def test_shuffle():
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
minibatches=1,
shuffle_init=True,
shuffle_epoch=False,
random_seed=1)
nn.fit(X_std, y)
y_pred = nn.predict(X_std)
acc = np.sum(y == y_pred, axis=0) / float(X_std.shape[0])
assert round(acc, 2) == 0.99, "Acc: %s" % acc
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
minibatches=1,
shuffle_init=True,
shuffle_epoch=True,
random_seed=1)
nn.fit(X_std, y)
y_pred = nn.predict(X_std)
acc = np.sum(y == y_pred, axis=0) / float(X_std.shape[0])
assert round(acc, 2) == 0.99, "Acc: %s" % acc
def test_shuffle():
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
minibatches=1,
shuffle_init=True,
shuffle_epoch=False,
random_seed=1)
nn.fit(X_std, y)
y_pred = nn.predict(X_std)
acc = np.sum(y == y_pred, axis=0) / float(X_std.shape[0])
assert round(acc, 2) == 0.99, "Acc: %s" % acc
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
minibatches=1,
shuffle_init=True,
shuffle_epoch=True,
random_seed=1)
nn.fit(X_std, y)
y_pred = nn.predict(X_std)
acc = np.sum(y == y_pred, axis=0) / float(X_std.shape[0])
assert round(acc, 2) == 0.99, "Acc: %s" % acc
def test_score_function():
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
minibatches=1,
shuffle_init=False,
shuffle_epoch=False,
random_seed=1)
nn.fit(X_std, y)
acc = nn.score(X_std, y)
assert round(acc, 2) == 0.99, "Acc: %s" % acc
def test_score_function():
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
minibatches=1,
shuffle_init=False,
shuffle_epoch=False,
random_seed=1)
nn.fit(X_std, y)
acc = nn.score(X_std, y)
assert round(acc, 2) == 0.99, "Acc: %s" % acc
def test_gradient_checking():
nn3 = NeuralNetMLP(n_output=len(np.unique(y)),
n_features=X_std.shape[1],
n_hidden=25,
l2=0.0,
l1=0.0,
epochs=1,
eta=0.01,
alpha=0.0,
decrease_const=0.0,
minibatches=1,
shuffle_init=False,
shuffle_epoch=False,
random_seed=1)
for epoch in range(10):
eucldist = nn3._gradient_checking(X=X_std, y=y)
assert eucldist < 1e-07, 'Gradient difference is %s' % eucldist
def test_gradient_descent():
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=10,
eta=0.01,
minibatches=1,
shuffle=True,
random_state=1)
nn.fit(X_std, y)
y_pred = nn.predict(X_std)
acc = np.sum(y == y_pred, axis=0) / float(X_std.shape[0])
#print(acc)
assert(round(acc, 2) == 0.87)
def test_gradient_checking():
nn3 = NeuralNetMLP(n_output=len(np.unique(y)),
n_features=X_std.shape[1],
n_hidden=25,
l2=0.0,
l1=0.0,
epochs=1,
eta=0.01,
alpha=0.0,
decrease_const=0.0,
minibatches=1,
shuffle_init=False,
shuffle_epoch=False,
random_seed=1)
for epoch in range(10):
eucldist = nn3._gradient_checking(X=X_std, y=y)
assert eucldist < 1e-07, 'Gradient difference is %s' % eucldist
def test_binary():
X0 = X_std[0:100] # class 0 and class 1
y0 = y[0:100] # class 0 and class 1
nn = NeuralNetMLP(n_output=2,
n_features=X0.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
minibatches=10,
shuffle_init=True,
shuffle_epoch=True,
random_seed=1)
nn.fit(X0, y0)
y_pred = nn.predict(X0)
acc = np.sum(y0 == y_pred, axis=0) / float(X0.shape[0])
assert round(acc, 2) == 1.0, "Acc: %s" % acc
def test_gradient_descent():
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
random_weights=[-1.0, 1.0],
minibatches=1,
shuffle_init=False,
shuffle_epoch=False,
random_seed=1)
nn.fit(X_std, y)
y_pred = nn.predict(X_std)
acc = np.sum(y == y_pred, axis=0) / float(X_std.shape[0])
assert(round(acc, 2) == 0.97)
def test_minibatch():
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=15,
alpha=2.0,
eta=0.05,
minibatches=10,
shuffle_init=True,
shuffle_epoch=False,
random_seed=1)
nn.fit(X_std, y)
y_pred = nn.predict(X_std)
acc = np.sum(y == y_pred, axis=0) / float(X_std.shape[0])
print(acc)
assert(round(acc, 2) == 0.97)
def test_binary():
X0 = X_std[0:100] # class 0 and class 1
y0 = y[0:100] # class 0 and class 1
nn = NeuralNetMLP(n_output=2,
n_features=X0.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
minibatches=10,
shuffle_init=True,
shuffle_epoch=True,
random_seed=1)
nn.fit(X0, y0)
y_pred = nn.predict(X0)
acc = np.sum(y0 == y_pred, axis=0) / float(X0.shape[0])
assert round(acc, 2) == 1.0, "Acc: %s" % acc
def test_minibatch():
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=15,
alpha=2.0,
eta=0.05,
minibatches=10,
shuffle_init=True,
shuffle_epoch=False,
random_seed=1)
nn.fit(X_std, y)
y_pred = nn.predict(X_std)
acc = np.sum(y == y_pred, axis=0) / float(X_std.shape[0])
print(acc)
assert (round(acc, 2) == 0.97)
def test_gradient_descent():
nn = NeuralNetMLP(n_output=3,
n_features=X.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=100,
eta=0.1,
random_weights=[-1.0, 1.0],
minibatches=1,
shuffle_init=False,
shuffle_epoch=False,
random_seed=1)
nn.fit(X_std, y)
y_pred = nn.predict(X_std)
acc = np.sum(y == y_pred, axis=0) / float(X_std.shape[0])
assert (round(acc, 2) == 0.97)
def test_binary():
X0 = X_std[0:100] # class 0 and class 1
y0 = y[0:100] # class 0 and class 1
nn = NeuralNetMLP(n_output=1,
n_features=X0.shape[1],
n_hidden=10,
l2=0.0,
l1=0.0,
epochs=10,
eta=0.01,
minibatches=10,
shuffle=True,
random_state=1)
nn.fit(X0, y0)
y_pred = nn.predict(X0)
acc = np.sum(y0 == y_pred, axis=0) / float(X0.shape[0])
#print(y_pred)
assert(round(acc, 2) == 1.0)
X, y = iris_data()
X = X[:, [0, 3]]
X = X[0:100]
y = y[0:100]
X[:, 0] = (X[:, 0] - X[:, 0].mean()) / X[:, 0].std()
X[:, 1] = (X[:, 1] - X[:, 1].mean()) / X[:, 1].std()
nn = NeuralNetMLP(n_output=len(np.unique(y)),
n_features=X.shape[1],
n_hidden=50,
l2=0.00,
l1=0.0,
epochs=300,
eta=0.01,
alpha=0.0,
decrease_const=0.0,
minibatches=1,
shuffle_init=False,
shuffle_epoch=False,
random_seed=1,
print_progress=3)
nn = nn.fit(X, y)
y_pred = nn.predict(X)
acc = np.sum(y == y_pred, axis=0) / X.shape[0]
print('Accuracy: %.2f%%' % (acc * 100))
plot_decision_regions(X, y, clf=nn, legend=2)
X, y = iris_data()
X = X[:, [0, 3]]
X = X[0:100]
y = y[0:100]
X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()
X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()
nn = NeuralNetMLP(n_output = len(np.unique(y)),
n_features = X.shape[1],
n_hidden = 50,
l2 = 0.00,
l1 = 0.0,
epochs = 300,
eta = 0.01,
alpha = 0.0,
decrease_const = 0.0,
minibatches = 1,
shuffle_init = False,
shuffle_epoch = False,
random_seed = 1,
print_progress = 3)
nn = nn.fit(X, y)
y_pred = nn.predict(X)
acc = np.sum(y == y_pred, axis=0) / X.shape[0]
print('Accuracy: %.2f%%' % (acc * 100))
