{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow.examples.tutorials.mnist import input_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting MNIST_data/train-images-idx3-ubyte.gz\n",
      "Extracting MNIST_data/train-labels-idx1-ubyte.gz\n",
      "Extracting MNIST_data/t10k-images-idx3-ubyte.gz\n",
      "Extracting MNIST_data/t10k-labels-idx1-ubyte.gz\n"
     ]
    }
   ],
   "source": [
    "mnist = input_data.read_data_sets(\"MNIST_data/\", one_hot=True)\n",
    "\n",
    "# 输入图片 28*28 个像素\n",
    "n_inputs = 28      # 输入的每行有 28 个数据，输入层 神经元的个数\n",
    "max_time = 28    # 输入的次数为 28 次\n",
    "lstm_size = 100   # 隐藏层 block 单元\n",
    "n_classes = 10     # 分类个数\n",
    "batch_size = 50   # 单批次的样本数量\n",
    "n_batch = mnist.train.num_examples / batch_size   # 一共会分成多少批次"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "x = tf.placeholder(tf.float32, [None, 784])\n",
    "y = tf.placeholder(tf.float32, [None, 10])\n",
    "weights = tf.Variable(tf.truncated_normal([lstm_size, n_classes], stddev=0.1))\n",
    "biases = tf.Variable(tf.constant(0.1, shape=[n_classes]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def LSTM(x, weights, biases):\n",
    "    inputs = tf.reshape(x, [-1, max_time, n_inputs])\n",
    "    # 定义隐藏层 block 单元\n",
    "    lstm_cell = tf.contrib.rnn.BasicLSTMCell(lstm_size)\n",
    "    # final_state[0]: cell state\n",
    "    # final_state[1]: hidden_state\n",
    "    outputs, final_state = tf.nn.dynamic_rnn(lstm_cell, inputs, dtype=tf.float32)\n",
    "    return tf.nn.softmax(tf.matmul(final_state[1], weights) + biases)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "prediction = LSTM(x, weights, biases)\n",
    "cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=prediction, labels=y))\n",
    "train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)\n",
    "correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(prediction, 1))\n",
    "accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))\n",
    "init = tf.global_variables_initializer()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iterator: 0 , Accuracy: 0.7244\n",
      "Iterator: 10 , Accuracy: 0.946\n",
      "Iterator: 20 , Accuracy: 0.9646\n",
      "Iterator: 30 , Accuracy: 0.9696\n",
      "Iterator: 40 , Accuracy: 0.9717\n",
      "Iterator: 50 , Accuracy: 0.9764\n",
      "Iterator: 60 , Accuracy: 0.9776\n",
      "Iterator: 70 , Accuracy: 0.9801\n",
      "Iterator: 80 , Accuracy: 0.9814\n",
      "Iterator: 90 , Accuracy: 0.9793\n",
      "Iterator: 100 , Accuracy: 0.9814\n"
     ]
    }
   ],
   "source": [
    "with tf.Session() as sess:\n",
    "    sess.run(init)\n",
    "    for epoch in range(101):\n",
    "        for batch in range(int(n_batch)):\n",
    "            batch_xs, batch_ys = mnist.train.next_batch(batch_size)\n",
    "            sess.run(train_step, feed_dict={x: batch_xs, y: batch_ys})\n",
    "        if epoch % 10 == 0:\n",
    "            acc = sess.run(accuracy, feed_dict={x:mnist.test.images, y:mnist.test.labels})\n",
    "            print(\"Iterator:\", str(epoch), \", Accuracy:\", str(acc))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}