Lecture22_DNN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# &#49888;&#44221;&#47581; (Deep Neural Network)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Tensor"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Tensor(&#53584;&#49436;)** : &#49688;&#52824;&#54805; &#45936;&#51060;&#53552;&#47484; &#50948;&#54620; &#52968;&#53580;&#51060;&#45320;&#51077;&#45768;&#45796;. &#51593;, &#51076;&#51032;&#51032; &#52264;&#50896; &#44060;&#49688;&#47484; &#44032;&#51648;&#45716; &#54665;&#47148;&#51032; &#51068;&#48152;&#54868;&#46108; &#47784;&#49845;&#51077;&#45768;&#45796;(&#45796;&#52264;&#50896; &#48176;&#50676;)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (1) 1D &#53584;&#49436; : &#48289;&#53552;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[12  3  6 14  7]\n",
      "&#52264;&#50896; :  1\n"
     ]
    }
   ],
   "source": [
    "x = np.array([12, 3, 6, 14, 7])\n",
    "print(x)\n",
    "print('&#52264;&#50896; : ', x.ndim)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (2) 2D &#53584;&#49436; : &#54665;&#47148;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 5 78  2 34  0]\n",
      " [ 6 79  3 35  1]\n",
      " [ 7 80  4 36  2]]\n",
      "&#52264;&#50896; :  2\n"
     ]
    }
   ],
   "source": [
    "x = np.array([[5, 78, 2, 34, 0],\n",
    "              [6, 79, 3, 35, 1],\n",
    "              [7, 80, 4, 36, 2]])\n",
    "print(x)\n",
    "print('&#52264;&#50896; : ', x.ndim)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (3) 3D &#53584;&#49436; : &#44256;&#52264;&#50896; &#53584;&#49436;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[[ 5 78  2 34  0]\n",
      "  [ 6 79  3 35  1]\n",
      "  [ 7 80  4 36  2]]\n",
      "\n",
      " [[ 5 78  2 34  0]\n",
      "  [ 6 79  3 35  1]\n",
      "  [ 7 80  4 36  2]]\n",
      "\n",
      " [[ 5 78  2 34  0]\n",
      "  [ 6 79  3 35  1]\n",
      "  [ 7 80  4 36  2]]]\n",
      "&#52264;&#50896; :  3\n"
     ]
    }
   ],
   "source": [
    "x = np.array([[[5, 78, 2, 34, 0],\n",
    "               [6, 79, 3, 35, 1],\n",
    "               [7, 80, 4, 36, 2]],\n",
    "              [[5, 78, 2, 34, 0],\n",
    "               [6, 79, 3, 35, 1],\n",
    "               [7, 80, 4, 36, 2]],\n",
    "              [[5, 78, 2, 34, 0],\n",
    "               [6, 79, 3, 35, 1],\n",
    "               [7, 80, 4, 36, 2]]])\n",
    "print(x)\n",
    "print('&#52264;&#50896; : ', x.ndim)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. &#49888;&#44221;&#47581;&#51032; &#44396;&#51312;\n",
    "<img src=\"./images/dnn_network.png\" style=\"height:200px width:400px\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- &#52789;(layer) : &#54616;&#45208; &#51060;&#49345;&#51032; &#53584;&#49436;&#47484; &#51077;&#47141;&#51004;&#47196; &#48155;&#50500; &#54616;&#45208; &#51060;&#49345;&#51032; &#53584;&#49436;&#47484; &#52636;&#47141;&#54616;&#45716; &#45936;&#51060;&#53552; &#52376;&#47532; &#47784;&#46280;&#51077;&#45768;&#45796;.\n",
    "- &#44032;&#51473;&#52824;(weight) : &#52789;&#51032; &#49345;&#53468;&#51060;&#47728;, &#54869;&#47456;&#51201; &#44221;&#49324; &#54616;&#44053;&#48277;&#50640; &#51032;&#54644; &#54617;&#49845;&#46104;&#45716; &#54616;&#45208; &#51060;&#49345;&#51032; &#53584;&#49436;&#51077;&#45768;&#45796;. &#50668;&#44592;&#50640;&#45716; &#45348;&#53944;&#50892;&#53356;&#44032; &#54617;&#49845;&#54620; &#51648;&#49885;&#51060; &#46308;&#50612; &#51080;&#49845;&#45768;&#45796;.\n",
    "- &#52789;&#51032; &#45348;&#53944;&#50892;&#53356; : &#52789;&#51012; &#49939;&#45716; &#44163;&#51077;&#45768;&#45796;.\n",
    "- &#49552;&#49892; &#54632;&#49688;(&#47785;&#51201; &#54632;&#49688;) : &#54984;&#47144;&#54616;&#45716; &#46041;&#50504; &#52572;&#49548;&#54868;&#46112; &#44050;&#51077;&#45768;&#45796;. &#51452;&#50612;&#51652; &#47928;&#51228;&#50640; &#45824;&#54620; &#49457;&#44277; &#51648;&#54364;&#44032; &#46121;&#45768;&#45796;.\n",
    "- &#50741;&#54000;&#47560;&#51060;&#51200; : &#49552;&#49892; &#54632;&#49688;&#47484; &#44592;&#48152;&#51004;&#47196; &#45348;&#53944;&#50892;&#53356;&#44032; &#50612;&#46523;&#44172; &#50629;&#45936;&#51060;&#53944;&#46112;&#51648;&#47484; &#44208;&#51221;&#54633;&#45768;&#45796;. &#53945;&#51221; &#51333;&#47448;&#51032; &#54869;&#47456;&#51201; &#44221;&#49324; &#54616;&#44053;&#48277;(SGD)&#51012; &#44396;&#54788;&#54633;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Keras"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#52992;&#46972;&#49828;&#45716; &#46373;&#47084;&#45789; &#47784;&#45944;&#51012; &#44036;&#54200;&#54616;&#44172; &#47564;&#46308;&#44256; &#54984;&#47144;&#49884;&#53420; &#49688; &#51080;&#45716; &#54028;&#51060;&#50028;&#51012; &#50948;&#54620; &#46373;&#47084;&#45789; &#54532;&#47112;&#51076;&#50892;&#53356;&#51077;&#45768;&#45796;.\n",
    "\n",
    "&#52992;&#46972;&#49828;&#51032; &#53945;&#51669;&#51008; &#45796;&#51020;&#44284; &#44057;&#49845;&#45768;&#45796;.\n",
    " - &#46041;&#51068;&#54620; &#53076;&#46300;&#47196; CPU&#50752; GPU&#50640;&#49436; &#49892;&#54665;&#54624; &#49688; &#51080;&#49845;&#45768;&#45796;.\n",
    " - &#49324;&#50857;&#54616;&#44592; &#49772;&#50868; API&#47484; &#44032;&#51648;&#44256; &#51080;&#50612; &#46373;&#47084;&#45789; &#47784;&#45944;&#51032; &#54532;&#47196;&#53664;&#53440;&#51077;&#51012; &#48736;&#47476;&#44172; &#47564;&#46308; &#49688; &#51080;&#49845;&#45768;&#45796;.\n",
    " - &#54633;&#49457;&#44273; &#49888;&#44221;&#47581;, &#49692;&#54872; &#49888;&#44221;&#47581;&#51012; &#51648;&#50896;&#54616;&#47728; &#51060; &#46168;&#51012; &#51088;&#50976;&#47213;&#44172; &#51312;&#54633;&#54616;&#50668; &#49324;&#50857;&#54624; &#49688; &#51080;&#49845;&#45768;&#45796;.\n",
    " - &#45796;&#51473; &#51077;&#47141;&#51060;&#45208;, &#45796;&#51473; &#52636;&#47141; &#47784;&#45944;, &#52789;&#51032; &#44277;&#50976;, &#47784;&#45944; &#44277;&#50976; &#46321; &#50612;&#46500; &#45348;&#53944;&#50892;&#53356; &#44396;&#51312;&#46020; &#47564;&#46308; &#49688; &#51080;&#49845;&#45768;&#45796;. &#51060; &#47568;&#51008; &#51201;&#45824;&#51201; &#49373;&#49457; &#49888;&#44221;&#47581;&#48512;&#53552; &#45684;&#47092; &#53916;&#47553; &#47672;&#49888;&#44620;&#51648; &#52992;&#46972;&#49828;&#45716; &#44592;&#48376;&#51201;&#51004;&#47196; &#50612;&#46500; &#46373;&#47084;&#45789; &#47784;&#45944;&#50640;&#46020; &#51201;&#54633;&#54616;&#45796;&#45716; &#46907;&#51077;&#45768;&#45796;.\n",
    " \n",
    "&#44396;&#44544;, &#45367;&#54540;&#47533;&#49828;, &#50864;&#48260;, &#50028;, &#50704;&#54532;, &#49828;&#53272;&#50612; &#44536;&#47532;&#44256; &#45796;&#50577;&#54620; &#48516;&#50556;&#51032; &#49828;&#53440;&#53944;&#50629; &#49688;&#48177; &#44060;&#44032; &#52992;&#46972;&#49828;&#47484; &#49324;&#50857;&#54633;&#45768;&#45796;. &#47672;&#49888; &#47084;&#45789; &#44221;&#50672; &#50937;&#49324;&#51060;&#53944;&#51064; &#52880;&#44544;&#50640;&#49436;&#46020; &#52992;&#46972;&#49828;&#51032; &#51064;&#44592;&#44032; &#45458;&#49845;&#45768;&#45796;. &#52572;&#44540;&#50640; &#44144;&#51032; &#47784;&#46304; &#46373;&#47084;&#45789; &#44221;&#47144; &#45824;&#54924;&#51032; &#50864;&#49849;&#51088;&#46308;&#51060; &#52992;&#46972;&#49828;&#47484; &#49324;&#50857;&#54616;&#44256; &#51080;&#49845;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"http://snowdeer.github.io/assets/machine-learning/012.jpg\" style=\"height:200px width:400px\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Keras&#51032; DNN&#51012; &#51060;&#50857;&#54616;&#50668; MNIST &#48516;&#47448;&#54616;&#44592;\n",
    "### (1) &#45936;&#51060;&#53552; &#47196;&#46377;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.datasets import mnist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "(train_images, train_labels), (test_images, test_labels) = \\\n",
    "    mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "&#54617;&#49845;&#49483; x :  (60000, 28, 28)\n",
      "&#54617;&#49845;&#49483; y :  (60000,)\n",
      "&#53580;&#49828;&#53944;&#49483; x :  (10000, 28, 28)\n",
      "&#53580;&#49828;&#53944;&#49483; y :  (10000,)\n"
     ]
    }
   ],
   "source": [
    "print('&#54617;&#49845;&#49483; x : ', train_images.shape)\n",
    "print('&#54617;&#49845;&#49483; y : ', train_labels.shape)\n",
    "print('&#53580;&#49828;&#53944;&#49483; x : ', test_images.shape)\n",
    "print('&#53580;&#49828;&#53944;&#49483; y : ', test_labels.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   3,\n",
       "         18,  18,  18, 126, 136, 175,  26, 166, 255, 247, 127,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,  30,  36,  94, 154, 170,\n",
       "        253, 253, 253, 253, 253, 225, 172, 253, 242, 195,  64,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,  49, 238, 253, 253, 253, 253,\n",
       "        253, 253, 253, 253, 251,  93,  82,  82,  56,  39,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,  18, 219, 253, 253, 253, 253,\n",
       "        253, 198, 182, 247, 241,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,  80, 156, 107, 253, 253,\n",
       "        205,  11,   0,  43, 154,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,  14,   1, 154, 253,\n",
       "         90,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0, 139, 253,\n",
       "        190,   2,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  11, 190,\n",
       "        253,  70,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  35,\n",
       "        241, 225, 160, 108,   1,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "         81, 240, 253, 253, 119,  25,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,  45, 186, 253, 253, 150,  27,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,  16,  93, 252, 253, 187,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,   0,   0, 249, 253, 249,  64,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,  46, 130, 183, 253, 253, 207,   2,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  39,\n",
       "        148, 229, 253, 253, 253, 250, 182,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  24, 114, 221,\n",
       "        253, 253, 253, 253, 201,  78,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,  23,  66, 213, 253, 253,\n",
       "        253, 253, 198,  81,   2,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,  18, 171, 219, 253, 253, 253, 253,\n",
       "        195,  80,   9,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,  55, 172, 226, 253, 253, 253, 253, 244, 133,\n",
       "         11,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0, 136, 253, 253, 253, 212, 135, 132,  16,   0,\n",
       "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0],\n",
       "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
       "          0,   0]], dtype=uint8)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_images[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_labels[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (2) DNN &#54617;&#49845;\n",
    " - &#54984;&#47144; &#45936;&#51060;&#53552; train_images, train_labels&#47484; &#45348;&#53944;&#50892;&#53356;&#50640; &#51452;&#51077;\n",
    " - &#45348;&#53944;&#50892;&#53356;&#45716; &#51060;&#48120;&#51648;&#50752; &#47112;&#51060;&#48660;&#51012; &#50672;&#44288;&#49884;&#53420;&#49688; &#51080;&#46020;&#47197; &#54617;&#49845;&#46120;\n",
    " - test_images&#50640; &#45824;&#54620; &#50696;&#52769;&#51012; &#45348;&#53944;&#50892;&#53356;&#50640; &#50836;&#52397;\n",
    " - &#50696;&#52769;&#51060; train_labels&#50752; &#47582;&#45716;&#51648; &#54869;&#51064;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Layer&#45716; &#51452;&#50612;&#51652; &#47928;&#51228;&#50640; &#45908; &#51032;&#48120;&#51080;&#45716; **&#54364;&#54788;(representation)** &#51012; &#51077;&#47141;&#46108; &#45936;&#51060;&#53552;&#47196;&#48512;&#53552; &#52628;&#52636;&#54633;&#45768;&#45796;. &#45824;&#48512;&#48516;&#51032; &#46373;&#47084;&#45789;&#51008; &#44036;&#45800;&#54620; &#52789;&#51012; &#50672;&#44208;&#54616;&#50668; &#44396;&#49457;&#46104;&#50612; &#51080;&#44256;, &#51216;&#51652;&#51201;&#51004;&#47196; &#45936;&#51060;&#53552;&#47484; &#51221;&#51228;&#54616;&#45716; &#54805;&#53468;&#47484; &#46916;&#44256; &#51080;&#49845;&#45768;&#45796;. &#46373;&#47084;&#45789; &#47784;&#45944;&#51008; &#45936;&#51060;&#53552; &#51221;&#51228; &#54596;&#53552;(&#52789;)&#44032; &#50672;&#49549;&#46104;&#50612; &#51080;&#45716; &#45936;&#51060;&#53552; &#54532;&#47196;&#49464;&#49905;&#51012; &#50948;&#54620; &#50668;&#44284;&#44592;&#50752; &#44057;&#49845;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> &#45348;&#53944;&#50892;&#53356; &#44396;&#49457;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import models\n",
    "from keras import layers\n",
    "\n",
    "network = models.Sequential()\n",
    "network.add(layers.Dense(512, activation='relu', input_shape=(28*28, )))\n",
    "network.add(layers.Dense(10, activation='softmax'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#52395;&#48264;&#51704; &#52789;&#51008; input&#51060; 784 &#52264;&#50896;&#51060;&#44256;, output&#51060; 512&#52264;&#50896;&#51077;&#45768;&#45796;.\n",
    "\n",
    "&#47560;&#51648;&#47561; &#52789;&#51008; 10&#44060;&#51032; &#54869;&#47456; &#51216;&#49688;&#44032; &#46308;&#50612; &#51080;&#45716; &#48176;&#50676;&#51012; &#48152;&#54872;&#54616;&#45716; **softmax(&#49548;&#54532;&#53944;&#47589;&#49828;)** &#52789;&#51077;&#45768;&#45796;. &#44033; &#51216;&#49688;&#45716; &#54788;&#51116; &#49707;&#51088; &#51060;&#48120;&#51648;&#44032; 10&#44060;&#51032; &#49707;&#51088; &#53364;&#47000;&#49828; &#51473; &#54616;&#45208;&#50640; &#49549;&#54624; &#54869;&#47456;&#51077;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> &#54984;&#47144;&#51012; &#50948;&#54620; &#52980;&#54028;&#51068;\n",
    "\n",
    " - &#49552;&#49892; &#54632;&#49688;(loss function) : &#54984;&#47144; &#45936;&#51060;&#53552;&#50640;&#49436; &#49888;&#44221;&#47581;&#51032; &#49457;&#45733;&#51012; &#52769;&#51221;&#54616;&#45716; &#48169;&#48277;&#51004;&#47196; &#45348;&#53944;&#50892;&#53356;&#44032; &#50739;&#51008; &#48169;&#54693;&#51004;&#47196; &#54617;&#49845;&#46112; &#49688; &#51080;&#46020;&#47197; &#46020;&#50752;&#51469;&#45768;&#45796;.\n",
    " - &#50741;&#54000;&#47560;&#51060;&#51200;(optimizer) : &#51077;&#47141;&#46108; &#45936;&#51060;&#53552;&#50752; &#49552;&#49892; &#54632;&#49688;&#47484; &#44592;&#48152;&#51004;&#47196; &#45348;&#53944;&#50892;&#53356;&#47484; &#50629;&#45936;&#51060;&#53944;&#54616;&#45716; &#47700;&#52964;&#45768;&#51608;&#51077;&#45768;&#45796;.\n",
    " - &#54984;&#47144;&#44284; &#53580;&#49828;&#53944; &#44284;&#51221;&#51012; &#47784;&#45768;&#53552;&#47553;&#54624; &#51648;&#54364; : &#50668;&#44592;&#50640;&#49436;&#45716; &#51221;&#54869;&#46020;(&#51221;&#54869;&#55176; &#48516;&#47448;&#46108; &#51060;&#48120;&#51648;&#51032; &#48708;&#50984;)&#47564; &#44256;&#47140;&#54616;&#44192;&#49845;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "network.compile(optimizer='rmsprop',\n",
    "                loss='categorical_crossentropy',\n",
    "                metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> &#45936;&#51060;&#53552; &#49828;&#52992;&#51068;&#47553;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# &#54984;&#47144;&#51012; &#49884;&#51089;&#54616;&#44592; &#51204;&#50640; &#47784;&#46304; &#44050;&#51012; 0&#44284; 1 &#49324;&#51060;&#47196; &#49828;&#52992;&#51068;&#51012; &#51312;&#51221;\n",
    "train_images = train_images.reshape((60000, 28 * 28))\n",
    "train_images = train_images.astype('float32') / 255\n",
    "\n",
    "test_images = test_images.reshape((10000, 28 * 28))\n",
    "test_images = test_images.astype('float32') / 255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# &#47112;&#51060;&#48660;\n",
    "from keras.utils import to_categorical\n",
    "\n",
    "train_labels = to_categorical(train_labels)\n",
    "test_labels = to_categorical(test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0., 0., 0., 0., 0., 1., 0., 0., 0., 0.], dtype=float32)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_labels[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(60000, 10)\n",
      "(10000, 10)\n"
     ]
    }
   ],
   "source": [
    "print(train_labels.shape)\n",
    "print(test_labels.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> &#54984;&#47144;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "60000/60000 [==============================] - 4s 71us/step - loss: 0.2557 - acc: 0.9259\n",
      "Epoch 2/5\n",
      "60000/60000 [==============================] - 4s 71us/step - loss: 0.1046 - acc: 0.9687\n",
      "Epoch 3/5\n",
      "60000/60000 [==============================] - 4s 70us/step - loss: 0.0694 - acc: 0.9787\n",
      "Epoch 4/5\n",
      "60000/60000 [==============================] - 4s 70us/step - loss: 0.0501 - acc: 0.9847\n",
      "Epoch 5/5\n",
      "60000/60000 [==============================] - 4s 72us/step - loss: 0.0373 - acc: 0.9886\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0xb26c18c88>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "network.fit(train_images, train_labels, epochs=5, batch_size=128)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 58us/step\n"
     ]
    }
   ],
   "source": [
    "test_loss, test_acc = network.evaluate(test_images, test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_acc 0.9791\n"
     ]
    }
   ],
   "source": [
    "print('test_acc', test_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 512)               401920    \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                5130      \n",
      "=================================================================\n",
      "Total params: 407,050\n",
      "Trainable params: 407,050\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "network.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Keras&#51032; DNN&#51012; &#51060;&#50857;&#54616;&#50668; &#50689;&#54868; &#47532;&#48624; &#48516;&#47448;&#54616;&#44592; (Classification)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#51064;&#53552;&#45367; &#50689;&#54868; &#45936;&#51060;&#53552;&#48288;&#51060;&#49828;(internet movie database)&#47196;&#48512;&#53552; &#44032;&#51256;&#50728; &#47532;&#48624; 5&#47564; &#44060;&#47196; &#51060;&#47336;&#50612;&#51652; IMDB &#45936;&#51060;&#53552;&#49483;&#51012; &#51060;&#50857;&#54616;&#50668; DNN &#48516;&#47448;&#44592;&#47484; &#49373;&#49457;&#54616;&#44192;&#49845;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (1) IMDB &#45936;&#51060;&#53552;&#49483; &#47196;&#46377;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.datasets import imdb\n",
    "\n",
    "# &#54984;&#47144; &#45936;&#51060;&#53552;&#50640;&#49436; &#44032;&#51109; &#51088;&#51452; &#45208;&#53440;&#45208;&#45716; &#45800;&#50612; 1&#47564; &#44060;&#47564; &#49324;&#50857;&#54620;&#45796;.\n",
    "(train_data, train_labels), (test_data, test_labels) = \\\n",
    "    imdb.load_data(num_words=10000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1, 14, 22, 16, 43, 530, 973, 1622, 1385, 65, 458, 4468, 66, 3941, 4, 173, 36, 256, 5, 25, 100, 43, 838, 112, 50, 670, 2, 9, 35, 480, 284, 5, 150, 4, 172, 112, 167, 2, 336, 385, 39, 4, 172, 4536, 1111, 17, 546, 38, 13, 447, 4, 192, 50, 16, 6, 147, 2025, 19, 14, 22, 4, 1920, 4613, 469, 4, 22, 71, 87, 12, 16, 43, 530, 38, 76, 15, 13, 1247, 4, 22, 17, 515, 17, 12, 16, 626, 18, 2, 5, 62, 386, 12, 8, 316, 8, 106, 5, 4, 2223, 5244, 16, 480, 66, 3785, 33, 4, 130, 12, 16, 38, 619, 5, 25, 124, 51, 36, 135, 48, 25, 1415, 33, 6, 22, 12, 215, 28, 77, 52, 5, 14, 407, 16, 82, 2, 8, 4, 107, 117, 5952, 15, 256, 4, 2, 7, 3766, 5, 723, 36, 71, 43, 530, 476, 26, 400, 317, 46, 7, 4, 2, 1029, 13, 104, 88, 4, 381, 15, 297, 98, 32, 2071, 56, 26, 141, 6, 194, 7486, 18, 4, 226, 22, 21, 134, 476, 26, 480, 5, 144, 30, 5535, 18, 51, 36, 28, 224, 92, 25, 104, 4, 226, 65, 16, 38, 1334, 88, 12, 16, 283, 5, 16, 4472, 113, 103, 32, 15, 16, 5345, 19, 178, 32]\n"
     ]
    }
   ],
   "source": [
    "print(train_data[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_labels[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#44033; &#45800;&#50612;&#45716; id&#47484; &#44032;&#51648;&#44256; &#51080;&#49845;&#45768;&#45796;. id&#47484; &#51060;&#50857;&#54616;&#50668; &#47532;&#48624;&#47484; &#47928;&#51109;&#51004;&#47196; &#48148;&#44984;&#50612; &#48389;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('fawn', 34701),\n",
       " ('tsukino', 52006),\n",
       " ('nunnery', 52007),\n",
       " ('sonja', 16816),\n",
       " ('vani', 63951)]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "word_index = imdb.get_word_index()\n",
    "# Word index&#45716; dict&#51077;&#45768;&#45796;.\n",
    "list(word_index.items())[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(34701, 'fawn'),\n",
       " (52006, 'tsukino'),\n",
       " (52007, 'nunnery'),\n",
       " (16816, 'sonja'),\n",
       " (63951, 'vani')]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reverse_word_index = dict(\n",
    "    [(value, key) for (key, value) in word_index.items()]\n",
    ")\n",
    "list(reverse_word_index.items())[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"? this film was just brilliant casting location scenery story direction everyone's really suited the part they played and you could just imagine being there robert ? is an amazing actor and now the same being director ? father came from the same scottish island as myself so i loved the fact there was a real connection with this film the witty remarks throughout the film were great it was just brilliant so much that i bought the film as soon as it was released for ? and would recommend it to everyone to watch and the fly fishing was amazing really cried at the end it was so sad and you know what they say if you cry at a film it must have been good and this definitely was also ? to the two little boy's that played the ? of norman and paul they were just brilliant children are often left out of the ? list i think because the stars that play them all grown up are such a big profile for the whole film but these children are amazing and should be praised for what they have done don't you think the whole story was so lovely because it was true and was someone's life after all that was shared with us all\""
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# &#47532;&#48624;&#47484; &#46356;&#53076;&#46377;&#54620;&#45796;. 0, 1, 2&#45716; &#54056;&#46377;/&#47928;&#49436;&#49884;&#51089;/&#49324;&#51204;&#50640; &#50630;&#51012;&#51012; &#50948;&#54620; &#51064;&#45937;&#49828;&#51060;&#44592; &#46412;&#47928;&#50640; 3&#51012; &#48764;&#51468;\n",
    "decode_review = ' '.join(\n",
    "    [reverse_word_index.get(i - 3, '?') for i in train_data[0]]\n",
    ")\n",
    "decode_review"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (2) &#45936;&#51060;&#53552; &#51456;&#48708; (&#51064;&#53076;&#46377;)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# &#47532;&#49828;&#53944;&#47484; &#50896;-&#54635; &#51064;&#53076;&#46377;&#54616;&#50668; 0&#44284; 1&#51032; &#48289;&#53552;&#47196; &#48320;&#54872;&#54633;&#45768;&#45796;.\n",
    "\n",
    "import numpy as np\n",
    "\n",
    "def vectorize_sequences(sequences, dimension=10000):\n",
    "    # &#53356;&#44592;&#44032; (len(sequences), dimension) &#51060;&#44256; &#47784;&#46304; &#50896;&#49548;&#44032; 0&#51064; &#54665;&#47148;&#51012; &#49373;&#49457;\n",
    "    results = np.zeros((len(sequences), dimension))\n",
    "    for i, sequence in enumerate(sequences):\n",
    "        # Result[i]&#50640;&#49436; &#53945;&#51221; &#51064;&#45937;&#49828;&#51032; &#50948;&#52824;&#47484; 1&#47196; &#47564;&#46304;&#45796;.\n",
    "        results[i, sequence] = 1.\n",
    "    return results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train = vectorize_sequences(train_data)\n",
    "x_test = vectorize_sequences(test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(25000, 10000)\n",
      "(25000, 10000)\n"
     ]
    }
   ],
   "source": [
    "print(x_train.shape)\n",
    "print(x_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0., 1., 1., ..., 0., 0., 0.])"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train = np.asarray(train_labels).astype('float')\n",
    "y_test = np.asarray(test_labels).astype('float')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(25000,)\n",
      "(25000,)\n"
     ]
    }
   ],
   "source": [
    "print(y_train.shape)\n",
    "print(y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (3) &#49888;&#44221;&#47581; &#47784;&#45944; &#49373;&#49457;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Dense &#52789;&#50640; &#51204;&#45804;&#54616;&#45716; &#47588;&#44060;&#48320;&#49688; $x$&#45716; hidden unit&#51032; &#44060;&#49688;&#44032; $x$&#46972;&#45716; &#51032;&#48120;&#51077;&#45768;&#45796;. &#51593; &#44032;&#51473;&#52824; &#54665;&#47148; $W$&#51032; &#53356;&#44592;&#44032; (input_dimension, $x$)&#46972;&#45716; &#46907;&#51077;&#45768;&#45796;. &#51077;&#47141; &#45936;&#51060;&#53552;&#50752; $W$&#47484; &#48289;&#53552;&#51032; &#45236;&#51201;&#51012; &#51060;&#50857;&#54616;&#50668; &#50672;&#49328;&#51012; &#54616;&#44172; &#46104;&#47732; &#51077;&#47141; &#45936;&#51060;&#53552;&#44032; 16&#52264;&#50896;&#51004;&#47196; &#54364;&#54788;&#46108; &#44277;&#44036;&#51004;&#47196; &#53804;&#50689;&#46121;&#45768;&#45796;. &#54364;&#54788; &#44277;&#44036;&#51032; &#52264;&#50896;&#51012; &#49888;&#44221;&#47581;&#51060; &#45236;&#51116;&#46108; &#54364;&#54788;&#51012; &#54617;&#49845;&#54624; &#46412; &#44032;&#51656; &#49688; &#51080;&#45716; &#51088;&#50976;&#46020;&#47196; &#51060;&#54644;&#54624; &#49688; &#51080;&#49845;&#45768;&#45796;. &#46384;&#46972;&#49436; hidden unit&#51032; &#49688;&#47484; &#45720;&#47532;&#47732;(&#54364;&#54788; &#44277;&#44036;&#51012; &#45908; &#44256;&#52264;&#50896;&#51004;&#47196; &#47564;&#46308;&#47732;) &#49888;&#44221;&#47581;&#51060; &#45908;&#50865; &#48373;&#51105;&#54620; &#54364;&#54788;&#51012; &#54617;&#49845;&#54624; &#49688; &#51080;&#51648;&#47564; &#44228;&#49328; &#48708;&#50857;&#51060; &#52964;&#51648;&#44256; &#50896;&#54616;&#51648; &#50506;&#45716; &#54056;&#53556;&#51012; &#54617;&#49845;&#54624; &#49688;&#46020; &#51080;&#49845;&#45768;&#45796;.\n",
    "\n",
    "Dense &#52789;&#51012; &#49939;&#51012; &#46412; &#54596;&#50836;&#54620; &#46160; &#44032;&#51648; &#51473;&#50836;&#54620; &#44396;&#51312;&#49345;&#51032; &#44208;&#51221;\n",
    " - &#50620;&#47560;&#45208; &#47566;&#51008; &#52789;&#51012; &#49324;&#50857;&#54624; &#44163;&#51064;&#44032;?\n",
    " - &#44033; &#52789;&#50640; &#50620;&#47560;&#45208; &#47566;&#51008; hidden unit&#51012; &#46168; &#44163;&#51064;&#44032;?\n",
    " \n",
    "&#44396;&#51312;\n",
    " - 16&#44060;&#51032; hidden unit&#51012; &#44032;&#51652; 2&#44060;&#51032; hidden layer\n",
    " - &#54788;&#51116; &#47532;&#48624;&#51032; &#44048;&#51221;&#51012; &#49828;&#52860;&#46972; &#44050;&#51032; &#50696;&#52769;&#51004;&#47196; &#52636;&#47141;&#54616;&#45716; 3&#48264;&#51704; &#52789;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import models\n",
    "from keras import layers\n",
    "\n",
    "model = models.Sequential()\n",
    "model.add(layers.Dense(16, activation='relu'))\n",
    "model.add(layers.Dense(16, activation='relu'))\n",
    "model.add(layers.Dense(1, activation='sigmoid'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "**&#54876;&#49457;&#54868; &#54632;&#49688;&#44032; &#50780; &#54596;&#50836;&#54620;&#44032;?**\n",
    "\n",
    "relu&#50752; &#44057;&#51008; &#54876;&#49457;&#54868; &#54632;&#49688;(&#46608;&#45716; &#48708;&#49440;&#54805;&#49457;, non-linearity)&#44032; &#50630;&#45796;&#47732; Dense &#52789;&#51008; &#49440;&#54805;&#51201;&#51064; &#50672;&#49328;&#51064; &#45236;&#51201;&#44284; &#45927;&#49480;&#51004;&#47196;&#47564; &#44396;&#49457;&#46121;&#45768;&#45796;. &#53076;&#46300;&#47196; &#48372;&#47732; &#45796;&#51020;&#44284; &#44057;&#49845;&#45768;&#45796;. `output = dot(W, input) + b` &#51060;&#47111;&#44172; &#45208;&#50724;&#47732; &#54644;&#45817; &#52789;&#51008; &#51077;&#47141;&#50640; &#45824;&#54620; &#49440;&#54805; &#48320;&#54872;(affine &#48320;&#54872;)&#47564;&#51012; &#54617;&#49845;&#54624; &#49688; &#51080;&#49845;&#45768;&#45796;. &#51060; &#52789;&#51032; &#44032;&#49444; &#44277;&#44036;&#51008; &#51077;&#47141; &#45936;&#51060;&#53552;&#47484; 16&#52264;&#50896;&#51032; &#44277;&#44036;&#51004;&#47196; &#48148;&#44984;&#45716; &#44032;&#45733;&#54620; &#47784;&#46304; &#49440;&#54805; &#48320;&#54872;&#51032; &#51665;&#54633;&#51077;&#45768;&#45796;. &#51060;&#47088; &#44032;&#49444; &#44277;&#44036;&#51008; &#47588;&#50864; &#51228;&#50557;&#51060; &#47566;&#51004;&#47728;, &#49440;&#54805; &#52789;&#51012; &#44618;&#44172; &#49939;&#50500;&#46020; &#50668;&#51204;&#55176; &#54616;&#45208;&#51032; &#49440;&#54805; &#50672;&#49328;&#51060;&#44592; &#46412;&#47928;&#50640; &#52789;&#51012; &#50668;&#47084; &#44060;&#47196; &#44396;&#49457;&#54616;&#45716; &#51109;&#51216;&#51060; &#50630;&#49845;&#45768;&#45796;.\n",
    "\n",
    "&#44032;&#49444; &#44277;&#44036;&#51012; &#54413;&#48512;&#54616;&#44172; &#47564;&#46308;&#50612; &#52789;&#51012; &#44618;&#44172; &#47564;&#46300;&#45716; &#51109;&#51216;&#51012; &#49332;&#47532;&#44592; &#50948;&#54644;&#49436;&#45716; &#48708;&#49440;&#54805;&#49457; &#46608;&#45716; &#54876;&#49457;&#54868; &#54632;&#49688;&#47484; &#52628;&#44032;&#54644;&#50556; &#54633;&#45768;&#45796;. Relu&#45716; &#46373;&#47084;&#45789;&#50640;&#49436; &#44032;&#51109; &#51064;&#44592; &#51080;&#45716; &#54876;&#49457;&#54868; &#54632;&#49688; &#51473;&#50640; &#54616;&#45208;&#51077;&#45768;&#45796;.\n",
    "___"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (4) &#54617;&#49845;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "partial_x_train, x_val, partial_y_train, y_val = \\\n",
    "    train_test_split(x_train, y_train, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/20\n",
      "20000/20000 [==============================] - 5s 233us/step - loss: 0.4653 - acc: 0.8104 - val_loss: 0.3353 - val_acc: 0.8794\n",
      "Epoch 2/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.2643 - acc: 0.9088 - val_loss: 0.2869 - val_acc: 0.8862\n",
      "Epoch 3/20\n",
      "20000/20000 [==============================] - 2s 88us/step - loss: 0.1989 - acc: 0.9312 - val_loss: 0.2953 - val_acc: 0.8758\n",
      "Epoch 4/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.1652 - acc: 0.9440 - val_loss: 0.2761 - val_acc: 0.8870\n",
      "Epoch 5/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.1382 - acc: 0.9529 - val_loss: 0.3471 - val_acc: 0.8634\n",
      "Epoch 6/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.1180 - acc: 0.9598 - val_loss: 0.3133 - val_acc: 0.8796\n",
      "Epoch 7/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.1028 - acc: 0.9672 - val_loss: 0.3447 - val_acc: 0.8730\n",
      "Epoch 8/20\n",
      "20000/20000 [==============================] - 2s 86us/step - loss: 0.0886 - acc: 0.9704 - val_loss: 0.3597 - val_acc: 0.8754\n",
      "Epoch 9/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.0757 - acc: 0.9751 - val_loss: 0.4088 - val_acc: 0.8720\n",
      "Epoch 10/20\n",
      "20000/20000 [==============================] - 2s 84us/step - loss: 0.0662 - acc: 0.9801 - val_loss: 0.4215 - val_acc: 0.8724\n",
      "Epoch 11/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.0560 - acc: 0.9840 - val_loss: 0.4405 - val_acc: 0.8672\n",
      "Epoch 12/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.0480 - acc: 0.9861 - val_loss: 0.4800 - val_acc: 0.8590\n",
      "Epoch 13/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.0410 - acc: 0.9882 - val_loss: 0.5339 - val_acc: 0.8622\n",
      "Epoch 14/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.0354 - acc: 0.9908 - val_loss: 0.5368 - val_acc: 0.8626\n",
      "Epoch 15/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.0306 - acc: 0.9916 - val_loss: 0.5757 - val_acc: 0.8634\n",
      "Epoch 16/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.0228 - acc: 0.9950 - val_loss: 0.6090 - val_acc: 0.8594\n",
      "Epoch 17/20\n",
      "20000/20000 [==============================] - 2s 85us/step - loss: 0.0250 - acc: 0.9937 - val_loss: 0.6578 - val_acc: 0.8602\n",
      "Epoch 18/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.0158 - acc: 0.9971 - val_loss: 0.6816 - val_acc: 0.8580\n",
      "Epoch 19/20\n",
      "20000/20000 [==============================] - 2s 84us/step - loss: 0.0160 - acc: 0.9963 - val_loss: 0.7594 - val_acc: 0.8536\n",
      "Epoch 20/20\n",
      "20000/20000 [==============================] - 2s 88us/step - loss: 0.0144 - acc: 0.9967 - val_loss: 0.7489 - val_acc: 0.8570\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(partial_x_train, partial_y_train, epochs=20, batch_size=512, validation_data=(x_val, y_val))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (5) &#54984;&#47144;&#44284; &#44160;&#51613; &#49552;&#49892; &#48143; &#51221;&#54869;&#46020; &#44536;&#47532;&#44592;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb2a3013c8>"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "history_dict = history.history\n",
    "loss = history_dict['loss']\n",
    "val_loss = history_dict['val_loss']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.plot(epochs, loss, 'b-o', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'r-o', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb2bc6e550>"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "history_dict = history.history\n",
    "acc = history_dict['acc']\n",
    "val_acc = history_dict['val_acc']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.plot(epochs, loss, 'b-o', label='Training acc')\n",
    "plt.plot(epochs, val_loss, 'r-o', label='Validation acc')\n",
    "plt.title('Training and validation acc')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('acc')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#54788;&#51116; &#45936;&#51060;&#53552;&#51032; &#44221;&#50864;&#45716; &#44284;&#45824;&#51201;&#54633;(overfitting)&#51060; &#48156;&#49373;&#54620; &#49345;&#54889;&#51077;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "**&#44284;&#45824;&#51201;&#54633;&#51012; &#47561;&#45716; &#48169;&#48277;**\n",
    " - &#45348;&#53944;&#50892;&#53356; &#53356;&#44592; &#52629;&#49548; : &#44284;&#45824;&#51201;&#54633;&#51012; &#47561;&#45716; &#44032;&#51109; &#45800;&#49692;&#54620; &#48169;&#48277;&#51077;&#45768;&#45796;. &#47784;&#45944;&#51032; &#53356;&#44592;, &#51593; &#47784;&#45944;&#50640; &#51080;&#45716; &#54617;&#49845; &#54028;&#46972;&#48120;&#53552;&#51032; &#49688;&#47484; &#51460;&#51060;&#45716; &#44163;&#51077;&#45768;&#45796;. &#54028;&#46972;&#48120;&#53552;&#51032; &#49688;&#45716; layers&#51032; &#49688;&#50752; &#44033; layer&#51032; hidden unit&#51032; &#49688;&#50640; &#51032;&#54616;&#50668; &#44208;&#51221;&#46121;&#45768;&#45796;. &#46373;&#47084;&#45789;&#50640;&#49436; &#47784;&#45944;&#50640; &#51080;&#45716; &#54617;&#49845; &#54028;&#46972;&#48120;&#53552;&#51032; &#49688;&#47484; &#51333;&#51333; &#47784;&#45944;&#51032; &#50857;&#47049;(capacity)&#46972;&#44256; &#47568;&#54633;&#45768;&#45796;. &#45936;&#51060;&#53552;&#50640; &#50508;&#47582;&#45716; &#47784;&#45944; &#53356;&#44592;&#47484; &#52286;&#51004;&#47140;&#47732; &#44033;&#44592; &#45796;&#47480; &#44396;&#51312;&#47484; &#44160;&#51613; set&#50640;&#49436; &#54217;&#44032;&#54644; &#48372;&#50500;&#50556; &#54633;&#45768;&#45796;. &#51068;&#48152;&#51201;&#51004;&#47196;&#45716; &#51201;&#51008; &#49688;&#51032; layer&#50752; &#51201;&#51008; &#49688;&#51032; hidden unit&#50640;&#49436;&#48512;&#53552; &#49884;&#51089;&#54633;&#45768;&#45796;.\n",
    " \n",
    " - &#44032;&#51473;&#52824; &#44508;&#51228; &#52628;&#44032; : &#50724;&#52868;&#51032; &#47732;&#46020;&#45216;(&#50612;&#46500; &#44163;&#50640; &#45824;&#54620; &#46160; &#44032;&#51648;&#51032; &#49444;&#47749;&#51060; &#51080;&#45796;&#47732; &#45908; &#51201;&#51008; &#44032;&#51221;&#51060; &#54596;&#50836;&#54620; &#44036;&#45800;&#54620; &#49444;&#47749;&#51060; &#50739;&#51012; &#44163;&#51060;&#46972;&#45716; &#51060;&#47200;). &#50612;&#46500; &#54984;&#47144; &#45936;&#51060;&#53552;&#50752; &#45348;&#53944;&#50892;&#53356; &#44396;&#51312;&#44032; &#51452;&#50668;&#51276;&#51012; &#46412; &#45936;&#51060;&#53552;&#47484; &#49444;&#47749;&#54624; &#49688; &#51080;&#45716; &#44032;&#51473;&#52824; &#44050;&#51032; &#51665;&#54633;&#51008; &#50668;&#47084; &#44060;(&#50668;&#47084; &#44060;&#51032; &#47784;&#45944;)&#51077;&#45768;&#45796;. &#50668;&#44592;&#49436; &#44036;&#45800;&#54620; &#47784;&#45944;&#51060; &#48373;&#51105;&#54620; &#47784;&#45944;&#48372;&#45796; &#45916; &#44284;&#45824;&#51201;&#54633;&#46112; &#44032;&#45733;&#49457;&#51060; &#45458;&#49845;&#45768;&#45796;. &#44036;&#45800;&#54620; &#47784;&#45944;&#51008; &#54028;&#46972;&#48120;&#53552; &#44050; &#48516;&#54252;&#51032; &#50644;&#53944;&#47196;&#54588;&#44032; &#51089;&#51008; &#47784;&#45944;&#51077;&#45768;&#45796;. &#51593; &#44032;&#51473;&#52824;&#44032; &#51089;&#51008; &#44050;&#51012; &#44032;&#51648;&#46020;&#47197; &#44053;&#51228;&#54616;&#45716; &#47784;&#45944;&#51077;&#45768;&#45796;. &#51060;&#47196; &#51064;&#54616;&#50668; &#54028;&#46972;&#48120;&#53552; &#44050;&#51032; &#48516;&#54252;&#44032; &#45908; &#44512;&#51068;&#54616;&#44172; &#46104;&#44256;, &#54028;&#46972;&#48120;&#53552; &#44050; &#48516;&#54252;&#51032; &#50644;&#53944;&#47196;&#54588;&#44032; &#51089;&#44172; &#46121;&#45768;&#45796;. &#51060;&#47484; &#44032;&#51473;&#52824; &#44508;&#51228;(weight regularization)&#46972;&#44256; &#54616;&#47728;, &#45348;&#53944;&#50892;&#53356;&#51032; &#49552;&#49892; &#54632;&#49688;&#50640; &#50672;&#44288;&#46108; &#48708;&#50857;&#51012; &#52628;&#44032;&#54633;&#45768;&#45796;.\n",
    "     - L1 &#44508;&#51228; : &#54028;&#46972;&#48120;&#53552;&#51032; &#51208;&#45824;&#44050;&#50640; &#48708;&#47168;&#54616;&#45716; &#48708;&#50857;&#51060; &#52628;&#44032;&#46121;&#45768;&#45796;.\n",
    "     - L2 &#44508;&#51228; : &#44032;&#51473;&#52824;&#51032; &#51228;&#44273;&#50640; &#48708;&#47168;&#54616;&#45716; &#48708;&#50857;&#51060; &#52628;&#44032;&#46121;&#45768;&#45796;.\n",
    " \n",
    " - Dropout &#52628;&#44032; : &#46300;&#47213;&#50500;&#50883;&#51008; &#49888;&#44221;&#47581;&#51012; &#50948;&#54644; &#49324;&#50857;&#46104;&#45716; &#44508;&#51228; &#44592;&#48277; &#51473;&#50640;&#49436; &#44032;&#51109; &#54952;&#44284;&#51201;&#51060;&#44256; &#45328;&#47532; &#49324;&#50857;&#46104;&#45716; &#48169;&#48277; &#51473; &#54616;&#45208;&#51077;&#45768;&#45796;. &#45348;&#53944;&#50892;&#53356; &#52789;&#50640; &#46300;&#47213;&#50500;&#50883;&#51012; &#51201;&#50857;&#54616;&#47732; &#54984;&#47144;&#54616;&#45716; &#46041;&#50504; &#47924;&#51089;&#50948;&#47196; &#52789;&#51032; &#51068;&#48512; &#52636;&#47141; &#53945;&#49457;&#51012; &#51228;&#50808;&#49884;&#53429;&#45768;&#45796;(0&#51004;&#47196; &#47564;&#46317;&#45768;&#45796;). &#54616;&#51648;&#47564; &#54617;&#49845; &#54980; &#53580;&#49828;&#53944; &#45800;&#44228;&#50640;&#49436;&#45716; &#50612;&#46500; unit&#46020; &#46300;&#47213;&#50500;&#50883;&#46104;&#51648; &#50506;&#49845;&#45768;&#45796;. &#44536; &#45824;&#49888;&#50640; &#52789;&#51032; &#52636;&#47141;&#51012; &#46300;&#47213;&#50500;&#50883; &#48708;&#50984;&#50640; &#48708;&#47168;&#54616;&#50668; &#51460;&#50668; &#51469;&#45768;&#45796;. &#50780;&#45264;&#54616;&#47732; &#54984;&#47144;&#54624; &#46412;&#48372;&#45796; &#45908; &#47566;&#51008; unit&#51060; &#54876;&#49457;&#54868;&#46104;&#44592; &#46412;&#47928;&#51077;&#45768;&#45796;.\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (6) &#44032;&#51473;&#52824; &#44508;&#51228; &#52628;&#44032; &#51201;&#50857;&#54616;&#50668; &#54617;&#49845;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "L2 &#44032;&#51473;&#52824;&#47484; &#52628;&#44032;&#54620; &#47784;&#45944;&#51012; &#49373;&#49457;&#54633;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import regularizers\n",
    "\n",
    "l2_model = models.Sequential()\n",
    "l2_model.add(layers.Dense(16, activation='relu', kernel_regularizer=regularizers.l2()))\n",
    "l2_model.add(layers.Dense(16, activation='relu', kernel_regularizer=regularizers.l2()))\n",
    "l2_model.add(layers.Dense(1, activation='sigmoid'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "l2_model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#54617;&#49845;&#51012; &#51652;&#54665;&#54616;&#47728; &#44284;&#45824;&#51201;&#54633;&#50668;&#48512;&#47484; &#49332;&#54196;&#48389;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/20\n",
      "20000/20000 [==============================] - 5s 257us/step - loss: 0.7581 - acc: 0.7904 - val_loss: 0.6229 - val_acc: 0.8452\n",
      "Epoch 2/20\n",
      "20000/20000 [==============================] - 2s 122us/step - loss: 0.5627 - acc: 0.8764 - val_loss: 0.5388 - val_acc: 0.8720\n",
      "Epoch 3/20\n",
      "20000/20000 [==============================] - 3s 127us/step - loss: 0.5101 - acc: 0.8872 - val_loss: 0.5297 - val_acc: 0.8596\n",
      "Epoch 4/20\n",
      "20000/20000 [==============================] - 3s 140us/step - loss: 0.4760 - acc: 0.8916 - val_loss: 0.4893 - val_acc: 0.8758\n",
      "Epoch 5/20\n",
      "20000/20000 [==============================] - 2s 98us/step - loss: 0.4541 - acc: 0.8932 - val_loss: 0.4716 - val_acc: 0.8758\n",
      "Epoch 6/20\n",
      "20000/20000 [==============================] - 2s 120us/step - loss: 0.4370 - acc: 0.8950 - val_loss: 0.4666 - val_acc: 0.8712\n",
      "Epoch 7/20\n",
      "20000/20000 [==============================] - 2s 122us/step - loss: 0.4277 - acc: 0.8953 - val_loss: 0.4733 - val_acc: 0.8684\n",
      "Epoch 8/20\n",
      "20000/20000 [==============================] - 2s 90us/step - loss: 0.4143 - acc: 0.8997 - val_loss: 0.4653 - val_acc: 0.8644\n",
      "Epoch 9/20\n",
      "20000/20000 [==============================] - 2s 89us/step - loss: 0.4073 - acc: 0.8996 - val_loss: 0.4425 - val_acc: 0.8752\n",
      "Epoch 10/20\n",
      "20000/20000 [==============================] - 2s 88us/step - loss: 0.3962 - acc: 0.9041 - val_loss: 0.4404 - val_acc: 0.8740\n",
      "Epoch 11/20\n",
      "20000/20000 [==============================] - 2s 89us/step - loss: 0.3897 - acc: 0.9050 - val_loss: 0.4369 - val_acc: 0.8736\n",
      "Epoch 12/20\n",
      "20000/20000 [==============================] - 2s 89us/step - loss: 0.3882 - acc: 0.9036 - val_loss: 0.4276 - val_acc: 0.8770\n",
      "Epoch 13/20\n",
      "20000/20000 [==============================] - 2s 89us/step - loss: 0.3818 - acc: 0.9073 - val_loss: 0.5121 - val_acc: 0.8352\n",
      "Epoch 14/20\n",
      "20000/20000 [==============================] - 2s 89us/step - loss: 0.3788 - acc: 0.9080 - val_loss: 0.4225 - val_acc: 0.8760\n",
      "Epoch 15/20\n",
      "20000/20000 [==============================] - 2s 89us/step - loss: 0.3710 - acc: 0.9095 - val_loss: 0.4262 - val_acc: 0.8754\n",
      "Epoch 16/20\n",
      "20000/20000 [==============================] - 2s 90us/step - loss: 0.3673 - acc: 0.9096 - val_loss: 0.4682 - val_acc: 0.8560\n",
      "Epoch 17/20\n",
      "20000/20000 [==============================] - 2s 91us/step - loss: 0.3718 - acc: 0.9050 - val_loss: 0.4577 - val_acc: 0.8584\n",
      "Epoch 18/20\n",
      "20000/20000 [==============================] - 2s 96us/step - loss: 0.3601 - acc: 0.9125 - val_loss: 0.4195 - val_acc: 0.8742\n",
      "Epoch 19/20\n",
      "20000/20000 [==============================] - 2s 103us/step - loss: 0.3568 - acc: 0.9137 - val_loss: 0.4223 - val_acc: 0.8742\n",
      "Epoch 20/20\n",
      "20000/20000 [==============================] - 2s 97us/step - loss: 0.3585 - acc: 0.9097 - val_loss: 0.4113 - val_acc: 0.8782\n"
     ]
    }
   ],
   "source": [
    "l2_history = l2_model.fit(\n",
    "    partial_x_train, partial_y_train, epochs=20, batch_size=512, validation_data=(x_val, y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb36273d30>"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "l2_history_dict = l2_history.history\n",
    "l2_loss = l2_history_dict['loss']\n",
    "l2_val_loss = l2_history_dict['val_loss']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.plot(epochs, l2_loss, 'b-o', label='L2 Training loss')\n",
    "plt.plot(epochs, l2_val_loss, 'r-o', label='L2 Validation loss')\n",
    "plt.title('L2 Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb35fe7978>"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "l2_acc = l2_history_dict['acc']\n",
    "l2_val_acc = l2_history_dict['val_acc']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.plot(epochs, l2_acc, 'b-o', label='Training Acc')\n",
    "plt.plot(epochs, l2_val_acc, 'r-o', label='Validation Acc')\n",
    "plt.title('L2 Training and validation acc')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (7) Dropout &#52628;&#44032; &#51201;&#50857;&#54616;&#50668; &#54617;&#49845;\n",
    "Dropout&#51012; &#52628;&#44032;&#54616;&#50668; &#47784;&#45944;&#51012; &#54617;&#49845;&#54633;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "do_model = models.Sequential()\n",
    "do_model.add(layers.Dense(16, activation='relu'))\n",
    "do_model.add(layers.Dropout(0.5))\n",
    "do_model.add(layers.Dense(16, activation='relu'))\n",
    "do_model.add(layers.Dropout(0.5))\n",
    "do_model.add(layers.Dense(1, activation='sigmoid'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "do_model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#54617;&#49845;&#51012; &#51652;&#54665;&#54616;&#47728; &#44284;&#45824;&#51201;&#54633; &#50668;&#48512;&#47484; &#54032;&#45800;&#54633;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/20\n",
      "20000/20000 [==============================] - 2s 113us/step - loss: 0.6061 - acc: 0.6812 - val_loss: 0.4815 - val_acc: 0.8524\n",
      "Epoch 2/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.4733 - acc: 0.8032 - val_loss: 0.3852 - val_acc: 0.8774\n",
      "Epoch 3/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.3938 - acc: 0.8545 - val_loss: 0.3377 - val_acc: 0.8724\n",
      "Epoch 4/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.3387 - acc: 0.8878 - val_loss: 0.2960 - val_acc: 0.8876\n",
      "Epoch 5/20\n",
      "20000/20000 [==============================] - 2s 88us/step - loss: 0.2926 - acc: 0.9068 - val_loss: 0.2842 - val_acc: 0.8872\n",
      "Epoch 6/20\n",
      "20000/20000 [==============================] - 2s 89us/step - loss: 0.2571 - acc: 0.9186 - val_loss: 0.3049 - val_acc: 0.8832\n",
      "Epoch 7/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.2289 - acc: 0.9290 - val_loss: 0.2904 - val_acc: 0.8824\n",
      "Epoch 8/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.2035 - acc: 0.9356 - val_loss: 0.3095 - val_acc: 0.8850\n",
      "Epoch 9/20\n",
      "20000/20000 [==============================] - 2s 89us/step - loss: 0.1887 - acc: 0.9420 - val_loss: 0.3285 - val_acc: 0.8820\n",
      "Epoch 10/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.1661 - acc: 0.9471 - val_loss: 0.3687 - val_acc: 0.8824\n",
      "Epoch 11/20\n",
      "20000/20000 [==============================] - 2s 88us/step - loss: 0.1555 - acc: 0.9525 - val_loss: 0.3686 - val_acc: 0.8816\n",
      "Epoch 12/20\n",
      "20000/20000 [==============================] - 2s 89us/step - loss: 0.1425 - acc: 0.9569 - val_loss: 0.3945 - val_acc: 0.8802\n",
      "Epoch 13/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.1308 - acc: 0.9597 - val_loss: 0.4237 - val_acc: 0.8772\n",
      "Epoch 14/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.1223 - acc: 0.9607 - val_loss: 0.4504 - val_acc: 0.8778\n",
      "Epoch 15/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.1189 - acc: 0.9635 - val_loss: 0.4767 - val_acc: 0.8758\n",
      "Epoch 16/20\n",
      "20000/20000 [==============================] - 2s 88us/step - loss: 0.1092 - acc: 0.9669 - val_loss: 0.5289 - val_acc: 0.8750\n",
      "Epoch 17/20\n",
      "20000/20000 [==============================] - 2s 88us/step - loss: 0.1042 - acc: 0.9677 - val_loss: 0.5264 - val_acc: 0.8794\n",
      "Epoch 18/20\n",
      "20000/20000 [==============================] - 2s 88us/step - loss: 0.1004 - acc: 0.9700 - val_loss: 0.5425 - val_acc: 0.8780\n",
      "Epoch 19/20\n",
      "20000/20000 [==============================] - 2s 87us/step - loss: 0.0989 - acc: 0.9704 - val_loss: 0.5613 - val_acc: 0.8752\n",
      "Epoch 20/20\n",
      "20000/20000 [==============================] - 2s 88us/step - loss: 0.0949 - acc: 0.9739 - val_loss: 0.6341 - val_acc: 0.8736\n"
     ]
    }
   ],
   "source": [
    "do_history = do_model.fit(\n",
    "    partial_x_train, partial_y_train, epochs=20, batch_size=512, validation_data=(x_val, y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb2f84df60>"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "do_history_dict = do_history.history\n",
    "do_loss = do_history_dict['loss']\n",
    "do_val_loss = do_history_dict['val_loss']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.plot(epochs, do_loss, 'b-o', label='Dropout Training loss')\n",
    "plt.plot(epochs, do_val_loss, 'r-o', label='Dropout Validation loss')\n",
    "plt.title('Dropout Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb2fa3c278>"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "do_history_dict = do_history.history\n",
    "do_acc = do_history_dict['acc']\n",
    "do_val_acc = do_history_dict['val_acc']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.plot(epochs, do_acc, 'b-o', label='Dropout Training acc')\n",
    "plt.plot(epochs, do_val_acc, 'r-o', label='Dropout Validation acc')\n",
    "plt.title('Dropout Training and validation acc')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('acc')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (8) Prediction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "&#49352;&#47196;&#50868; &#47784;&#45944;&#50640; &#45824;&#54616;&#50668; &#50696;&#52769;&#51012; &#54616;&#45716; &#48169;&#48277;&#51008; &#45796;&#51020;&#44284; &#44057;&#49845;&#45768;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.01350514],\n",
       "       [1.        ],\n",
       "       [0.94708514],\n",
       "       ...,\n",
       "       [0.06980336],\n",
       "       [0.00491353],\n",
       "       [0.2844553 ]], dtype=float32)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.predict(x_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> &#52376;&#51020; &#54617;&#49845; &#47784;&#45944;&#51032; &#52572;&#51333; &#49457;&#45733;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25000/25000 [==============================] - 2s 74us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.84676"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "_, acc = model.evaluate(x_test, y_test)\n",
    "acc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> L2 &#51201;&#50857; &#47784;&#45944;&#51032; &#52572;&#51333; &#49457;&#45733;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25000/25000 [==============================] - 5s 186us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.88104"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "_, l2_acc = l2_model.evaluate(x_test, y_test)\n",
    "l2_acc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Dropout &#51201;&#50857; &#47784;&#45944;&#51032; &#52572;&#51333; &#49457;&#45733;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25000/25000 [==============================] - 2s 63us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.87004"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "_, do_acc = do_model.evaluate(x_test, y_test)\n",
    "do_acc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Keras&#51032; DNN&#51012; &#51060;&#50857;&#54616;&#50668; &#48372;&#49828;&#53556; &#51452;&#53469; &#44032;&#44201; &#50696;&#52769;&#54616;&#44592; (Regression)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (1) &#45936;&#51060;&#53552; &#47196;&#46377;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.datasets import boston_housing\n",
    "\n",
    "(train_data, train_targets), (test_data, test_targets) = boston_housing.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(404, 13)\n",
      "(404,)\n",
      "(102, 13)\n",
      "(102,)\n"
     ]
    }
   ],
   "source": [
    "print(train_data.shape)\n",
    "print(train_targets.shape)\n",
    "print(test_data.shape)\n",
    "print(test_targets.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  1.23247,   0.     ,   8.14   ,   0.     ,   0.538  ,   6.142  ,\n",
       "        91.7    ,   3.9769 ,   4.     , 307.     ,  21.     , 396.9    ,\n",
       "        18.72   ])"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (2) &#45936;&#51060;&#53552; scailing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean = train_data.mean(axis=0)\n",
    "train_data -= mean\n",
    "std = train_data.std(axis=0)\n",
    "train_data /= std\n",
    "\n",
    "test_data -= mean\n",
    "test_data /= std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.27224633, -0.48361547, -0.43576161, -0.25683275, -0.1652266 ,\n",
       "       -0.1764426 ,  0.81306188,  0.1166983 , -0.62624905, -0.59517003,\n",
       "        1.14850044,  0.44807713,  0.8252202 ])"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "partial_train_x, val_x, partial_train_y, val_y = \\\n",
    "    train_test_split(train_data, train_targets, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (3) Dropout&#51012; &#51201;&#50857;&#54616;&#50668; &#47784;&#45944; &#54617;&#49845;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "do_model = models.Sequential()\n",
    "do_model.add(layers.Dense(64, activation='relu', input_shape=(train_data.shape[1], )))\n",
    "do_model.add(layers.Dropout(0.5))\n",
    "do_model.add(layers.Dense(64, activation='relu'))\n",
    "do_model.add(layers.Dropout(0.5))\n",
    "do_model.add(layers.Dense(1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "do_model.compile(optimizer='adam', loss='mse', metrics=['mae'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 323 samples, validate on 81 samples\n",
      "Epoch 1/15\n",
      "323/323 [==============================] - 0s 1ms/step - loss: 615.1355 - mean_absolute_error: 22.8565 - val_loss: 458.3810 - val_mean_absolute_error: 20.1482\n",
      "Epoch 2/15\n",
      "323/323 [==============================] - 0s 44us/step - loss: 595.0456 - mean_absolute_error: 22.3929 - val_loss: 442.1359 - val_mean_absolute_error: 19.7339\n",
      "Epoch 3/15\n",
      "323/323 [==============================] - 0s 45us/step - loss: 574.6992 - mean_absolute_error: 21.8846 - val_loss: 426.1490 - val_mean_absolute_error: 19.3136\n",
      "Epoch 4/15\n",
      "323/323 [==============================] - 0s 59us/step - loss: 550.2793 - mean_absolute_error: 21.4078 - val_loss: 409.9798 - val_mean_absolute_error: 18.8758\n",
      "Epoch 5/15\n",
      "323/323 [==============================] - 0s 47us/step - loss: 536.5190 - mean_absolute_error: 21.0277 - val_loss: 393.1164 - val_mean_absolute_error: 18.4071\n",
      "Epoch 6/15\n",
      "323/323 [==============================] - 0s 53us/step - loss: 508.8611 - mean_absolute_error: 20.4011 - val_loss: 374.8941 - val_mean_absolute_error: 17.8896\n",
      "Epoch 7/15\n",
      "323/323 [==============================] - 0s 45us/step - loss: 487.8515 - mean_absolute_error: 19.8998 - val_loss: 354.8473 - val_mean_absolute_error: 17.3065\n",
      "Epoch 8/15\n",
      "323/323 [==============================] - 0s 39us/step - loss: 466.8695 - mean_absolute_error: 19.2835 - val_loss: 332.6787 - val_mean_absolute_error: 16.6733\n",
      "Epoch 9/15\n",
      "323/323 [==============================] - 0s 43us/step - loss: 433.8161 - mean_absolute_error: 18.4796 - val_loss: 307.9683 - val_mean_absolute_error: 15.9688\n",
      "Epoch 10/15\n",
      "323/323 [==============================] - 0s 53us/step - loss: 408.0958 - mean_absolute_error: 17.7469 - val_loss: 280.8030 - val_mean_absolute_error: 15.1758\n",
      "Epoch 11/15\n",
      "323/323 [==============================] - 0s 41us/step - loss: 365.0056 - mean_absolute_error: 16.6681 - val_loss: 251.4380 - val_mean_absolute_error: 14.2834\n",
      "Epoch 12/15\n",
      "323/323 [==============================] - 0s 56us/step - loss: 327.4106 - mean_absolute_error: 15.6856 - val_loss: 220.7128 - val_mean_absolute_error: 13.2725\n",
      "Epoch 13/15\n",
      "323/323 [==============================] - 0s 47us/step - loss: 293.5142 - mean_absolute_error: 14.5735 - val_loss: 190.1204 - val_mean_absolute_error: 12.1641\n",
      "Epoch 14/15\n",
      "323/323 [==============================] - 0s 45us/step - loss: 248.1203 - mean_absolute_error: 13.2052 - val_loss: 160.6790 - val_mean_absolute_error: 10.9926\n",
      "Epoch 15/15\n",
      "323/323 [==============================] - 0s 59us/step - loss: 215.5930 - mean_absolute_error: 12.0478 - val_loss: 132.6605 - val_mean_absolute_error: 9.8003\n"
     ]
    }
   ],
   "source": [
    "do_history = do_model.fit(\n",
    "    partial_train_x, partial_train_y, epochs=15, batch_size=64, validation_data=[val_x, val_y])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb340c8390>"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "do_history_dict = do_history.history\n",
    "do_loss = do_history_dict['loss']\n",
    "do_val_loss = do_history_dict['val_loss']\n",
    "\n",
    "epochs = range(1, len(do_loss) + 1)\n",
    "\n",
    "plt.plot(epochs, do_loss, 'b-o', label='Dropout Training loss')\n",
    "plt.plot(epochs, do_val_loss, 'r-o', label='Dropout Validation loss')\n",
    "plt.title('Dropout Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb392622b0>"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "do_history_dict = do_history.history\n",
    "do_mae = do_history_dict['mean_absolute_error']\n",
    "do_val_mae = do_history_dict['val_mean_absolute_error']\n",
    "\n",
    "epochs = range(1, len(do_mae) + 1)\n",
    "\n",
    "plt.plot(epochs, do_mae, 'b-o', label='Dropout Training mae')\n",
    "plt.plot(epochs, do_val_mae, 'r-o', label='Dropout Validation mae')\n",
    "plt.title('Dropout Training and validation mae')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('mae')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (4) L2 &#44508;&#51228;&#54868;&#47484; &#51201;&#50857;&#54616;&#50668; &#54617;&#49845;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "l2_model = models.Sequential()\n",
    "l2_model.add(layers.Dense(64, \n",
    "                          activation='relu', \n",
    "                          input_shape=(train_data.shape[1], ), \n",
    "                          kernel_regularizer=regularizers.l2()))\n",
    "l2_model.add(layers.Dense(64, \n",
    "                          activation='relu',\n",
    "                          kernel_regularizer=regularizers.l2()))\n",
    "l2_model.add(layers.Dense(1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "l2_model.compile(optimizer='adam', loss='mse', metrics=['mae'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 323 samples, validate on 81 samples\n",
      "Epoch 1/15\n",
      "323/323 [==============================] - 0s 1ms/step - loss: 601.3030 - mean_absolute_error: 22.5638 - val_loss: 447.8407 - val_mean_absolute_error: 19.8846\n",
      "Epoch 2/15\n",
      "323/323 [==============================] - 0s 39us/step - loss: 577.5930 - mean_absolute_error: 22.0327 - val_loss: 427.5497 - val_mean_absolute_error: 19.3763\n",
      "Epoch 3/15\n",
      "323/323 [==============================] - 0s 36us/step - loss: 552.5320 - mean_absolute_error: 21.4724 - val_loss: 405.5666 - val_mean_absolute_error: 18.8199\n",
      "Epoch 4/15\n",
      "323/323 [==============================] - 0s 53us/step - loss: 524.9810 - mean_absolute_error: 20.8493 - val_loss: 380.7875 - val_mean_absolute_error: 18.1822\n",
      "Epoch 5/15\n",
      "323/323 [==============================] - 0s 48us/step - loss: 492.0087 - mean_absolute_error: 20.1036 - val_loss: 352.8376 - val_mean_absolute_error: 17.4417\n",
      "Epoch 6/15\n",
      "323/323 [==============================] - 0s 53us/step - loss: 455.0300 - mean_absolute_error: 19.2255 - val_loss: 321.1123 - val_mean_absolute_error: 16.5716\n",
      "Epoch 7/15\n",
      "323/323 [==============================] - 0s 50us/step - loss: 412.4494 - mean_absolute_error: 18.1858 - val_loss: 285.9044 - val_mean_absolute_error: 15.5635\n",
      "Epoch 8/15\n",
      "323/323 [==============================] - 0s 55us/step - loss: 365.3852 - mean_absolute_error: 16.9691 - val_loss: 247.3850 - val_mean_absolute_error: 14.3844\n",
      "Epoch 9/15\n",
      "323/323 [==============================] - 0s 50us/step - loss: 312.2033 - mean_absolute_error: 15.5151 - val_loss: 207.5302 - val_mean_absolute_error: 13.0372\n",
      "Epoch 10/15\n",
      "323/323 [==============================] - 0s 70us/step - loss: 259.0007 - mean_absolute_error: 13.8928 - val_loss: 167.2648 - val_mean_absolute_error: 11.4934\n",
      "Epoch 11/15\n",
      "323/323 [==============================] - 0s 38us/step - loss: 206.6567 - mean_absolute_error: 12.0501 - val_loss: 128.6479 - val_mean_absolute_error: 9.8957\n",
      "Epoch 12/15\n",
      "323/323 [==============================] - 0s 50us/step - loss: 158.4191 - mean_absolute_error: 10.1349 - val_loss: 96.4848 - val_mean_absolute_error: 8.3859\n",
      "Epoch 13/15\n",
      "323/323 [==============================] - 0s 36us/step - loss: 118.7918 - mean_absolute_error: 8.3438 - val_loss: 71.5792 - val_mean_absolute_error: 7.0183\n",
      "Epoch 14/15\n",
      "323/323 [==============================] - 0s 46us/step - loss: 91.8106 - mean_absolute_error: 7.1438 - val_loss: 55.7590 - val_mean_absolute_error: 6.0939\n",
      "Epoch 15/15\n",
      "323/323 [==============================] - 0s 39us/step - loss: 75.7886 - mean_absolute_error: 6.4223 - val_loss: 45.2276 - val_mean_absolute_error: 5.4943\n"
     ]
    }
   ],
   "source": [
    "l2_history = l2_model.fit(\n",
    "    partial_train_x, partial_train_y, epochs=15, batch_size=64, validation_data=[val_x, val_y])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb38005ac8>"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "l2_history_dict = l2_history.history\n",
    "l2_loss = l2_history_dict['loss']\n",
    "l2_val_loss = l2_history_dict['val_loss']\n",
    "\n",
    "epochs = range(1, len(l2_loss) + 1)\n",
    "\n",
    "plt.plot(epochs, do_loss, 'b-o', label='L2 Training loss')\n",
    "plt.plot(epochs, do_val_loss, 'r-o', label='L2 Validation loss')\n",
    "plt.title('L2 Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0xb37e139e8>"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "l2_history_dict = l2_history.history\n",
    "l2_mae = l2_history_dict['mean_absolute_error']\n",
    "l2_val_mae = l2_history_dict['val_mean_absolute_error']\n",
    "\n",
    "epochs = range(1, len(l2_mae) + 1)\n",
    "\n",
    "plt.plot(epochs, l2_mae, 'b-o', label='L2 Training mae')\n",
    "plt.plot(epochs, l2_val_mae, 'r-o', label='L2 Validation mae')\n",
    "plt.title('Dropout Training and validation mae')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('mae')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (5) Prediction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Dropout&#51012; &#51060;&#50857;&#54620; &#47784;&#45944;&#51032; &#52572;&#51333; &#49457;&#45733;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "102/102 [==============================] - 0s 55us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "11.577999152389227"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "_, acc = do_model.evaluate(test_data, test_targets)\n",
    "acc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> L2 &#44508;&#51228;&#54868;&#47484; &#49324;&#50857;&#54620; &#47784;&#45944;&#51032; &#52572;&#51333; &#49457;&#45733;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "102/102 [==============================] - 0s 55us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "6.579368797003054"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "_, acc = l2_model.evaluate(test_data, test_targets)\n",
    "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}