def test_RGCN_apply_dense_directed():
G, features = create_graph_features(is_directed=True)
As = get_As(G)
As = [np.expand_dims(A.todense(), 0) for A in As]
generator = RelationalFullBatchNodeGenerator(G, sparse=False)
rgcnModel = RGCN([2],
generator,
num_bases=10,
activations=["relu"],
dropout=0.5)
x_in, x_out = rgcnModel.in_out_tensors()
model = keras.Model(inputs=x_in, outputs=x_out)
# Check fit method
out_indices = np.array([[0, 1]], dtype="int32")
preds_1 = model.predict([features[None, :, :], out_indices] + As)
assert preds_1.shape == (1, 2, 2)
# Check fit method
preds_2 = model.predict(generator.flow(["a", "b"]))
assert preds_2.shape == (1, 2, 2)
assert preds_1 == pytest.approx(preds_2)
def test_RGCN_apply_sparse_directed():
G, features = create_graph_features(is_directed=True)
As = get_As(G)
As = [A.tocoo() for A in As]
A_indices = [
np.expand_dims(np.hstack((A.row[:, None], A.col[:, None])), 0) for A in As
]
A_values = [np.expand_dims(A.data, 0) for A in As]
generator = RelationalFullBatchNodeGenerator(G, sparse=True)
rgcnModel = RGCN([2], generator, num_bases=10, activations=["relu"], dropout=0.5)
x_in, x_out = rgcnModel.build()
model = keras.Model(inputs=x_in, outputs=x_out)
# Check fit method
out_indices = np.array([[0, 1]], dtype="int32")
preds_1 = model.predict([features[None, :, :], out_indices] + A_indices + A_values)
assert preds_1.shape == (1, 2, 2)
# Check fit_generator method
preds_2 = model.predict_generator(generator.flow(["a", "b"]))
assert preds_2.shape == (1, 2, 2)
assert preds_1 == pytest.approx(preds_2)
def test_RGCN_init():
G, features = create_graph_features()
generator = RelationalFullBatchNodeGenerator(G)
rgcnModel = RGCN([2], generator, num_bases=10, activations=["relu"], dropout=0.5)
assert rgcnModel.layer_sizes == [2]
assert rgcnModel.activations == ["relu"]
assert rgcnModel.dropout == 0.5
assert rgcnModel.num_bases == 10
def test_kernel_and_bias_defaults():
graph, _ = create_graph_features()
generator = RelationalFullBatchNodeGenerator(graph, sparse=False)
rgcn = RGCN([2, 2], generator, num_bases=10)
for layer in rgcn._layers:
if isinstance(layer, RelationalGraphConvolution):
assert isinstance(layer.kernel_initializer,
tf.initializers.GlorotUniform)
assert isinstance(layer.bias_initializer, tf.initializers.Zeros)
assert layer.kernel_regularizer is None
assert layer.bias_regularizer is None
assert layer.kernel_constraint is None
assert layer.bias_constraint is None
def test_RGCN_save_load(tmpdir, num_bases, sparse):
graph, _ = create_graph_features()
generator = RelationalFullBatchNodeGenerator(graph, sparse=sparse)
rgcn = RGCN([2, 2], generator, num_bases=num_bases)
test_utils.model_save_load(tmpdir, rgcn)
