def test__back_propagation(self):
data = [[1, 1, 2]]
target = [1, 0, 0]
nn = NeuralNetwork()
nn.create_network(len(data[0]), 2, (1,), weights=[[-.1, .2, .1, -.4], [-.15, -.2, .3]])
r = nn._feed_forward(data[0])
uw = nn._back_propagation(r, target[0])
def test__back_propagation(self):
data = [[1, 1, 2]]
target = [1, 0, 0]
nn = NeuralNetwork()
nn.create_network(len(data[0]),
2, (1, ),
weights=[[-.1, .2, .1, -.4], [-.15, -.2, .3]])
r = nn._feed_forward(data[0])
uw = nn._back_propagation(r, target[0])
def run_iris():
# Load iris data set
iris = datasets.load_iris()
n_inputs = len(iris.data[0])
n_outputs = len(iris.target_names)
network = NeuralNetwork()
network.create_network(n_inputs, n_outputs, (3, 4))
data_scaled = preprocessing.scale(iris.data)
print("Their neural network results: {}".format(cross_val_score(nn, data_scaled, iris.target, 3)))
print("My neural network results: {}".format(cross_val_score(network, data_scaled, iris.target, 3)))
def run_diabetes():
data = []
target = []
# Read data
with open('pima-indians-diabetes.data') as diabetes_file:
diabetes_reader = csv.reader(diabetes_file, quoting=csv.QUOTE_NONNUMERIC)
for row in diabetes_reader:
data.append(row[:8])
target.append(int(row[8]))
n_inputs = len(data[0])
n_outputs = len(set(target))
network = NeuralNetwork()
network.create_network(n_inputs, n_outputs, (3, 4))
data_scaled = preprocessing.scale(np.array(data))
print("Neural network results accuracy: {}".format(cross_val_score(network, data_scaled, np.array(target), 3)))
