12. 手写数字识别准确率输出.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "\n",
    "INPUT_NODE = 784\n",
    "OUTPUT_NODE = 10\n",
    "LAYER1_NODE = 500\n",
    "\n",
    "def forward(x, regularizer):\n",
    "    w1 = get_weight([INPUT_NODE, LAYER1_NODE], regularizer)\n",
    "    b1 = get_bias([LAYER1_NODE])\n",
    "    w2 = get_weight([LAYER1_NODE, OUTPUT_NODE], regularizer)\n",
    "    b2 = get_bias([OUTPUT_NODE])\n",
    "    \n",
    "    a = tf.nn.relu(tf.matmul(x, w1) + b1)\n",
    "    y = tf.matmul(a, w2) + b2\n",
    "    return y\n",
    "\n",
    "def get_weight(shape, regularizer):\n",
    "    w = tf.Variable(tf.truncated_normal(shape, stddev=0.1), dtype=tf.float32)\n",
    "    if regularizer != None:\n",
    "        tf.add_to_collection(\"losses\", tf.contrib.layers.l2_regularizer(regularizer)(w))\n",
    "    return w\n",
    "\n",
    "def get_bias(shape):\n",
    "    b = tf.Variable(tf.zeros(shape))\n",
    "    return b"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From D:\\soft\\Ana\\lib\\site-packages\\tensorflow\\contrib\\learn\\python\\learn\\datasets\\base.py:198: retry (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use the retry module or similar alternatives.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import os\n",
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "import code_12_mnist_forward\n",
    "\n",
    "STEPS = 50000\n",
    "BATCH_SIZE = 200\n",
    "REGULARIZER = 0.0001\n",
    "LEARNING_RATE_BASE = 0.1\n",
    "LEARNING_RATE_DECAY = 0.99\n",
    "MOVING_AVERAGE_DECAY = 0.99\n",
    "MODEL_SAVE_PATH = \"./model/\"\n",
    "MODEL_NAME = \"mnist_model\"\n",
    "\n",
    "def backward(mnist):\n",
    "    x = tf.placeholder(tf.float32, shape=(None, code_12_mnist_forward.INPUT_NODE))\n",
    "    y = tf.placeholder(tf.float32, shape=(None, code_12_mnist_forward.OUTPUT_NODE))\n",
    "    \n",
    "    y_hat = code_12_mnist_forward.forward(x, REGULARIZER)\n",
    "    global_step = tf.Variable(0, trainable=False)\n",
    "    loss = tf.reduce_mean(tf.nn.sparse_softmax_cross_entropy_with_logits(logits=y_hat, labels=tf.argmax(y, 1)))\n",
    "    loss_total = loss + tf.add_n(tf.get_collection(\"losses\"))\n",
    "    \n",
    "    learning_rate = tf.train.exponential_decay(LEARNING_RATE_BASE, \n",
    "                                               global_step, \n",
    "                                               mnist.train.num_examples / BATCH_SIZE, \n",
    "                                               LEARNING_RATE_DECAY, \n",
    "                                               staircase=True)\n",
    "    \n",
    "    train_step = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss_total, global_step=global_step)\n",
    "    \n",
    "    ema = tf.train.ExponentialMovingAverage(MOVING_AVERAGE_DECAY, global_step)\n",
    "    ema_op = ema.apply(tf.trainable_variables())\n",
    "    with tf.control_dependencies([train_step, ema_op]):\n",
    "        train_op = tf.no_op(name=\"train\")\n",
    "        \n",
    "    saver = tf.train.Saver()\n",
    "    \n",
    "    with tf.Session() as sess:\n",
    "        init = tf.global_variables_initializer()\n",
    "        sess.run(init)\n",
    "        for i in range(STEPS):\n",
    "            xs, ys = mnist.train.next_batch(BATCH_SIZE)\n",
    "            _, loss_val, step = sess.run([train_op, loss_total, global_step], feed_dict={x:xs, y:ys})\n",
    "            if i % 1000 == 0:\n",
    "                print(\"After %6d steps, loss is %f\" % (step, loss_val))\n",
    "                saver.save(sess, os.path.join(MODEL_SAVE_PATH, MODEL_NAME), global_step=global_step)\n",
    "                \n",
    "def main():\n",
    "    mnist = input_data.read_data_sets(\"../data/\", one_hot=True)\n",
    "    backward(mnist)\n",
    "    \n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import time\n",
    "import tensorflow as tf\n",
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "import code_12_mnist_forward\n",
    "import code_12_mnist_backward\n",
    "\n",
    "TEST_INTERVEL_SEC = 10\n",
    "\n",
    "def test(mnist):\n",
    "    with tf.Graph().as_default() as g:  # 使用tf.Graph()复现计算图\n",
    "        x = tf.placeholder(tf.float32, [None, code_12_mnist_forward.INPUT_NODE])\n",
    "        y = tf.placeholder(tf.float32, [None, code_12_mnist_forward.OUTPUT_NODE])\n",
    "        y_hat = code_12_mnist_forward.forward(x, None)\n",
    "        \n",
    "        ema = tf.train.ExponentialMovingAverage(code_12_mnist_backward.MOVING_AVERAGE_DECAY)\n",
    "        ema_restore = ema.variables_to_restore()\n",
    "        saver = tf.train.Saver(ema_restore)\n",
    "        \n",
    "        correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_hat, 1))\n",
    "        accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))\n",
    "        \n",
    "        while True:\n",
    "            with tf.Session() as sess:\n",
    "                ckpt = tf.train.get_checkpoint_state(code_12_mnist_backward.MODEL_SAVE_PATH)\n",
    "                if ckpt and ckpt.model_checkpoint_path:\n",
    "                    saver.restore(sess, ckpt.model_checkpoint_path) # 恢复模型到当前会话\n",
    "                    global_step = ckpt.model_checkpoint_path.split(\"/\")[-1].split(\"-\")[-1]\n",
    "                    accuracy_val = sess.run(accuracy, feed_dict={x:mnist.test.images, y:mnist.test.labels})\n",
    "                    print(\"After training %s steps, test accuracy is %f\" % (global_step, accuracy_val))\n",
    "                else:\n",
    "                    print(\"No checkpoint file find\")\n",
    "                    return\n",
    "            time.sleep(TEST_INTERVEL_SEC)\n",
    "            \n",
    "def main():\n",
    "    mnist = input_data.read_data_sets(\"../data/\", one_hot=True)\n",
    "    test(mnist)\n",
    "    \n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}