def test_train_labels_no_labels(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 2)
data = [[1, 1, 1, 1, 1], [1, 1, 1, 1, 1]]
labels = []
with self.assertRaises(ValueError):
d.train_labels(data, labels, 1, 1)
def test_train_labels_no_labels(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 2)
data = [[1, 1, 1, 1, 1], [1, 1, 1, 1, 1]]
labels = []
with self.assertRaises(ValueError):
d.train_labels(data, labels, 1, 1)
def test_get_topology_constructs(self):
layer_sizes = [144, 50, 50, 2000]
number_labels = 1
d = DBN(layer_sizes, number_labels)
layer_sizes.append(number_labels)
self.assertListEqual(d.get_topology(), layer_sizes)
def test_rbms_list_setup_correct(self):
layer_sizes = [144, 50, 50, 2000]
d = DBN(layer_sizes, 1)
self.assertEquals(len(d._rbms), d.number_layers)
def test_number_inputs_setup_correct(self):
layer_sizes = [144, 50, 50, 2000]
d = DBN(layer_sizes, 1)
self.assertEquals(d.number_inputs, 144)
def test_empty_config_constructor(self):
layer_sizes = []
with self.assertRaises(IndexError):
d = DBN(layer_sizes, 1)
def test_classify_wrong_shape(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 2)
data = [[1, 1, 1, 1]]
with self.assertRaises(ValueError):
probs = d.classify(data)
def test_dbn_constructor(self):
layer_sizes = [10, 20]
d = DBN(layer_sizes, 1)
def test_pre_train_negative_batch(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 10)
data = [[1, 1, 1, 1, 1], [1, 1, 1, 1, 1]]
with self.assertRaises(ValueError):
d.pre_train(data, 1, -1)
def test_classify(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 2)
data = [[1, 1, 1, 1, 1]]
probs = d.classify(data)
def test_train_labels(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 2)
data = [[1, 1, 1, 1, 1], [1, 1, 1, 1, 1]]
labels = [[1, 0], [0, 1]]
d.train_labels(data, labels, 1, 1)
def test_pre_train_negative_batch(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 10)
data = [[1, 1, 1, 1, 1], [1, 1, 1, 1, 1]]
with self.assertRaises(ValueError):
d.pre_train(data, 1, -1)
def test_pre_train_wrong_data_size(self):
layer_sizes = [5, 20]
d = DBN(layer_sizes, 10)
data = [[1, 1, 1, 1, 1, 1, 1]]
with self.assertRaises(ValueError):
d.pre_train(data, 1, 1)
def test_pre_train_correct_data_size(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 10)
data = [[1, 1, 1, 1, 1], [1, 1, 1, 1, 1]]
d.pre_train(data, 1, 1)
def test_get_topology_constructs(self):
layer_sizes = [144, 50, 50, 2000]
number_labels = 1
d = DBN(layer_sizes, number_labels)
layer_sizes.append(number_labels)
self.assertListEqual(d.get_topology(), layer_sizes)
def test_pre_train_correct_data_size(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 10)
data = [[1, 1, 1, 1, 1], [1, 1, 1, 1, 1]]
d.pre_train(data, 1, 1)
def test_pre_train_wrong_data_size(self):
layer_sizes = [5, 20]
d = DBN(layer_sizes, 10)
data = [[1, 1, 1, 1, 1, 1, 1]]
with self.assertRaises(ValueError):
d.pre_train(data, 1, 1)
def test_classify_wrong_shape(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 2)
data = [[1, 1, 1, 1]]
with self.assertRaises(ValueError):
probs = d.classify(data)
def test_train_labels(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 2)
data = [[1, 1, 1, 1, 1], [1, 1, 1, 1, 1]]
labels = [[1, 0], [0, 1]]
d.train_labels(data, labels, 1, 1)
def test_one_layer_constructor(self):
layer_sizes = [10]
d = DBN(layer_sizes, 1)
def test_negative_dimension_constructor(self):
layer_sizes = [-10, 500]
with self.assertRaises(ValueError):
d = DBN(layer_sizes, 1)
def test_classify(self):
layer_sizes = [5, 20, 10]
d = DBN(layer_sizes, 2)
data = [[1, 1, 1, 1, 1]]
probs = d.classify(data)
