h2o_test_opti.py

import tensorflow as tf
import numpy as np
import math


# input
X = tf.placeholder(tf.float32, [None, 2], name="X")
Y = tf.placeholder(tf.float32, [None, 1], name="Y")

# hidden
beta = tf.get_variable("beta", shape=[2], initializer=tf.contrib.layers.xavier_initializer())
powered = tf.pow(X,beta)
productLayer = tf.contrib.keras.layers.multiply(powered)

# output
w_o = tf.get_variable("w_o", shape=[2], initializer=tf.contrib.layers.xavier_initializer())
b_o = tf.get_variable("bias", shape=[1], initializer=tf.zeros([1]))

output = tf.add(tf.matmul(productLayer,w_o), b_o)
loss = tf.reduce_sum(tf.square(output - Y)) # tf.nn.l2_loss(yhat - Y)

training_op = tf.train.AdamOptimizer(learning_rate=0.01).minimize(loss)

## TRAINING PROCEDURE
h2o = lambda h,o: h**2 * o

trainIn = np.random.rand(10000, 2)
trainOut = h2o(trainIn)
validIn = np.random.rand(4000, 2)
validOut = h2o(validIn)

trainOut = np.reshape(trainOut, (-1, 1))
validOut = np.reshape(validOut, (-1, 1))

mse = math.inf
sess = tf.Session()
sess.run(tf.global_variables_initializer())
for i in range(1000):
    if i % 100 == 0:
        print("epoch {}, validation MSE {}".format((i + 1), mse))
    _, mse = sess.run([training_op, loss], feed_dict={X: trainIn, Y: trainOut})