def test_weights():
input_dimensions = 4
number_of_nodes = 9
model = LinearAssociator(input_dimensions=input_dimensions,
number_of_nodes=number_of_nodes,
transfer_function="Hard_limit")
weights = model.get_weights()
assert weights.ndim == 2 and \
weights.shape[0] == number_of_nodes and \
weights.shape[1] == (input_dimensions)
model.set_weights(np.ones((number_of_nodes, input_dimensions)))
weights = model.get_weights()
assert weights.ndim == 2 and \
weights.shape[0] == number_of_nodes and \
weights.shape[1] == (input_dimensions)
assert np.array_equal(model.get_weights(),
np.ones((number_of_nodes, input_dimensions)))
model.initialize_weights(seed=3)
weights = np.array([[1.78862847, 0.43650985, 0.09649747, -1.8634927],
[-0.2773882, -0.35475898, -0.08274148, -0.62700068],
[-0.04381817, -0.47721803, -1.31386475, 0.88462238],
[0.88131804, 1.70957306, 0.05003364, -0.40467741],
[-0.54535995, -1.54647732, 0.98236743, -1.10106763],
[-1.18504653, -0.2056499, 1.48614836, 0.23671627],
[-1.02378514, -0.7129932, 0.62524497, -0.16051336],
[-0.76883635, -0.23003072, 0.74505627, 1.97611078],
[-1.24412333, -0.62641691, -0.80376609, -2.41908317]])
np.testing.assert_array_almost_equal(model.get_weights(),
weights,
decimal=3)
def test_train_hardlim_delta():
input_dimensions = 5
number_of_nodes = 5
for i in range(10):
model = LinearAssociator(input_dimensions=input_dimensions, number_of_nodes=number_of_nodes,
transfer_function="Hard_limit")
model.initialize_weights(seed=i + 1)
X_train = np.random.randn(input_dimensions, 100)
out = model.predict(X_train)
model.set_weights(np.random.randn(*model.get_weights().shape))
model.train(X_train, out, batch_size=10, num_epochs=200, alpha=0.1, gamma=0.1, learning="Delta")
new_out = model.predict(X_train)
np.testing.assert_array_almost_equal(out, new_out, decimal=4)
def test_pseudo_inverse_fit():
input_dimensions = 5
number_of_nodes = 5
model = LinearAssociator(input_dimensions=input_dimensions, number_of_nodes=number_of_nodes,
transfer_function="Linear")
model.initialize_weights(seed=1)
X_train = np.random.randn(input_dimensions, 10)
out = model.predict(X_train)
model.set_weights(np.zeros_like(model.get_weights()))
model.fit_pseudo_inverse(X_train, out)
new_out = model.predict(X_train)
np.testing.assert_array_almost_equal(out, new_out, decimal=4)