def test_mlp_convergence():
# Run until convergence
# assert that network can converge
model = mlp.MLP((2, 4, 2))
dataset = datasets.get_xor()
model.train(*dataset, retries=5, error_break=0.002)
assert validation.get_error(model, *dataset) <= 0.02
def test_mlp():
# Run for a couple of iterations
# assert that new error is less than original
model = mlp.MLP((2, 2, 2))
dataset = datasets.get_xor()
error = validation.get_error(model, *dataset)
model.train(*dataset, iterations=10)
assert validation.get_error(model, *dataset) < error
def test_mlp_classifier_convergence():
# Run until convergence
# assert that network can converge
model = mlp.MLP(
(2, 3, 2), transfers=SoftmaxTransfer(), error_func=CrossEntropyError())
dataset = datasets.get_and()
model.train(*dataset, retries=5, error_break=0.002)
assert validation.get_error(model, *dataset) <= 0.02
def test_mlp_classifier():
# Run for a couple of iterations
# assert that new error is less than original
model = mlp.MLP(
(2, 2, 2), transfers=SoftmaxTransfer(), error_func=CrossEntropyError())
dataset = datasets.get_xor()
error = validation.get_error(model, *dataset)
model.train(*dataset, iterations=20)
assert validation.get_error(model, *dataset) < error
def test_mlp_perceptron():
# Given known inputs and weights, test expected outputs
model = mlp.MLP((2, 1), transfers=mlp.LinearTransfer())
model._weight_matrices[0][0][0] = 0.0
model._weight_matrices[0][1][0] = 0.5
model._weight_matrices[0][2][0] = -0.5
assert (model.activate([1, 1]) == [0.0]).all()
model._weight_matrices[0][1][0] = 1.0
model._weight_matrices[0][2][0] = 2.0
assert (model.activate([1, 1]) == [3.0]).all()
def test_MLP_activate_matrix():
model = mlp.MLP((2, 2, 2), transfers=[LinearTransfer(), LinearTransfer()])
# Set weights for deterministic results
model._bias_vec = numpy.ones(model._bias_vec.shape)
model._weight_matrices = [numpy.ones(weight_matrix.shape) for weight_matrix in model._weight_matrices]
# Activate
assert helpers.approx_equal(model.activate([[0, 0], [0.5, 0.5]]), [[2, 2], [4, 4]])
assert helpers.approx_equal(model.activate([[1, 0], [0.5, 1]]), [[4, 4], [5, 5]])
assert helpers.approx_equal(model.activate([[1, 1], [0, 0.5]]), [[6, 6], [3, 3]])
def test_mlp_bias():
# Should have bias for each layer
model = mlp.MLP((2, 4, 3))
# +1 input for bias
assert model._weight_matrices[0].shape == (3, 4)
assert model._weight_matrices[1].shape == (5, 3)
# First input should always be 1
model.activate([0, 0])
assert model._weight_inputs[0][0] == 1.0
assert model._weight_inputs[1][0] == 1.0
assert model._weight_inputs[2][0] == 1.0
def test_MLP_reset():
shape = (random.randint(1, 10), random.randint(1, 10), random.randint(1, 10))
model = mlp.MLP(shape)
model_2 = mlp.MLP(shape)
# Resetting different with the same seed should give the same model
prev_seed = random.randint(0, 2**32-1)
try:
random.seed(0)
numpy.random.seed(0)
model.reset()
random.seed(0)
numpy.random.seed(0)
model_2.reset()
assert model.serialize() == model_2.serialize()
finally:
random.seed(prev_seed)
numpy.random.seed(prev_seed)
def test_mlp_jacobian_softmax_out_ce():
_check_jacobian(lambda s1, s2, s3: mlp.MLP((s1, s2, s3),
transfers=mlp.SoftmaxTransfer(),
error_func=CrossEntropy()))
def test_mlp_jacobian_lin_out_mse():
_check_jacobian(lambda s1, s2, s3: mlp.MLP(
(s1, s2, s3), transfers=mlp.LinearTransfer(), error_func=MSE()))
def test_mlp_obj_and_obj_jac_match_softmax_out_ce():
_check_obj_and_obj_jac_match(lambda s1, s2, s3: mlp.MLP(
(s1, s2, s3),
transfers=mlp.SoftmaxTransfer(),
error_func=CrossEntropy()),
classification=True)
def test_mlp_obj_and_obj_jac_match_softmax_out_mse():
_check_obj_and_obj_jac_match(lambda s1, s2, s3: mlp.MLP(
(s1, s2, s3), transfers=mlp.SoftmaxTransfer(), error_func=MSE()))
def test_mlp_obj_and_obj_jac_match_lin_out_mse():
_check_obj_and_obj_jac_match(lambda s1, s2, s3: mlp.MLP(
(s1, s2, s3), transfers=mlp.LinearTransfer(), error_func=MSE()))
def test_mlp_jacobian_softmax_out_mse():
_check_jacobian(lambda s1, s2, s3: mlp.MLP(
(s1, s2, s3), transfers=SoftmaxTransfer(), error_func=MeanSquaredError()))
def test_mlp_jacobian_relu_out_ce():
_check_jacobian(lambda s1, s2, s3: mlp.MLP(
(s1, s2, s3), transfers=mlp.ReluTransfer(), error_func=CrossEntropyError()))
