tensorflow.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0, array([ 0.61638772], dtype=float32), array([ 0.02258076], dtype=float32), 0.02007756)\n",
      "(20, array([ 0.25472665], dtype=float32), array([ 0.21643446], dtype=float32), 0.001806676)\n",
      "(40, array([ 0.14641654], dtype=float32), array([ 0.27493113], dtype=float32), 0.0001625907)\n",
      "(60, array([ 0.11392454], dtype=float32), array([ 0.29247957], dtype=float32), 1.4632288e-05)\n",
      "(80, array([ 0.10417723], dtype=float32), array([ 0.29774395], dtype=float32), 1.3168243e-06)\n",
      "(100, array([ 0.10125311], dtype=float32), array([ 0.29932323], dtype=float32), 1.1850291e-07)\n",
      "(120, array([ 0.10037591], dtype=float32), array([ 0.299797], dtype=float32), 1.0663627e-08)\n",
      "(140, array([ 0.10011276], dtype=float32), array([ 0.2999391], dtype=float32), 9.5966568e-10)\n",
      "(160, array([ 0.10003382], dtype=float32), array([ 0.29998174], dtype=float32), 8.6334516e-11)\n",
      "(180, array([ 0.10001013], dtype=float32), array([ 0.29999453], dtype=float32), 7.7622622e-12)\n",
      "(200, array([ 0.10000303], dtype=float32), array([ 0.29999837], dtype=float32), 6.9795726e-13)\n"
     ]
    }
   ],
   "source": [
    "# 线性回归示例\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "\n",
    "# 根据真实关系创建100个样本  y = x * 0.1 + 0.3\n",
    "x_data = np.random.rand(100).astype(np.float32)\n",
    "y_data = x_data * 0.1 + 0.3\n",
    "\n",
    "# 其中截距项b和斜率项W是待估计的参数 y_data = W * x_data + b\n",
    "# 需要先定义参数，并给初始值\n",
    "W = tf.Variable(tf.random_uniform([1], -1.0, 1.0))\n",
    "b = tf.Variable(tf.zeros([1]))\n",
    "# 定义参数和变量间的关系\n",
    "y = W * x_data + b\n",
    "\n",
    "# 定义均方误做为损失函数，后面它是最优化的目标函数\n",
    "# reduce_mean是对一个tensor求均值并坍缩为一个常数\n",
    "loss = tf.reduce_mean(tf.square(y - y_data))\n",
    "# 定义优化器，参数0.5为学习率\n",
    "optimizer = tf.train.GradientDescentOptimizer(0.5)\n",
    "# 定义优化器的最小化运算目标\n",
    "train = optimizer.minimize(loss)\n",
    "\n",
    "# 初始化所有变量\n",
    "init = tf.initialize_all_variables()\n",
    "\n",
    "# 启动整个运算流程\n",
    "sess = tf.Session()\n",
    "sess.run(init)\n",
    "\n",
    "#  通过循环迭代计算\n",
    "for step in xrange(201):\n",
    "    sess.run(train)\n",
    "    if step % 20 == 0:\n",
    "        print(step, sess.run(W), sess.run(b), sess.run(loss))\n",
    "\n",
    "# Learns best fit is W: [0.1], b: [0.3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "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": [
    "# 数字图像识别示例--用多分类逻辑回归方法\n",
    "# 读取内置数据集\n",
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "mnist = input_data.read_data_sets('MNIST_data', one_hot=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "sess = tf.InteractiveSession()\n",
    "# 定义X和Y数据\n",
    "x = tf.placeholder(tf.float32, shape=[None, 784])\n",
    "y_ = tf.placeholder(tf.float32, shape=[None, 10])\n",
    "# 定义参数\n",
    "W = tf.Variable(tf.zeros([784,10]))\n",
    "b = tf.Variable(tf.zeros([10]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 初始化\n",
    "sess.run(tf.initialize_all_variables())\n",
    "# 定义Y的生成过程，x和w矩阵相乘后通过softmax映射M\n",
    "y = tf.nn.softmax(tf.matmul(x,W) + b)\n",
    "# 定义交叉熵\n",
    "cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 定义优化器\n",
    "train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# 开始运算，每50个样本做一批喂入算法\n",
    "for i in range(1000):\n",
    "  batch = mnist.train.next_batch(50)\n",
    "  train_step.run(feed_dict={x: batch[0], y_: batch[1]})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9171\n"
     ]
    }
   ],
   "source": [
    "# 计算正确率\n",
    "correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1))\n",
    "accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))\n",
    "print(accuracy.eval(feed_dict={x: mnist.test.images, y_: mnist.test.labels}))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}