def test_EFN_masking(mask_val):
n, m1, m2 = (50, 10, 20)
input_dim = 3
X_train = [np.random.rand(n, m1), np.random.rand(n, m1, input_dim)]
y_train = np.random.rand(n, 2)
efn = archs.EFN(input_dim=input_dim,
Phi_sizes=[10],
F_sizes=[10],
mask_val=mask_val)
efn.fit(X_train, y_train, epochs=1)
X_test = [np.random.rand(n, m2), np.random.rand(n, m2, input_dim)]
X_test_mask = [
np.concatenate((X_test[0], mask_val * np.ones((n, 5))), axis=1),
np.concatenate((X_test[1], mask_val * np.ones((n, 5, input_dim))),
axis=1)
]
assert epsilon_diff(efn.predict(X_test), efn.predict(X_test_mask), 10**-15)
kf = K.function(inputs=efn.inputs, outputs=efn.latent)
pure_mask = kf([0 * X_test[0] + mask_val, 0 * X_test[1] + mask_val])[0]
assert epsilon_diff(pure_mask, 0, 10**-15)
def test_EFN_global_features(nglobal):
n, m = 50, 10
X_train = [
np.random.rand(n, m),
np.random.rand(n, m, 2),
np.random.rand(n, nglobal)
]
Y_train = np.random.rand(n, 2)
X_val = [
np.random.rand(n // 10, m),
np.random.rand(n // 10, m, 2),
np.random.rand(n // 10, nglobal)
]
Y_val = np.random.rand(n // 10, 2)
efn = archs.EFN(input_dim=2,
Phi_sizes=[10],
F_sizes=[10],
num_global_features=nglobal,
summary=False)
hist = efn.fit(X_train,
Y_train,
epochs=1,
batch_size=5,
validation_data=(X_val, Y_val))
efn._global_feature_tensor
def test_EFN_required(input_dim, Phi_sizes, F_sizes):
n, m = 50, 10
X_train = [np.random.rand(n, m), np.random.rand(n, m, input_dim)]
Y_train = np.random.rand(n, 2)
efn = archs.EFN(input_dim=input_dim,
Phi_sizes=Phi_sizes,
F_sizes=F_sizes,
summary=False)
efn.fit(X_train, Y_train, epochs=1, batch_size=10)
efn.inputs, efn.latent, efn.weights, efn.outputs, efn.Phi, efn.F
def test_EFN_modelcheck(model_path, save_while_training, save_weights_only,
modelcheck_opts):
n, m = 50, 10
X_train = [np.random.rand(n, m), np.random.rand(n, m, 2)]
Y_train = np.random.rand(n, 2)
X_val = [np.random.rand(n // 10, m), np.random.rand(n // 10, m, 2)]
Y_val = np.random.rand(n // 10, 2)
efn = archs.EFN(input_dim=2,
Phi_sizes=[10],
F_sizes=[10],
summary=False,
filepath=model_path,
save_while_training=save_while_training,
save_weights_only=save_weights_only,
modelcheck_opts=modelcheck_opts)
hist = efn.fit(X_train,
Y_train,
epochs=2,
batch_size=10,
validation_data=[X_val, Y_val])