Using Tensorflow and Keras to approximate an XOR

keras_xor.py

import numpy as np
from tensorflow.contrib.keras.api.keras.models import Sequential, Model
from tensorflow.contrib.keras.api.keras.layers import Dense, Activation, Dropout
from tensorflow.contrib.keras.api.keras.optimizers import SGD

model = Sequential()
# tanh activation allows [-1,1]
model.add(Dense(6, input_dim=2, activation='tanh'))
# hard_sigmoid will converge much faster than sigmoid with this example
model.add(Dense(1, activation='hard_sigmoid'))

sgd = SGD(lr=0.25)
model.compile(optimizer=sgd, loss='binary_crossentropy')

X = np.array([[0,0], [1,0], [0,1], [1,1]])
y = np.array([[0], [1], [1], [0]])

# Only need 100 epochs using hard_sigmoid and it converges to an exact solution!
model.fit(X, y, batch_size=1, epochs=100, verbose=1)
outputs = model.predict(X)

for input, output in zip(X, outputs):
    print("{} => {}".format(input, output))

# [0 0] => [ 0.]
# [1 0] => [ 1.]
# [0 1] => [ 1.]
# [1 1] => [ 0.]