import tensorflow as tf

Input = tf.placeholder('float',shape=[None, 2], name="Input")
Target = tf.placeholder('float',shape=[None, 1], name="Target")
inputBias = tf.Variable(initial_value=tf.random_normal(shape=[3], stddev=0.4), dtype='float', name="input_bias")

weights = tf.Variable(initial_value=tf.random_normal(shape=[2,3], stddev=0.4), dtype='float', name="input_weights")
hiddenBias = tf.Variable(initial_value=tf.random_normal(shape=[1], stddev=0.4), dtype='float', name="hidden_bias")

outputWeights = tf.Variable(initial_value=tf.random_normal(shape=[3,1], stddev=0.4), dtype='float',
                            name="output_weights")

hiddenLayer = tf.matmul(Input, weights) + inputBias
hiddenLayer = tf.sigmoid(hiddenLayer, name="hidden_layer_activation")

output = tf.matmul(hiddenLayer, outputWeights) + hiddenBias
hiddenLayer = tf.sigmoid(output, name="output_layer_activation")

#or output = tf.sigmoid(output, name="output_layer_activation")


cost = tf.squared_difference(Target, output)
cost = tf.reduce_mean(cost)
optimizer = tf.train.AdamOptimizer().minimize(cost)

inp = [[1, 1],[1, 0],[0, 1],[0, 0]]
out = [[0],[1],[1],[0]]
epochs= 4000

with tf.Session() as sess:
    tf.global_variables_initializer().run()
    for i in range(epochs):
        err, _ = sess.run([cost,optimizer],feed_dict={Input:inp, Target:out})
        print(i,err)