def test_setup_2_3_1():
nn = js_neural_network()
# data = [{imput: [], output: []}]
data = [{
'input': [1, 1],
'output': [0]
}, {
'input': [1, 0],
'output': [1]
}, {
'input': [0, 1],
'output': [1]
}, {
'input': [0, 0],
'output': [0]
}]
examples = nn.normalize(data)
nn.setup(examples=examples, default_weight=0.1)
assert len(nn.weights) == 2
np.testing.assert_array_equal(nn.weights[0],
np.array([[0.1, 0.1, 0.1], [0.1, 0.1, 0.1]]))
np.testing.assert_array_equal(nn.weights[1], np.array([[0.1], [0.1],
[0.1]]))
assert nn.weights[0].shape == (2, 3)
assert nn.weights[1].shape == (3, 1)
def test_or_2_3_3():
print('Testing ------- OR --------')
nn = js_neural_network() # (iterations=1)
# data = [{imput: [], output: []}]
data = [
{
'input': [1, 1, 1, 1],
'output': [1, 0, 0]
},
{
'input': [1, 1, 0, 1],
'output': [0, 1, 0]
},
{
'input': [1, 0, 0, 1],
'output': [0, 0, 1]
},
# { 'input': [0, 0], 'output': [0, 0, 0] }
]
# examples = nn.normalize(data)
# nn.setup(examples=examples, default_weight=0.1)
nn.learn(data) # , default_weight=0.1)
to_predict = [1, 1, 1, 1]
res = nn.predict(to_predict)
print('Prediction for {} = {}'.format(to_predict, res))
to_predict = [1, 1, 0, 1]
res = nn.predict(to_predict)
print('Prediction for {} = {}'.format(to_predict, res))
to_predict = [1, 0, 0, 1]
res = nn.predict(to_predict)
print('Prediction for {} = {}'.format(to_predict, res))
to_predict = [1, 0, 1, 1]
res = nn.predict(to_predict)
print('Prediction for {} = {}'.format(to_predict, res))
def test_train_ocr_121():
characters = alphabet_121()
nn = js_neural_network()
nn.learn(characters, default_weight=0.1)
for ch in altered_c:
res = nn.predict(ch['glyph'])
print('Prediction for {} = {}'.format(ch['character'], res))
def test_train_ocr():
characters = alphabet()
nn = js_neural_network(iterations=1) # ()
nn.learn(characters)
for ch in altered_c:
res = nn.predict(ch['glyph'])
print('Prediction for {} = {}'.format(ch['character'], res))
def test_wine():
X_train, y_train, X_test, y_test = get_data()
nn = gen_neural_network.js_neural_network(iterations=5000,
hidden_units=13,
learning_rate=0.001,
hidden_layers=1)
wine_data = {'input': np.array(X_train), 'output': np.array(y_train)}
nn.learn(wine_data, default_weight=None, normalize_data=True)
y_test = np.array(y_test)
count = 0
for i, x in enumerate(np.array(X_test)):
prediction = nn.predict(x)
if all(prediction == y_test[i]):
count += 1
else:
print('x: {} prediction: {} expected: {}'.format(
'x', prediction, y_test[i]))
print('correct {} out of {}'.format(count, i))
def test_setup_2_3_2_random_weight():
nn = js_neural_network()
# data = [{imput: [], output: []}]
data = [{
'input': [1, 1],
'output': [0, 1]
}, {
'input': [1, 0],
'output': [1, 0]
}, {
'input': [0, 1],
'output': [1, 0]
}, {
'input': [0, 0],
'output': [0, 1]
}]
examples = nn.normalize(data)
nn.setup(examples=examples)
assert len(nn.weights) == 2
assert nn.weights[0].shape == (3, 2)
assert nn.weights[1].shape == (2, 3)
def test_normalize():
nn = js_neural_network()
# data = [{imput: [], output: []}]
data = [{
'input': [1, 1],
'output': [0]
}, {
'input': [1, 0],
'output': [1]
}, {
'input': [0, 1],
'output': [1]
}, {
'input': [0, 0],
'output': [0]
}]
examples = nn.normalize(data)
assert len(examples) == 2
assert examples['input'] is not None
assert examples['output'] is not None
np.testing.assert_array_equal(examples['input'],
np.array([[1, 1], [1, 0], [0, 1], [0, 0]]))
np.testing.assert_array_equal(examples['output'],
np.array([[0], [1], [1], [0]]))
