weight(edge_index)
layer = GCLSTM(in_channels=in_channels, out_channels=out_channels, K=K)
H, C = layer(X, edge_index)
assert H.shape == (number_of_nodes, out_channels)
assert C.shape == (number_of_nodes, out_channels)
H, C = layer(X, edge_index, edge_weight)
assert H.shape == (number_of_nodes, out_channels)
assert C.shape == (number_of_nodes, out_channels)
H, C = layer(X, edge_index, edge_weight, H, C)
assert H.shape == (number_of_nodes, out_channels)
assert C.shape == (number_of_nodes, out_channels)
def test_gc_lstm_layer():
"""
Testing the GCLSTM Layer.
"""
number_of_nodes = 100
edge_per_node = 10
in_channels = 64
out_channels = 16
K = 2
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
X, edge_index = create_mock_data(number_of_nodes, edge_per_node,
in_channels)
X = X.to(device)
edge_index = edge_index.to(device)
edge_weight = create_mock_edge_weight(edge_index).to(device)
layer = GCLSTM(in_channels=in_channels, out_channels=out_channels,
K=K).to(device)
H, C = layer(X, edge_index)
assert H.shape == (number_of_nodes, out_channels)
assert C.shape == (number_of_nodes, out_channels)
H, C = layer(X, edge_index, edge_weight)
assert H.shape == (number_of_nodes, out_channels)
assert C.shape == (number_of_nodes, out_channels)
H, C = layer(X, edge_index, edge_weight, H, C)
assert H.shape == (number_of_nodes, out_channels)
assert C.shape == (number_of_nodes, out_channels)
def __init__(self, h, z, k):
super(IGNNK, self).__init__()
self.time_dimension = h
self.hidden_dimnesion = z
self.order = k
self.recurrent = TGCN(h, 32).to(torch.device("cuda:0"))
# self.recurrent = AGCRN(number_of_nodes = 64,
# in_channels = h,
# out_channels = 32,
# K = 2,
# embedding_dimensions = 4)
self.recurrent = GCLSTM(64, 32, 1)
self.GNN1 = D_GCN(self.time_dimension, self.hidden_dimnesion,
self.order)
def __init__(self, node_features):
super(RecurrentGCN, self).__init__()
self.recurrent = GCLSTM(node_features, 32, 1)
self.linear = torch.nn.Linear(32, 1)
def __init__(self, node_features, num_classes):
super(RecurrentGCN, self).__init__()
self.recurrent_1 = GCLSTM(node_features, 32, 5)
self.recurrent_2 = GCLSTM(32, 16, 5)
self.linear = torch.nn.Linear(16, num_classes)