Merge pull request #11 from tnakae/ShuffleTrainingData

import tensorflow as tf

import numpy as np

from sklearn.preprocessing import StandardScaler

from sklearn.externals import joblib

        with tf.Graph().as_default() as graph:

            self.graph = graph

            tf.set_random_seed(self.seed)

            np.random.seed(seed=self.seed)

            # Create Placeholder

            self.input = input = tf.placeholder(

            self.sess.run(init)

            # Training

            idx = np.arange(x.shape[0])

            np.random.shuffle(idx)

            for epoch in range(self.epoch_size):

                for batch in range(n_batch):

                    i_start = batch * self.minibatch_size

                    i_end = (batch + 1) * self.minibatch_size

                    x_batch = x[i_start:i_end]

                    x_batch = x[idx[i_start:i_end]]

                    self.sess.run(minimizer, feed_dict={

                        input:x_batch, drop:self.est_dropout_ratio})