def test_predict_different_inputs(self):
ojanet = algorithms.Oja(minimized_data_size=1,
verbose=False,
step=0.01)
data = np.array([[1, 2, 3]]).T
target = np.array([[1, 2, 3]]).T
ojanet.train(data, epsilon=0.01, epochs=100)
self.assertInvalidVectorPred(ojanet, data.ravel(), target, decimal=2)
def test_oja_minimization(self):
ojanet = algorithms.Oja(minimized_data_size=1,
step=0.01,
weights=np.ones((2, 1)) * 0.1,
verbose=False)
ojanet.train(self.data, epsilon=1e-5)
minimized_data = ojanet.predict(self.data)
self.assertTrue(np.all(np.round(minimized_data, 2) == self.result))
reconstructed = ojanet.reconstruct(minimized_data)
self.assertTrue(np.allclose(reconstructed, self.data))
def test_reconstruct_different_inputs(self):
ojanet = algorithms.Oja(minimized_data_size=1,
verbose=False,
step=0.01)
data = np.array([[1, 2, 3]]).T
target = np.array([[1, 2, 3]]).T
input_vector = data.ravel()
ojanet.train(data, epsilon=0.01, epochs=100)
test_vectors = vectors_for_testing(input_vector)
for i, test_vector in enumerate(test_vectors, start=1):
np.testing.assert_array_almost_equal(
ojanet.reconstruct(test_vector), target, decimal=1)
def test_oja_minimization(self):
ojanet = algorithms.Oja(minimized_data_size=1,
step=0.01,
weights=np.ones((2, 1)) * 0.1,
verbose=False)
ojanet.train(self.data, epsilon=1e-5, epochs=100)
minimized_data = ojanet.predict(self.data)
np.testing.assert_array_almost_equal(minimized_data,
self.result,
decimal=2)
reconstructed = ojanet.reconstruct(minimized_data)
np.testing.assert_array_almost_equal(reconstructed,
self.data,
decimal=3)
def test_oja_exceptions(self):
ojanet = algorithms.Oja(minimized_data_size=1, step=0.01,
verbose=False)
with self.assertRaises(ValueError):
# Can't reconstruct without training
ojanet.reconstruct(np.random.random((4, 1)))
with self.assertRaises(ValueError):
# Can't predict without training
ojanet.predict(np.random.random((4, 1)))
ojanet.train(self.data, epsilon=1e-5)
with self.assertRaises(ValueError):
# Invalid #feature for reconstruct
ojanet.reconstruct(np.random.random((3, 3)))
with self.assertRaises(ValueError):
# Invalid #feature for train
ojanet.train(np.random.random((4, 10)))
def test_train_different_inputs(self):
self.assertInvalidVectorTrain(
algorithms.Oja(minimized_data_size=1, verbose=False, step=0.01),
np.array([1, 2, 3]))
a = oya(w_start, y_start1, iris_data, 4)
a = np.array(a)
check = matrixmult(np.array(a[0]), np.array(a[1]))
check = np.array(check)
iris_data = np.array(iris_data)
check_t = trans(check)
print('final results(iris)')
print(check_t)
print('iris data')
print(iris_data)
print('decompressing(iris)')
print(check_t - iris_data)
iris_data = np.array(iris_data)
ojanet = algorithms.Oja(minimized_data_size=1, step=0.01, verbose=False)
ojanet.train(iris_data, epochs=100)
minimized = ojanet.predict(iris_data)
print('Minimized(iris, neupy)', minimized)
print('Reconstruct(iris)', ojanet.reconstruct(minimized))
wine_data = np.array(wine_data)
ojanet = algorithms.Oja(minimized_data_size=1, step=0.01, verbose=False)
ojanet.train(wine_data, epochs=100)
minimized = ojanet.predict(wine_data)
print('Minimized(wine, neupy)', minimized)
print('Reconstruct(wine)', ojanet.reconstruct(minimized))
