def test_outputs_and():
gate = 'and'
learning_rate = 0.1
# error = 10/float(epochs)
error = .2
accuracy = 95
expected_inputs = np.array([[0, 0],
[0, 1],
[1, 0],
[1, 1]])
expected_outputs = np.array([[0, 0, 0, 1]]).T
gate_ = LogicGate(gate, 'tanh')
gate_.evolved_train(accuracy, learning_rate)
i = 0
for elem in expected_inputs:
pr_out = gate_.madame_irma.evolved_predict(elem)
print(elem, ' ---> ', pr_out[0])
assert(pr_out[0] > expected_outputs[i][0] - error and pr_out[0] < expected_outputs[i][0] + error)
i += 1
def test_outputs_and():
gate = 'and'
learning_rate = 0.1
error = .25
epochs = 100000
mode = 'perceptron'
expected_inputs = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])
expected_outputs = np.array([[0, 0, 0, 1]]).T
gate_ = LogicGate(gate, 'tanh')
gate_.train(epochs, learning_rate, mode)
i = 0
for elem in expected_inputs:
pr_out = gate_.madame_irma.predict(elem)
print(elem, ' ---> ', pr_out[0])
assert (pr_out[0] > expected_outputs[i][0] - error
and pr_out[0] < expected_outputs[i][0] + error)
i += 1
def test_outputs_xor():
gate = 'xor'
learning_rate = 0.1
error = .34
epochs = 100000
mode = 'network'
expected_inputs = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])
expected_outputs = np.array([[0, 1, 1, 0]]).T
gate_ = LogicGate(gate, 'sigmoid')
gate_.train(epochs, learning_rate, mode)
i = 0
for elem in expected_inputs:
pr_out = gate_.madame_irma.predict(elem)
print(elem, ' ---> ', pr_out[0])
assert (pr_out[0] > expected_outputs[i][0] - error
and pr_out[0] < expected_outputs[i][0] + error)
i += 1