04_CNN_advances/cnn_mnist_simple.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 卷积神经网络示例与各层可视化"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "当前TensorFlow版本为 [1.3.0]\n",
      "所有包载入完毕\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "%matplotlib inline  \n",
    "print (\"当前TensorFlow版本为 [%s]\" % (tf.__version__))\n",
    "print (\"所有包载入完毕\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 载入 MNIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting data/train-images-idx3-ubyte.gz\n",
      "Extracting data/train-labels-idx1-ubyte.gz\n",
      "Extracting data/t10k-images-idx3-ubyte.gz\n",
      "Extracting data/t10k-labels-idx1-ubyte.gz\n",
      "MNIST ready\n"
     ]
    }
   ],
   "source": [
    "mnist = input_data.read_data_sets('data/', one_hot=True)\n",
    "trainimg   = mnist.train.images\n",
    "trainlabel = mnist.train.labels\n",
    "testimg    = mnist.test.images\n",
    "testlabel  = mnist.test.labels\n",
    "print (\"MNIST ready\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 定义模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NETWORK READY\n"
     ]
    }
   ],
   "source": [
    "# NETWORK TOPOLOGIES\n",
    "n_input    = 784\n",
    "n_channel  = 64 \n",
    "n_classes  = 10  \n",
    "\n",
    "# INPUTS AND OUTPUTS\n",
    "x = tf.placeholder(\"float\", [None, n_input])\n",
    "y = tf.placeholder(\"float\", [None, n_classes])\n",
    "    \n",
    "# NETWORK PARAMETERS\n",
    "stddev = 0.1\n",
    "weights = {\n",
    "    'c1': tf.Variable(tf.random_normal([7, 7, 1, n_channel], stddev=stddev)),\n",
    "    'd1': tf.Variable(tf.random_normal([14*14*64, n_classes], stddev=stddev))\n",
    "}\n",
    "biases = {\n",
    "    'c1': tf.Variable(tf.random_normal([n_channel], stddev=stddev)),\n",
    "    'd1': tf.Variable(tf.random_normal([n_classes], stddev=stddev))\n",
    "}\n",
    "print (\"NETWORK READY\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 定义图结构"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "FUNCTIONS READY\n"
     ]
    }
   ],
   "source": [
    "# MODEL\n",
    "def CNN(_x, _w, _b):\n",
    "    # RESHAPE\n",
    "    _x_r = tf.reshape(_x, shape=[-1, 28, 28, 1])\n",
    "    # CONVOLUTION\n",
    "    _conv1 = tf.nn.conv2d(_x_r, _w['c1'], strides=[1, 1, 1, 1], padding='SAME')\n",
    "    # ADD BIAS\n",
    "    _conv2 = tf.nn.bias_add(_conv1, _b['c1'])\n",
    "    # RELU\n",
    "    _conv3 = tf.nn.relu(_conv2)\n",
    "    # MAX-POOL\n",
    "    _pool  = tf.nn.max_pool(_conv3, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')\n",
    "    # VECTORIZE\n",
    "    _dense = tf.reshape(_pool, [-1, _w['d1'].get_shape().as_list()[0]])\n",
    "    # DENSE\n",
    "    _logit = tf.add(tf.matmul(_dense, _w['d1']), _b['d1'])\n",
    "    _out = {\n",
    "        'x_r': _x_r, 'conv1': _conv1, 'conv2': _conv2, 'conv3': _conv3\n",
    "        , 'pool': _pool, 'dense': _dense, 'logit': _logit\n",
    "    }\n",
    "    return _out\n",
    "\n",
    "# PREDICTION\n",
    "cnnout = CNN(x, weights, biases)\n",
    "\n",
    "# LOSS AND OPTIMIZER\n",
    "cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(\n",
    "        labels=y, logits=cnnout['logit']))\n",
    "optm = tf.train.AdamOptimizer(learning_rate=0.001).minimize(cost) \n",
    "corr = tf.equal(tf.argmax(cnnout['logit'], 1), tf.argmax(y, 1))    \n",
    "accr = tf.reduce_mean(tf.cast(corr, \"float\"))\n",
    "\n",
    "# INITIALIZER\n",
    "init = tf.global_variables_initializer()\n",
    "print (\"FUNCTIONS READY\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 存储"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SAVER READY\n"
     ]
    }
   ],
   "source": [
    "savedir = \"nets/cnn_mnist_simple/\"\n",
    "saver = tf.train.Saver(max_to_keep=3)\n",
    "save_step = 4\n",
    "if not os.path.exists(savedir):\n",
    "    os.makedirs(savedir)\n",
    "print (\"SAVER READY\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 运行"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 004/020 cost: 0.035408186\n",
      "TRAIN ACCURACY: 0.980\n",
      "TEST ACCURACY: 0.986\n",
      "[nets/cnn_mnist_simple/net-4.ckpt] SAVED.\n",
      "Epoch: 008/020 cost: 0.014208857\n",
      "TRAIN ACCURACY: 0.990\n",
      "TEST ACCURACY: 0.987\n",
      "[nets/cnn_mnist_simple/net-8.ckpt] SAVED.\n",
      "Epoch: 012/020 cost: 0.006352575\n",
      "TRAIN ACCURACY: 1.000\n",
      "TEST ACCURACY: 0.987\n",
      "[nets/cnn_mnist_simple/net-12.ckpt] SAVED.\n",
      "Epoch: 016/020 cost: 0.003182677\n",
      "TRAIN ACCURACY: 1.000\n",
      "TEST ACCURACY: 0.987\n",
      "[nets/cnn_mnist_simple/net-16.ckpt] SAVED.\n",
      "Epoch: 020/020 cost: 0.001347903\n",
      "TRAIN ACCURACY: 1.000\n",
      "TEST ACCURACY: 0.987\n",
      "[nets/cnn_mnist_simple/net-20.ckpt] SAVED.\n",
      "OPTIMIZATION FINISHED\n"
     ]
    }
   ],
   "source": [
    "# PARAMETERS\n",
    "training_epochs = 20\n",
    "batch_size      = 100\n",
    "display_step    = 4\n",
    "# LAUNCH THE GRAPH\n",
    "sess = tf.Session()\n",
    "sess.run(init)\n",
    "# OPTIMIZE\n",
    "for epoch in range(training_epochs):\n",
    "    avg_cost = 0.\n",
    "    total_batch = int(mnist.train.num_examples/batch_size)\n",
    "    # ITERATION\n",
    "    for i in range(total_batch):\n",
    "        batch_xs, batch_ys = mnist.train.next_batch(batch_size)\n",
    "        feeds = {x: batch_xs, y: batch_ys}\n",
    "        sess.run(optm, feed_dict=feeds)\n",
    "        avg_cost += sess.run(cost, feed_dict=feeds)\n",
    "    avg_cost = avg_cost / total_batch\n",
    "    # DISPLAY\n",
    "    if (epoch+1) % display_step == 0:\n",
    "        print (\"Epoch: %03d/%03d cost: %.9f\" % (epoch+1, training_epochs, avg_cost))\n",
    "        feeds = {x: batch_xs, y: batch_ys}\n",
    "        train_acc = sess.run(accr, feed_dict=feeds)\n",
    "        print (\"TRAIN ACCURACY: %.3f\" % (train_acc))\n",
    "        feeds = {x: mnist.test.images, y: mnist.test.labels}\n",
    "        test_acc = sess.run(accr, feed_dict=feeds)\n",
    "        print (\"TEST ACCURACY: %.3f\" % (test_acc))\n",
    "    # SAVE\n",
    "    if (epoch+1) % save_step == 0:\n",
    "        savename = savedir+\"net-\"+str(epoch+1)+\".ckpt\"\n",
    "        saver.save(sess, savename)\n",
    "        print (\"[%s] SAVED.\" % (savename))\n",
    "print (\"OPTIMIZATION FINISHED\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 恢复"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DO NOTHING\n"
     ]
    }
   ],
   "source": [
    "do_restore = 0\n",
    "if do_restore == 1:\n",
    "    sess = tf.Session()\n",
    "    epoch = 20\n",
    "    savename = savedir+\"net-\"+str(epoch)+\".ckpt\"\n",
    "    saver.restore(sess, savename)\n",
    "    print (\"NETWORK RESTORED\")\n",
    "else:\n",
    "    print (\"DO NOTHING\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CNN如何工作"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "input_r = sess.run(cnnout['x_r'], feed_dict={x: trainimg[0:1, :]})\n",
    "conv1   = sess.run(cnnout['conv1'], feed_dict={x: trainimg[0:1, :]})\n",
    "conv2   = sess.run(cnnout['conv2'], feed_dict={x: trainimg[0:1, :]})\n",
    "conv3   = sess.run(cnnout['conv3'], feed_dict={x: trainimg[0:1, :]})\n",
    "pool    = sess.run(cnnout['pool'], feed_dict={x: trainimg[0:1, :]})\n",
    "dense   = sess.run(cnnout['dense'], feed_dict={x: trainimg[0:1, :]})\n",
    "out     = sess.run(cnnout['logit'], feed_dict={x: trainimg[0:1, :]}) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 输入"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Size of 'input_r' is (1, 28, 28, 1)\n",
      "Label is 7\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQIAAAD3CAYAAAD2Z1pOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nmBnw/9WyODOUS7mGAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f91a117e750>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print (\"Size of 'input_r' is %s\" % (input_r.shape,))\n",
    "label = np.argmax(trainlabel[0, :])\n",
    "print (\"Label is %d\" % (label))\n",
    "\n",
    "# PLOT\n",
    "plt.matshow(input_r[0, :, :, 0], cmap=plt.get_cmap('gray'))\n",
    "plt.title(\"Label of this image is \" + str(label) + \"\")\n",
    "plt.colorbar()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CONV 卷积层"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SIZE OF 'CONV1' IS (1, 28, 28, 64)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQAAAAD0CAYAAAB0BvjdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f91b46f5e50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQoAAAD0CAYAAABjJGgUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nFCJSSoFCREopUIhIqf8Dl57AgiW6U7EAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9188202fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQoAAAD0CAYAAABjJGgUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAGmVJREFUeJzt3W+MXNWZ5/Hvg42NsR03/oP/YTAGG43JKiaxnOwmmiWb\nSWRIsh6SWQQzyqAoGmeikJ1IeTGITBR2NC94sUkmo7AgJ7ECE/4Mq4TByniCgI3F5AVZtxEy2CbC\nEBv/w8bmX9vGYLuffXFvZ2va9zn3VlfVrerq30cqdfW5de85Vd399L3nueccc3dERFLO63YDRKT3\nKVCISCkFChEppUAhIqUUKESklAKFiJRSoBCRUgoUPcTM7jCzn3a7HSKjKVB0iJlNNbMfm9leMxsy\ns2fN7LqG7dea2f5utrFZZvZ+M3vMzI6ame7Um0Amd7sBfWwysA/4z8ArwPXAw2b2H9x9Tzcb1oLT\nwMPA/wL+ucttGTfWrl3rR48erfTabdu2PebuazvcpOa5ux41PYDtwOeB6cA7wDBwPH8sAu4g+0O8\nDxgCdgCrE8e7GngceB04DNyel08F/h44mD/+Hpiab7sW2A98AzgCHAK+mG/7MPAqMKmhjhuA7aPq\nvTL71en+ZzoeHh/60Id8eHi40gMY7HZ7ix669KiJmc0HVgA73P0EcB1w0N1n5I+D+Uv/K/AQMABs\nAn4QHG8m8ATwS7IgcyXwZL75m8BHgFXAB4A1wN807L4AmAUsBr4E3GVmF7n7b4ATwH9peO2fAg+0\n8NaF6v+Qe5UCRQ3M7HzgfuBed3+h5OW/dvfN7n4W+EeyP/QinwFedffvuPspdx/K/9AB/gz4W3c/\n4u6vAf8D+ELDvqfz7afdfTPZGc1V+bYHgZvzds8ku2R6sKk3LOdQoJAkMzuP7A/+PeDWCru82vD8\nJHCBmRX1JS0BXgqOsQjY2/D93rxsxDF3PzOqnhn58weAz5nZVOBzwDPu3ngsaVIzp/i9SoGig8zM\ngB8D84HPu/vphs2t/lbsA5YF2w4ClzV8f2leVsrdd5IFluvQZUfbKFCMgZmtNbPfmtluM7utC/Xv\nMbPn8pTlYAeruhv4A7I/6r1m9nzDtlPAQjN7ycweN7OLmjz2L/L9v56nYmea2YfzbQ8Cf2Nm88xs\nLtkZzbzG+vN7Ng6Y2bNkZxtrGo79APBXwB8C/7thHzOzC4Ap+fcX5GceITNbYma/MrOdZrbDzP4q\nL5+dv+8Xx/j+K0nU//v3nz+u70T9IxQommRmk4C7yP5jrQRuNrOVdbcD+Li7r3L31Z04uJldBnyZ\nrEPxT8hO7f/AzP4sf8nngWfIOi0/BvxdM8d39yHgk8BnyS5XXgQ+nm/+O2CQLMvyHPBvwCcKDvM9\nd19FdrbxfxvKHyRL6/4fd2/M611Glq3ZkX//DvDbkqaeAb7h7ivJOli/mv+8bwOedPflZJ2wnfqH\nEdUP+fvPH5s7VD8w/gNFN+6jWAPsdveXAczsIWAdsLMLbemY/LreRr43s6XAL9z9/rxoHXCtux8y\ns4XAFne/atQx9jQeo6CO5ykIAO5+Cvjv+aOxftx9C3CJmd3R8Pqlo/Z/hYJ/ImXtCdp4iCwFi7sP\nmdkusmzLOrJULcC9wBbgr5s5dov118bdGR4errPKtuvGpcdislPxEfup+QdH1j/whJltM7P1Ndc9\nYn7+SwzZGcH8LrTha2a23cw2durUv1EerK4BfkMX3v+o+qHG9z/ezygmamfmx/JT7uvITkX/sJuN\n8ew3pO7fkrvJOkNXkf3H/U4nKzOzGcDPgK+7+9uN2+p4/wX11/r+FSiad4AstTfikrysNu5+IP96\nBHiEf9+RV5fD+SUH+dcjdVbu7ofd/ay7DwM/pIOfQX4fyc+A+93953lxbe+/qP46339enwJFk7YC\ny83scjObAtxEdgdiLcxsen4jEWY2HfgU8Hx6r47YBNySP78FeLTOykf+SHM30KHPoCFFvMvdv9uw\nqZb3H9Vf1/sfMd4DRe2dme5+xsxuBR4DJgEb3X1HyW7tNB94JPv9YTLwgLv/spMVmtmDZB13c/MR\no98G7iQbJPYlsvsWbqy5/mvNbBXZKf8esgxNJ3yU7K7Q5/JULMDt1Pf+o/pvrun993wQqMLG+xsQ\n6XXXXHONb9mypdJrBwYGtnUqZd+KidqZKVKr4eHhSo8yeYbmyKib9xq3m5n9Q34z43Yz+2A72q9A\nIdJhVfsnKp7d/wRIzVdxHbA8f6wny+60TIFCpAbtChTu/hTZ/CORdcB9nnkaGBjVcTsmmuFKpAY1\n9gVGNzQeKn55NV07o+jiHZGqX/XXXn8TZxRzzWyw4dHVz2lENy89uv0BqH7VX5smAsVRd1/d8NjQ\nZFUduaFRfRQiHdbmzswym4A/z7MfHwHe8v8/pmbMWuqjMLO1wPfJbpz6kbvfmXr99OnTfWBgAIBZ\ns2axePHirt3EofpVfyv1Hzx48Ki7z6v6+naNHg1unjsfwN3vATaTTV+4m2zmsi+2o94xB4qGeSU+\nSdZhstXMNnk2Q1KhgYEBvvKVr4y1SpGe8a1vfaup6QHb1Znp7jeXbHfgq22prEErlx6/n1fC3d8j\nmzl6XXuaJdJfxvtYj1YCRS/MKyHS82ruo+iIjt9Hkad31kN2XSgyEfVyEKiilTOKSmkYd98wkuqZ\nPn16C9WJjF8T+Yzi9/NKkAWIm8imdxeRUcb7nJljDhQ9MK+EyLjQ62cLVbTUR+HZFOcdneZcpB9M\n6EAhItUoUIhIKQUKESmlQCEiSRO+M1NEqpmw6VERqU5nFCJSSoFCRJLURyEilShQiEgpBQoRKaVA\nISJJ7q70qIiU0xmFiJRSoBCRUgoUIlJKgUJEknTD1QSW6sU+//zzmzrW2bNnmz7Wu+++23S7zKzp\n+idPLv4VicrHe+9+pyhQiEip8R5AFShEaqAzChFJUh+FiFSiQCEipSZ0oDCzPcAQcBY44+6ry/Yp\n6tQ577zmVzZsNrMA8N5771Vu04ixZBeijMDp06cLy1O/RFOmTCksP3nyZGH5O++8Ex5r6tSpheWn\nTp0K9zlz5kxhebQ8ZGrZyAsvvDDc1u8mdKDIfdzdj7bhOCJ9S4FCRJL6YfRoK6uZAzjwhJltM7P1\nRS8ws/VmNmhmgydOnGixOpHxaSKvZg7wMXc/YGYXA4+b2Qvu/lTjC9x9A7ABYPHixb37SYh0UC8H\ngSpaOqNw9wP51yPAI8CadjRKpN9M2DMKM5sOnOfuQ/nzTwF/m9on+jCiLAHEmYKoHOD48eOF5VGm\nIHp9atvQ0FC4T/QDj7IuAwMD4bFmzJhRWJ7KbkSiTNGbb74Z7hPVE2VjonKAOXPmFJYvW7Ys3Gcs\nGbFe084gYGZrge8Dk4Afufudo7ZfCzwK/C4v+rm7J/8uq2jl0mM+8Eg+0Ggy8IC7/7LVBon0o3YE\nCjObBNwFfBLYD2w1s03uvnPUS//N3T/TcoUNxhwo3P1l4ANtbItI32rTGcUaYHf+t4eZPQSsA0YH\nirYb/+d1IuPA8PBwpQcwdyRLmD8as4mLgX0N3+/Py0b7T2a23cz+1cyubkf7dR+FSIc12UdxtMod\nzgnPAJe6+3Ezux74Z2B5C8cDdEYhUos2ZT0OAEsavr8kL2us5213P54/3wycb2ZzW22/AoVIDdoU\nKLYCy83scjObAtwEbGp8gZktsDzDYGZryP7Gj7Xa/lovPcysMEUXDTwCiO7mjNKGEKdbL7jggpIW\nnitKKY7lWNHgq1R6OBrI1ezAM4gHZV188cXhPtHnv2fPnsLygwcPhsc6erR4SFDq9uYrr7yysHy8\npU3b0Znp7mfM7FbgMbL06EZ332Fmf5lvvwf4E+ArZnYGeAe4ydtQufooRGrQrvso8suJzaPK7ml4\n/gPgB22prIEChUiH9cOgMAUKkRr08u3ZVShQiNRAgUJESilQtEFqUFY0tdpll10W7hNlF7Zv315Y\nHk13B80vgANxj3x0nZoaSDVr1qzC8ve9732F5amsx6RJkwrLU+9l6dKlheULFiwoLP/Vr34VHuuV\nV14pLE/9EUVtu/zyywvLo0WOuqnXR4ZW0ROBQqTfKVCISCkFChEppfSoiCSpj0JEKlGgEJFSChRN\nMLPC1GEqPRkNZEqtFBYNpLrkkksKy1P1z5s3r7A8mv8S4Nix4sF6r7/+etP1R/N8RunJ1OcSpQ6j\nzwVg0aJF4bYiqQFmP/3pTwvLDxw4UFgOcXp07tzikdNR2rjbFChEpJQChYgkqTNTRCpRelRESumM\nQkSSJsSlh5ltBD4DHHH39+dls4F/ApYCe4Ab3f2NsmMNDw8X9uL/7ne/K3h15umnny4s37VrV7jP\nRRddVFgeZSNSA6miXv+ZM2eG+0SZmiiDklr1KzplffvttwvLU9MKTps2rbB89uzZ4T7RlH+rVq0q\nLE8N1osyOBs3bgz32bt3b2F5NFhNWY/OqDLx4E+AtaPKbgOedPflwJP59yISGO9rj5YGinx18tE3\nAKwD7s2f3wv8cZvbJdJXxnugGGsfxXx3P5Q/f5VsHdJC+UpH6yGeW0Gk3/VyEKii5c5Md3czCz8F\nd98AbABYtGjR+P60RMagHybXHeviCIfNbCFA/vVI+5ok0n8m6qXHJuAW4M7866NVdjp16lRhtuK1\n114L94myC6np2wYGBgrL58yZU1geZRAgnqYumlYO4sxDNKYjygZAPKYkGjeS+s8VZVeizALAyy+/\nXFgejRtZsWJFeKzPfvazheWDg4PhPlu2bCksP3z4cGH5ypUrw2N18796LweBKqqkRx8EriVbZXk/\n8G2yAPGwmX0J2Avc2MlGiox3fR8o3P3mYNMn2twWkb7U65cVVejOTJEaKFCISCkFChEpNd7TowoU\nIh2mPoqxVFiQ1oxWfYI4PZlaXeyNN4rHp0UDhlKpzmilsmjgF8T/PaJ9okFsEKeO9+3bV1gepU0B\njhwpvt0lSjWm2hathvbpT386PFb0WV599dXhPtHgv2jVsdQUhamUeqcpUIhIKQUKESmlQCEipRQo\nRCRpIg8KE5EmtGtQmJmtNbPfmtluMztnwijL/EO+fbuZfbAd7a/1jGL69OmsXr36nPKoBx3iU7Yo\nswHx1HZR1uPUqVPhsaIBY0ePHg33iRbhiaaVS2VdomNFU86lFgCKMgWpTEmUEbn//vsLy1PTCn7w\ng8W/s6msTzTl3QsvvFBYHr1HgGXLloXbOq0dlx5mNgm4C/gksB/Yamab3H1nw8uuA5bnjw8Dd+df\nW6IzCpEatOmMYg2w291fdvf3gIfIZptrtA64zzNPAwMjU0K0QoFCpMOqBokKgWIx0HgDzf68rNnX\nNE2dmSI1aOLSY66ZNU7QsSGfJa6rFChEatBEoDjq7ud25GUOAEsavr8kL2v2NU3TpYdIDYaHhys9\nSmwFlpvZ5WY2BbiJbLa5RpuAP8+zHx8B3mqYCHvMaj2jcHfOnj17TnnqA4qmj0tlSqJ7+qOp4KZO\nnRoe6+KLLy4sTy200+yiPW+++WZ4rGjKuQULFjRVDvGYmtR7eeuttwrLozE4O3fuLCyHOOsULYwE\ncMUVVxSWR9P3pTJY3dKuQWHufsbMbgUeAyYBG919h5n9Zb79HmAzcD2wGzgJfLHlitGlh0gt2nVn\nprtvJgsGjWX3NDx34KttqayBAoVIDXQLt4iUUqAQkVIKFCKSpBmuRKSS8T56tMoCQBuBzwBH3P39\nedkdwF8AI/O03Z73xia5e3Kqsmak0qPRIKton9SgrKi9qdW9ol+KKNWZ+m8TpYejKfJSqdYo1Zma\n1m/GjBmF5VdddVVh+cKF8bCC6HOOpuhLmT17dmF56ufSTeP9jKLKDVc/AdYWlH/P3Vflj9IgITKR\n9f3ao+7+lJkt7XxTRPpTrweBKlq5hftr+cQYG80snFDAzNab2aCZDZ44caKF6kTGr/F+RjHWQHE3\nsAxYBRwCvhO90N03uPtqd18dTdcu0u/Ge6AYU9bD3X8/7ZGZ/RD4RdtaJNKHejkIVDGmQGFmCxtG\npN0APN++JlWTGsgUDViKsh6pTExUT5QNSB0vWrQoymxAnCmJBj+lFrlZsWJFYXnqlziaWm7x4uK5\nUJYsWVJYntqWynpEU/tF9ad+Lt3SD5PrVkmPPghcSzahxn7g28C1ZrYKcGAP8OUOtlFk3Ov7Mwp3\nv7mg+McdaItI3+r7QCEirVOgEJFSChQiktTrqc8qFChEaqBA0SWpNGCUiop+WKlBYdEgp1R6dnBw\nsLD8wIHiyZAPHjwYHiuaZzJKA0arkUG86lYqpRkdL/qMo/YCXHrppYXlL774YrhPtCJbNJdpNFis\n2/o+PSoirdMZhYgkqY9CRCpRoBCRUgoUIlJKgaJLUqt7RT+UoaGhwvLU6lLRoKRo1TGAV199tbD8\n2LFjheVXX311eKwoixBlI1LZoGifqF0Qf2bTpk0rLB8YGAiPFbUtNSgsmsMkOlZqpbRUpqqTJsSg\nMBFpnc4oRKSUAoWIlFKgEJFSChQikqQbrrootQBQNBXdSy+9VFi+c+fO8FgrV64sLL/iiivCfZYt\nW1ZYvnz58sLyefPmhceK3mfUiz537tzwWNHiQM8++2y4T5SRiBYTiqaoAzh79mxheSqDNHPmzKbK\nU1mfbmU9QGcUIlKB0qMiUkpnFCKSpD4KEalEgUJESilQiEipOgKFmc0G/glYSrbezo3u/kbB6/YA\nQ8BZ4Iy7ry47dpUFgJYA9wHzyRb82eDu36/aqE554424qv379xeWR6m+RYsWhceaNWtWYXk0rRvE\nA6Nee+21wvLU9HVRenTv3r1N1ZHaJ5qiD+JUczQVXbQaGsRp6JMnT4b7XHRR8frX0Tq23UyBptR0\nRnEb8KS732lmt+Xf/3Xw2o+7e/E8gwWqLFJ8BviGu68EPgJ81cxWNjRqOfBk/r2IjDIyerTKo0Xr\ngHvz5/cCf9zqAUeUBgp3P+Tuz+TPh4BdwOJONkqk3zSxmvlcMxtseKxvopr5DWsCv0p2FVDYHOAJ\nM9tW9fhN9VGY2VLgGuA3VRuVN2Q9xKfxIv2uiUuPo6k+AzN7AiiadOObo+pzM4sq/Zi7HzCzi4HH\nzewFd38q1ajKgcLMZgA/A77u7m83rrKdapS7bwA2ACxevHh8d/2KjEE776Nw9z+KtpnZYTNb6O6H\nzGwhUNgp5+4H8q9HzOwRYA2QDBRV+igws/PJgsT97v7zvPhw3hhSjRKRpi49WrEJuCV/fgvw6OgX\nmNl0M5s58hz4FPB82YGrZD2MbPXyXe7+3YJG3Rk1qpOiAUYQZwqihW7mz48u5eLLpSgbAHHPfzTl\nXipTEWUEovefmoru9OnTheWpAW6NZ46Nos8sNa1gVH9UB8SD3KZMmVJYnvq5dFNNWY87gYfN7EvA\nXuBGADNbBPzI3a8n6yJ4JP/MJwMPuPsvyw5c5dLjo8AXgOfMbGSY4e1Ro0TkXHUECnc/BnyioPwg\ncH3+/GXgA80euzRQuPuvgSjkn9MoEfn3NLmuiFSiW7hFpJQChYiUUqAQkVIKFF0yZ86ccFs0kChK\nm6bSc9FAsj179oT7RPNGRquORak+iH/BojkjU+nBaCBd6pc4Sp1GacvU/JvR+5w9e3a4z4UXXlhY\nHv3MejE9qolrRKQSBQoRKaX0qIiU0hmFiCSpj0JEKlGg6JLUoLAouxBlPVIrVUVTrqUyFZMmTQq3\ntUv0HqNpAAGGhoYKy6PBchCvlBYN/kpNRTdt2rTC8mhavZRezG6kKFCISCkFChFJ0qAwEalEZxQi\nUkqBQkRKKVB0Seqa7913321bPVF2ISpvt6lTpxaW79u3r7B869at4bGirMOKFSvCfaIxNSdOnCgs\nX758eXis6DNLzc4eZaSiafV6lQKFiCTphisRqUSBQkRKKT0qIqV0RiEiSeqjEJFK+j5QmNkS4D6y\nFYYc2ODu3zezO4C/AEaWubrd3Td3qqETVZQejUQpUIgHsqUG2EVTAUYp6FT90bR2qT+i8ZYGjfR9\noADOAN9w92fyNQu3mdnj+bbvufv/7FzzRPpD3wcKdz8EHMqfD5nZLqB49lgRKTTeA0Wl1cxHmNlS\n4BrgN3nR18xsu5ltNLPCqa/NbL2ZDZrZYHQ3n0g/Gxk9WuXRqyoHCjObAfwM+Lq7vw3cDSwDVpGd\ncXynaD933+Duq919dTQJjEi/G8l8lD16VaWsh5mdTxYk7nf3nwO4++GG7T8EftGRFor0gV4OAlVU\nyXoY8GNgl7t/t6F8Yd5/AXAD8HxnmjixHT9+vLA8Wsxn8uT4R7po0aLC8tQCPFEWI8p6DAwMhMeK\npsmLpugDZT16RZUzio8CXwCeM7ORZaBuB242s1VkKdM9wJc70kKRca7XLyuqqJL1+DVQtH6b7pkQ\nqajvA4WItE6BQkRK9XLqs4qm7qMQkeZVTY22etZhZv/NzHaY2bCZrU68bq2Z/dbMdpvZbVWOrUAh\nUoOa7qN4Hvgc8FT0AjObBNwFXAesJEtKFK/y1ECXHj3urbfeKiw/efJkYfnChQvDY1166aWF5a+/\n/nq4T1RPNMArNZdov8x/ORZ19FG4+y6A7I6G0Bpgt7u/nL/2IWAdsDO1k84oRGrQQ3dmLgYaZ2be\nT4WxWzqjEKlBE0FgrpkNNny/wd03jHxjZk8ACwr2+6a7P9pCE5MUKEQ6rMmzhaPuHnZEuvsftdic\nA0DjqtSX5GVJChQiNeih9OhWYLmZXU4WIG4C/rRsJ/VRiNSgpvToDWa2H/iPwL+Y2WN5+SIz25y3\n4wxwK/AYsAt42N13lB67zjvGzOw1YG/+7VzgaG2Vn0v1q/5W6r/M3edVeeGUKVN87ty5lQ566NCh\nbalLj26p9dKj8YM1s8FufiCqX/XXVf+EGBQmIq1ToBCRUgoUY7eh/CWqX/X3R/09lPUYk1o7M0Um\nosmTJ3tq5q9Gx44dU2emyEQ13v8hK1CI1ECBQkRKKVCISCkFChFJ0g1XIlLJeE+PKlCI1EBnFCJS\nSoFCRJLURyEilShQiEgpBQoRKaVAISJJ7q70qIiU0xmFiJRSoBCRUgoUIlLmMXevNg13d2cmD2mG\nKxEppQWARKSUAoWIlFKgEJFSChQiUkqBQkRKKVCISCkFChEppUAhIqUUKESk1P8DOKe83ZzK2xUA\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f91881c68d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print (\"SIZE OF 'CONV1' IS %s\" % (conv1.shape,))\n",
    "for i in range(3):\n",
    "    plt.matshow(conv1[0, :, :, i], cmap=plt.get_cmap('gray'))\n",
    "    plt.title(str(i) + \"th conv1\")\n",
    "    plt.colorbar()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CONV + BIAS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SIZE OF 'CONV2' IS (1, 28, 28, 64)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQAAAAD0CAYAAAB0BvjdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f91786e0dd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQoAAAD0CAYAAABjJGgUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9188096890>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQoAAAD0CAYAAABjJGgUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f91880fe550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print (\"SIZE OF 'CONV2' IS %s\" % (conv2.shape,))\n",
    "for i in range(3):\n",
    "    plt.matshow(conv2[0, :, :, i], cmap=plt.get_cmap('gray'))\n",
    "    plt.title(str(i) + \"th conv2\")\n",
    "    plt.colorbar()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CONV + BIAS + RELU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SIZE OF 'CONV3' IS (1, 28, 28, 64)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQIAAAD3CAYAAAD2Z1pOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9188147b90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQgAAAD3CAYAAADhRcqHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ns1IOEGZWygHCzEo5QJhZKQcIMyv1fwa5MfdD25KAAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f91a117e710>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQIAAAD3CAYAAAD2Z1pOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9178128050>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print (\"SIZE OF 'CONV3' IS %s\" % (conv3.shape,))\n",
    "for i in range(3):\n",
    "    plt.matshow(conv3[0, :, :, i], cmap=plt.get_cmap('gray'))\n",
    "    plt.title(str(i) + \"th conv3\")\n",
    "    plt.colorbar()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## POOL"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SIZE OF 'POOL' IS (1, 14, 14, 64)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQIAAAD3CAYAAAD2Z1pOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9178133e50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQgAAAD3CAYAAADhRcqHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f91882ad0d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQIAAAD3CAYAAAD2Z1pOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nmeFAYGY4EJgZDgRmhgOBmeFAYGbA/wOKPf/qcTR0OQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f91786c3ed0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print (\"SIZE OF 'POOL' IS %s\" % (pool.shape,))\n",
    "for i in range(3):\n",
    "    plt.matshow(pool[0, :, :, i], cmap=plt.get_cmap('gray'))\n",
    "    plt.title(str(i) + \"th pool\")\n",
    "    plt.colorbar()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## DENSE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SIZE OF 'DENSE' IS (1, 12544)\n",
      "SIZE OF 'OUT' IS (1, 10)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0YAAACFCAYAAAB/sOifAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9178663ed0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print (\"SIZE OF 'DENSE' IS %s\" % (dense.shape,))\n",
    "print (\"SIZE OF 'OUT' IS %s\" % (out.shape,))\n",
    "plt.matshow(out, cmap=plt.get_cmap('gray'))\n",
    "plt.title(\"OUT\")\n",
    "plt.colorbar()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CONVOLUTION FILTER 卷积核"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SIZE OF 'WC1' IS (7, 7, 1, 64)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQQAAAD0CAYAAAB97VinAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9178133590>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQQAAAD0CAYAAAB97VinAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f91347b2c10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQQAAAD0CAYAAAB97VinAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f917814e290>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "wc1 = sess.run(weights['c1'])\n",
    "print (\"SIZE OF 'WC1' IS %s\" % (wc1.shape,))\n",
    "for i in range(3):\n",
    "    plt.matshow(wc1[:, :, 0, i], cmap=plt.get_cmap('gray'))\n",
    "    plt.title(str(i) + \"th conv filter\")\n",
    "    plt.colorbar()\n",
    "    plt.show()"
   ]
  }
 ],
 "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": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}