Node 25. Regularization.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "sophisticated-basketball",
   "metadata": {},
   "source": [
    "# 학습목표\n",
    "\n",
    "- 정칙화(Regularization)의 개념을 이해하고 정규화(Normalization)와 구분합니다.\n",
    "- L1 regularization과 L2 regularization의 차이를 설명합니다.\n",
    "- 실습을 통하여 Lp norm, Dropout, Batch Normalization에 대해 학습합니다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "czech-southwest",
   "metadata": {},
   "source": [
    "## Regularization과 Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "written-saturday",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sepal length (cm)</th>\n",
       "      <th>sepal width (cm)</th>\n",
       "      <th>petal length (cm)</th>\n",
       "      <th>petal width (cm)</th>\n",
       "      <th>species</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.1</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4.9</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4.7</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.2</td>\n",
       "      <td>setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4.6</td>\n",
       "      <td>3.1</td>\n",
       "      <td>1.5</td>\n",
       "      <td>0.2</td>\n",
       "      <td>setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>3.6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>setosa</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   sepal length (cm)  sepal width (cm)  petal length (cm)  petal width (cm)  \\\n",
       "0                5.1               3.5                1.4               0.2   \n",
       "1                4.9               3.0                1.4               0.2   \n",
       "2                4.7               3.2                1.3               0.2   \n",
       "3                4.6               3.1                1.5               0.2   \n",
       "4                5.0               3.6                1.4               0.2   \n",
       "\n",
       "  species  \n",
       "0  setosa  \n",
       "1  setosa  \n",
       "2  setosa  \n",
       "3  setosa  \n",
       "4  setosa  "
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.datasets import load_iris\n",
    "import pandas as pd \n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "iris = load_iris()\n",
    "iris_df = pd.DataFrame(data=iris.data, columns=iris.feature_names)\n",
    "target_df = pd.DataFrame(data=iris.target, columns=['species'])\n",
    "\n",
    "# 0, 1, 2로 되어있는 target 데이터를 \n",
    "# 알아보기 쉽게 'setosa', 'versicolor', 'virginica'로 바꿉니다 \n",
    "def converter(species):\n",
    "    if species == 0:\n",
    "        return 'setosa'\n",
    "    elif species == 1:\n",
    "        return 'versicolor'\n",
    "    else:\n",
    "        return 'virginica'\n",
    "\n",
    "target_df['species'] = target_df['species'].apply(converter)\n",
    "\n",
    "iris_df = pd.concat([iris_df, target_df], axis=1)\n",
    "iris_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "dominant-essex",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[6.0, 5.1, 5.9, 5.6, 5.8, 6.6, 4.5, 6.3, 5.8, 6.1, 5.1, 5.3, 5.5, 5.0, 5.1, 5.3, 5.5, 6.7, 6.9, 5.0, 5.7, 4.9, 6.7, 4.9, 5.7, 6.0, 4.8, 4.9, 5.6, 5.8, 6.1, 6.4, 5.6, 5.1, 5.6, 6.1, 5.6, 5.5, 4.8, 5.4, 5.6, 5.1, 5.1, 5.9, 5.7, 5.2, 5.0, 5.2, 5.4, 5.1]\n",
      "[6.3, 5.8, 7.1, 6.3, 6.5, 7.6, 4.9, 7.3, 6.7, 7.2, 6.5, 6.4, 6.8, 5.7, 5.8, 6.4, 6.5, 7.7, 7.7, 6.0, 6.9, 5.6, 7.7, 6.3, 6.7, 7.2, 6.2, 6.1, 6.4, 7.2, 7.4, 7.9, 6.4, 6.3, 6.1, 7.7, 6.3, 6.4, 6.0, 6.9, 6.7, 6.9, 5.8, 6.8, 6.7, 6.7, 6.3, 6.5, 6.2, 5.9]\n"
     ]
    }
   ],
   "source": [
    "X = [iris_df['petal length (cm)'][a] for a in iris_df.index if iris_df['species'][a]=='virginica']\n",
    "Y = [iris_df['sepal length (cm)'][a] for a in iris_df.index if iris_df['species'][a]=='virginica']\n",
    "\n",
    "print(X)\n",
    "print(Y)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "retired-harrison",
   "metadata": {},
   "source": [
    "> #### 직관적인 이해를 위해 scatterplot 으로 그려보기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "increasing-invite",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(X,Y)\n",
    "plt.title('petal-sepal scatter before normalization') \n",
    "plt.xlabel('petal length (cm)')\n",
    "plt.ylabel('sepal length (cm)')\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "oriented-initial",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.preprocessing import minmax_scale\n",
    "\n",
    "X_scale = minmax_scale(X)\n",
    "Y_scale = minmax_scale(Y)\n",
    "\n",
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(X_scale,Y_scale)\n",
    "plt.title('petal-sepal scatter after normalization') \n",
    "plt.xlabel('petal length (cm)')\n",
    "plt.ylabel('sepal length (cm)')\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "historical-industry",
   "metadata": {},
   "source": [
    "--------------------\n",
    "--------------------\n",
    "--------------------\n",
    "--------------------"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "sublime-dietary",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "\n",
    "plt.figure()\n",
    "sns.scatterplot(x=X, y=Y)\n",
    "sns.scatterplot(x=X_scale, y=Y_scale)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "useful-switzerland",
   "metadata": {},
   "source": [
    "--------------------\n",
    "--------------------\n",
    "--------------------\n",
    "--------------------"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "protecting-caribbean",
   "metadata": {},
   "source": [
    "> #### regularization에 대해 알아보기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "deadly-cooler",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "기울기 : 1.00, 절편 : 1.06\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LinearRegression\n",
    "import numpy as np \n",
    "\n",
    "X = np.array(X)\n",
    "Y = np.array(Y)\n",
    "\n",
    "# Iris Dataset을 Linear Regression으로 학습합니다. \n",
    "linear= LinearRegression()\n",
    "linear.fit(X.reshape(-1,1), Y)\n",
    "\n",
    "# Linear Regression의 기울기와 절편을 확인합니다. \n",
    "a, b=linear.coef_, linear.intercept_\n",
    "print(\"기울기 : %0.2f, 절편 : %0.2f\" %(a,b))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "gentle-powder",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.99573864]\n",
      "1.059659090909089\n"
     ]
    }
   ],
   "source": [
    "print(a)\n",
    "print(b)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "precious-jacksonville",
   "metadata": {},
   "source": [
    "> 위의 기울기, 절편을 통해 fit된 직선을 만들어보기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "quality-george",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(X,Y)\n",
    "plt.plot(X,linear.predict(X.reshape(-1,1)),'-b')\n",
    "plt.title('petal-sepal scatter with linear regression') \n",
    "plt.xlabel('petal length (cm)')\n",
    "plt.ylabel('sepal length (cm)')\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "green-lloyd",
   "metadata": {},
   "source": [
    "### L1, L2 Regularization 사용"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "appropriate-stand",
   "metadata": {},
   "source": [
    "> L1을 사용하는 Lasso 먼저 확인"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "empty-buyer",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "기울기 : 0.00, 절편 : 6.59\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#L1 regularization은 Lasso로 import 합니다.\n",
    "from sklearn.linear_model import Lasso\n",
    "\n",
    "L1 = Lasso()\n",
    "L1.fit(X.reshape(-1,1), Y)\n",
    "a, b=L1.coef_, L1.intercept_\n",
    "print(\"기울기 : %0.2f, 절편 : %0.2f\" %(a,b))\n",
    "\n",
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(X,Y)\n",
    "plt.plot(X,L1.predict(X.reshape(-1,1)),'-b')\n",
    "plt.title('petal-sepal scatter with L1 regularization(Lasso)') \n",
    "plt.xlabel('petal length (cm)')\n",
    "plt.ylabel('sepal length (cm)')\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "vanilla-tumor",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.]\n",
      "6.587999999999998\n"
     ]
    }
   ],
   "source": [
    "print(a)\n",
    "print(b)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "separate-charles",
   "metadata": {},
   "source": [
    "> - 기울기가 0이다.\n",
    "> - 방정식을 제대로 풀지 못하고 있다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "outstanding-colon",
   "metadata": {},
   "source": [
    "> L2를 쓰는 Ridge로 확인"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "literary-methodology",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "기울기 : 0.93, 절편 : 1.41\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#L2 regularization은 Ridge로 import 합니다. \n",
    "from sklearn.linear_model import Ridge\n",
    "\n",
    "L2 = Ridge()\n",
    "L2.fit(X.reshape(-1,1), Y)\n",
    "a, b = L2.coef_, L2.intercept_\n",
    "print(\"기울기 : %0.2f, 절편 : %0.2f\" %(a,b))\n",
    "\n",
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(X,Y)\n",
    "plt.plot(X,L2.predict(X.reshape(-1,1)),'-b')\n",
    "plt.title('petal-sepal scatter with L2 regularization(Ridge)') \n",
    "plt.xlabel('petal length (cm)')\n",
    "plt.ylabel('sepal length (cm)')\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "practical-reflection",
   "metadata": {},
   "source": [
    "> #### note:\n",
    "> - `Lasso`, `Ridge`가 다르게 나오는데 `Ridge`가 `Linear Regression`과 비슷하게 나오는 이유는\n",
    "> - Linear Regression이 `L2 norm`과 관련 있기 때문이다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "coral-insulin",
   "metadata": {},
   "source": [
    "## L1 Regularization"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "normal-spring",
   "metadata": {},
   "source": [
    "- L1 norm 에 대한 이해를 통해 `Lasso`를 언제 쓰는지 알아보자"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "juvenile-ultimate",
   "metadata": {},
   "source": [
    "#### 이전 스텝에서 왜 L1 Regularization 만 답이 나오지 않았을까?"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "buried-round",
   "metadata": {},
   "source": [
    "`p=1`인 경우 $\\beta_0$ 에 대해 미분하는 과정에서 $\\lambda$가 사라지므로 Regularization의 효과를 볼 수 없습니다. $X$가 **1차원 값인 선형회귀분석 같은 경우에는 L1 Regularization이 의미가 없다**는 것을 말합니다. 그러므로, L1 Regularization을 사용할 때는 **X가 2차원 이상인 여러 컬럼 값이 있는 데이터일 때** 실제 효과를 볼 수 있습니다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "hawaiian-while",
   "metadata": {},
   "source": [
    "### 컬럼 수가 많은 데이터에서의 L1 regularization 비교"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "postal-difficulty",
   "metadata": {},
   "source": [
    "- 총 13개의 컬럼을 갖는 wind data로 비교해보자"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "incorrect-highland",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_wine\n",
    "\n",
    "wine = load_wine()\n",
    "wine_df = pd.DataFrame(data=wine.data, columns=wine.feature_names)\n",
    "target_df = pd.DataFrame(data=wine.target, columns=['Y'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "manual-electric",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>alcohol</th>\n",
       "      <th>malic_acid</th>\n",
       "      <th>ash</th>\n",
       "      <th>alcalinity_of_ash</th>\n",
       "      <th>magnesium</th>\n",
       "      <th>total_phenols</th>\n",
       "      <th>flavanoids</th>\n",
       "      <th>nonflavanoid_phenols</th>\n",
       "      <th>proanthocyanins</th>\n",
       "      <th>color_intensity</th>\n",
       "      <th>hue</th>\n",
       "      <th>od280/od315_of_diluted_wines</th>\n",
       "      <th>proline</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>14.23</td>\n",
       "      <td>1.71</td>\n",
       "      <td>2.43</td>\n",
       "      <td>15.6</td>\n",
       "      <td>127.0</td>\n",
       "      <td>2.80</td>\n",
       "      <td>3.06</td>\n",
       "      <td>0.28</td>\n",
       "      <td>2.29</td>\n",
       "      <td>5.64</td>\n",
       "      <td>1.04</td>\n",
       "      <td>3.92</td>\n",
       "      <td>1065.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>13.20</td>\n",
       "      <td>1.78</td>\n",
       "      <td>2.14</td>\n",
       "      <td>11.2</td>\n",
       "      <td>100.0</td>\n",
       "      <td>2.65</td>\n",
       "      <td>2.76</td>\n",
       "      <td>0.26</td>\n",
       "      <td>1.28</td>\n",
       "      <td>4.38</td>\n",
       "      <td>1.05</td>\n",
       "      <td>3.40</td>\n",
       "      <td>1050.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>13.16</td>\n",
       "      <td>2.36</td>\n",
       "      <td>2.67</td>\n",
       "      <td>18.6</td>\n",
       "      <td>101.0</td>\n",
       "      <td>2.80</td>\n",
       "      <td>3.24</td>\n",
       "      <td>0.30</td>\n",
       "      <td>2.81</td>\n",
       "      <td>5.68</td>\n",
       "      <td>1.03</td>\n",
       "      <td>3.17</td>\n",
       "      <td>1185.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>14.37</td>\n",
       "      <td>1.95</td>\n",
       "      <td>2.50</td>\n",
       "      <td>16.8</td>\n",
       "      <td>113.0</td>\n",
       "      <td>3.85</td>\n",
       "      <td>3.49</td>\n",
       "      <td>0.24</td>\n",
       "      <td>2.18</td>\n",
       "      <td>7.80</td>\n",
       "      <td>0.86</td>\n",
       "      <td>3.45</td>\n",
       "      <td>1480.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>13.24</td>\n",
       "      <td>2.59</td>\n",
       "      <td>2.87</td>\n",
       "      <td>21.0</td>\n",
       "      <td>118.0</td>\n",
       "      <td>2.80</td>\n",
       "      <td>2.69</td>\n",
       "      <td>0.39</td>\n",
       "      <td>1.82</td>\n",
       "      <td>4.32</td>\n",
       "      <td>1.04</td>\n",
       "      <td>2.93</td>\n",
       "      <td>735.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   alcohol  malic_acid   ash  alcalinity_of_ash  magnesium  total_phenols  \\\n",
       "0    14.23        1.71  2.43               15.6      127.0           2.80   \n",
       "1    13.20        1.78  2.14               11.2      100.0           2.65   \n",
       "2    13.16        2.36  2.67               18.6      101.0           2.80   \n",
       "3    14.37        1.95  2.50               16.8      113.0           3.85   \n",
       "4    13.24        2.59  2.87               21.0      118.0           2.80   \n",
       "\n",
       "   flavanoids  nonflavanoid_phenols  proanthocyanins  color_intensity   hue  \\\n",
       "0        3.06                  0.28             2.29             5.64  1.04   \n",
       "1        2.76                  0.26             1.28             4.38  1.05   \n",
       "2        3.24                  0.30             2.81             5.68  1.03   \n",
       "3        3.49                  0.24             2.18             7.80  0.86   \n",
       "4        2.69                  0.39             1.82             4.32  1.04   \n",
       "\n",
       "   od280/od315_of_diluted_wines  proline  \n",
       "0                          3.92   1065.0  \n",
       "1                          3.40   1050.0  \n",
       "2                          3.17   1185.0  \n",
       "3                          3.45   1480.0  \n",
       "4                          2.93    735.0  "
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "wine_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "immediate-elephant",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Y\n",
       "0  0\n",
       "1  0\n",
       "2  0\n",
       "3  0\n",
       "4  0"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "coupled-sword",
   "metadata": {},
   "source": [
    "> - 먼저 linear regression으로 문제를 풀어보고\n",
    "> - 해당 문제에서의 mean absolute error, mean sqaured error, root mean sqaured error를 구해보자"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "academic-elder",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of linear regression\n",
      "Mean Absolute Error: 0.25128973939722626\n",
      "Mean Squared Error: 0.1062458740952556\n",
      "Mean Root Squared Error: 0.32595379134971814\n",
      "\n",
      "\n",
      " coefficient linear regression\n",
      "[[-8.09017190e-02  4.34817880e-02 -1.18857931e-01  3.65705449e-02\n",
      "  -4.68014203e-04  1.41423581e-01 -4.54107854e-01 -5.13172664e-01\n",
      "   9.69318443e-02  5.34311136e-02 -1.27626604e-01 -2.91381844e-01\n",
      "  -5.72238959e-04]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.metrics import mean_absolute_error, mean_squared_error\n",
    "\n",
    "# 데이터를 준비하고\n",
    "X_train, X_test, y_train, y_test = train_test_split(wine_df, target_df, test_size=0.3, random_state=101)\n",
    "\n",
    "# 모델을 훈련시킵니다.\n",
    "model = LinearRegression()\n",
    "model.fit(X_train, y_train)\n",
    "\n",
    "# 테스트를 해볼까요?\n",
    "model.predict(X_test)\n",
    "pred = model.predict(X_test)\n",
    "\n",
    "# 테스트 결과는 이렇습니다!\n",
    "print(\"result of linear regression\")\n",
    "print('Mean Absolute Error:', mean_absolute_error(y_test, pred))\n",
    "print('Mean Squared Error:', mean_squared_error(y_test, pred))\n",
    "print('Mean Root Squared Error:', np.sqrt(mean_squared_error(y_test, pred)))\n",
    "\n",
    "print(\"\\n\\n coefficient linear regression\")\n",
    "print(model.coef_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "portuguese-league",
   "metadata": {},
   "source": [
    "> - L1 regularization으로 문제를 풀어보자"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "overall-review",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of Lasso\n",
      "Mean Absolute Error: 0.24233731936122138\n",
      "Mean Squared Error: 0.0955956894578189\n",
      "Mean Root Squared Error: 0.3091855259513597\n",
      "\n",
      "\n",
      " coefficient of Lasso\n",
      "[-0.          0.01373795 -0.          0.03065716  0.00154719 -0.\n",
      " -0.34143614 -0.          0.          0.06755943 -0.         -0.14558153\n",
      " -0.00089635]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import Lasso\n",
    "from sklearn.metrics import mean_absolute_error, mean_squared_error\n",
    "\n",
    "# 모델을 준비하고 훈련시킵니다.\n",
    "L1 = Lasso(alpha=0.05)\n",
    "L1.fit(X_train, y_train)\n",
    "\n",
    "# 테스트를 해봅시다.\n",
    "pred = L1.predict(X_test)\n",
    "\n",
    "# 모델 성능은 얼마나 좋을까요?\n",
    "print(\"result of Lasso\")\n",
    "print('Mean Absolute Error:', mean_absolute_error(y_test, pred))\n",
    "print('Mean Squared Error:', mean_squared_error(y_test, pred))\n",
    "print('Mean Root Squared Error:', np.sqrt(mean_squared_error(y_test, pred)))\n",
    "\n",
    "print(\"\\n\\n coefficient of Lasso\")\n",
    "print(L1.coef_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "coupled-maple",
   "metadata": {},
   "source": [
    "### 결과 분석"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fourth-tragedy",
   "metadata": {},
   "source": [
    "#### 아이펠 설명\n",
    "\n",
    "coefficient 부분을 보시면 Linear Regression과 L1 Regularization의 차이가 좀 더 두드러짐을 알 수 있습니다. Linear Regression에서는 모든 컬럼의 가중치를 탐색하여 구하는 반면, L1 Regularization에서는 총 13개 중 7개를 제외한 나머지의 값들이 모두 0임을 확인할 수 있습니다. Error 부분에서는 큰 차이가 없었지만, 우리가 어떤 컬럼이 결과에 영향을 더 크게 미치는지 확실히 확인할 수 있습니다. 이러한 경우 다른 문제에서도 error의 차이가 크게 나지 않는다면, 차원 축소와 비슷한 개념으로 변수의 값을 7개만 남겨도 충분히 결과를 예측할 수 있습니다. 다만 Linear Regression과 L1, L2 Regularization의 차이 중 하나는 $\\alpha$라는 하이퍼파라미터(수식에서는 $\\lambda$)가 하나 더 들어간다는 것이고, 그 값에 따라 error에 영향을 미친다는 점입니다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "angry-filename",
   "metadata": {},
   "source": [
    "## L2 Regularization"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "authorized-current",
   "metadata": {},
   "source": [
    "### L2 Regularization(Ridge) 의 정의\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "imposed-orbit",
   "metadata": {},
   "source": [
    "### L1 / L2 Regularization의 차이점"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "middle-enterprise",
   "metadata": {},
   "source": [
    "> iteration 숫자를 제약 조건으로 주고 실험 예정"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "committed-sunday",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "=3\n"
     ]
    }
   ],
   "source": [
    "from sklearn.datasets import load_wine\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import mean_absolute_error, mean_squared_error\n",
    "\n",
    "wine = load_wine()\n",
    "wine_df = pd.DataFrame(data=wine.data, columns=wine.feature_names)\n",
    "target_df = pd.DataFrame(data=wine.target, columns=['Y'])\n",
    "X_train, X_test, y_train, y_test = train_test_split(wine_df, target_df, test_size= 0.3, random_state=101)\n",
    "print('=3')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "stuffed-arena",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of Lasso\n",
      "Mean Absolute Error: 0.24845768841769436\n",
      "Mean Squared Error: 0.10262989110341268\n",
      "Mean Root Squared Error: 0.32035900346862844\n",
      "\n",
      "\n",
      " coefficient of Lasso\n",
      "[-0.          0.         -0.          0.03295564  0.00109495  0.\n",
      " -0.4027847   0.          0.          0.06023131 -0.         -0.12001119\n",
      " -0.00078971]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/sklearn/linear_model/_coordinate_descent.py:529: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Duality gap: 3.924145836492522, tolerance: 0.007479838709677421\n",
      "  positive)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import Lasso\n",
    "\n",
    "L1 = Lasso(alpha=0.05, max_iter=5)\n",
    "L1.fit(X_train, y_train)\n",
    "pred = L1.predict(X_test)\n",
    "\n",
    "print(\"result of Lasso\")\n",
    "print('Mean Absolute Error:', mean_absolute_error(y_test, pred))\n",
    "print('Mean Squared Error:', mean_squared_error(y_test, pred))\n",
    "print('Mean Root Squared Error:', np.sqrt(mean_squared_error(y_test, pred)))\n",
    "\n",
    "print(\"\\n\\n coefficient of Lasso\")\n",
    "print(L1.coef_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "printable-facing",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of Ridge\n",
      "Mean Absolute Error: 0.2511466959936429\n",
      "Mean Squared Error: 0.10568076460795563\n",
      "Mean Root Squared Error: 0.3250857803841251\n",
      "\n",
      "\n",
      " coefficient of Ridge\n",
      "[[-8.12456257e-02  4.35541496e-02 -1.21661565e-01  3.65979773e-02\n",
      "  -3.94014013e-04  1.39168707e-01 -4.50691113e-01 -4.87216747e-01\n",
      "   9.54111059e-02  5.37077039e-02 -1.28602933e-01 -2.89832790e-01\n",
      "  -5.73136185e-04]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import Ridge\n",
    "\n",
    "L2 = Ridge(alpha=0.05,max_iter=5)\n",
    "L2.fit(X_train, y_train)\n",
    "pred = L2.predict(X_test)\n",
    "\n",
    "print(\"result of Ridge\")\n",
    "print('Mean Absolute Error:', mean_absolute_error(y_test, pred))\n",
    "print('Mean Squared Error:', mean_squared_error(y_test, pred))\n",
    "print('Mean Root Squared Error:', np.sqrt(mean_squared_error(y_test, pred)))\n",
    "\n",
    "print(\"\\n\\n coefficient of Ridge\")\n",
    "print(L2.coef_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "every-failure",
   "metadata": {},
   "source": [
    "#### 아이펠 설명\n",
    "\n",
    "iteration 값을 5로만 설정해 보아도, L2 Regularization의 문제는 Linear Regression과 같은 값이 나오지만, L1 Regularization에서는 앞 step과 같은 값을 확인할 수는 없었습니다. 심지어 수렴하지 않았다는 경고까지 나오는군요! 이는 아직 다른 조건들을 만족하는 답을 찾지 못하였다는 뜻입니다.\n",
    "\n",
    "정리하면, `L1 Regularization`은 **가중치가 적은** 벡터에 해당하는 계수를 `0`으로 보내면서 차원 축소와 비슷한 역할을 하는 것이 특징이며, `L2 Regularization`은 0이 아닌 0에 가깝게 보내지만 **제곱 텀**이 있기 때문에 L1 Regularization보다는 **수렴 속도가 빠르다**는 장점이 있습니다. 예를 들어, $A=[1,1,1,1,1]$ , $B=[5,0,0,0,0]$ 의 경우 L1-norm은 같지만, L2-norm은 같지 않습니다. 즉, **제곱 텀에서 결과에 큰 영향을 미치는 값은 더 크게, 결과에 영향이 적은 값들은 더 작게 보내면서 수렴 속도가 빨라지는 것**입니다.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "overhead-antenna",
   "metadata": {},
   "source": [
    "## Extra : Lp norm"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "massive-supplement",
   "metadata": {},
   "source": [
    "### vector norm"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "valuable-anderson",
   "metadata": {},
   "source": [
    "- p를 바꿔가면서 실험해보기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "palestinian-force",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of numpy package norm function : 10.00008 \n",
      "result of making norm : 10.00008 \n"
     ]
    }
   ],
   "source": [
    "x=np.array([1,10,1,1,1])\n",
    "p=5\n",
    "norm_x=np.linalg.norm(x, ord=p)\n",
    "making_norm = (sum(x**p))**(1/p)\n",
    "print(\"result of numpy package norm function : %0.5f \"%norm_x) \n",
    "print(\"result of making norm : %0.5f \"%making_norm)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "deadly-fisher",
   "metadata": {},
   "source": [
    "> - p=inf 일 경우에는 가장 벡터 요소 중 큰 숫자를 출력"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "generous-legislation",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of infinite norm : 10.00000 \n"
     ]
    }
   ],
   "source": [
    "norm_x=np.linalg.norm(x, ord=np.inf)\n",
    "print(\"result of infinite norm : %0.5f \"%norm_x)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "prescribed-portal",
   "metadata": {},
   "source": [
    "--------------\n",
    "--------------\n",
    "--------------\n",
    "--------------"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "pediatric-colon",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of numpy package norm function : 14.00000 \n",
      "result of making norm : 14.00000 \n"
     ]
    }
   ],
   "source": [
    "p=1\n",
    "norm_x=np.linalg.norm(x, ord=p)\n",
    "making_norm = (sum(x**p))**(1/p)\n",
    "print(\"result of numpy package norm function : %0.5f \"%norm_x) \n",
    "print(\"result of making norm : %0.5f \"%making_norm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "collective-arbor",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of numpy package norm function : 10.19804 \n",
      "result of making norm : 10.19804 \n"
     ]
    }
   ],
   "source": [
    "p=2\n",
    "norm_x=np.linalg.norm(x, ord=p)\n",
    "making_norm = (sum(x**p))**(1/p)\n",
    "print(\"result of numpy package norm function : %0.5f \"%norm_x) \n",
    "print(\"result of making norm : %0.5f \"%making_norm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "dying-celebration",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of numpy package norm function : 10.01332 \n",
      "result of making norm : 10.01332 \n"
     ]
    }
   ],
   "source": [
    "p=3\n",
    "norm_x=np.linalg.norm(x, ord=p)\n",
    "making_norm = (sum(x**p))**(1/p)\n",
    "print(\"result of numpy package norm function : %0.5f \"%norm_x) \n",
    "print(\"result of making norm : %0.5f \"%making_norm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "organized-syndication",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result of numpy package norm function : 2.00000 \n",
      "result of making norm : 2.00000 \n",
      "=======================================\n",
      "result of numpy package norm function : 1.41421 \n",
      "result of making norm : 1.41421 \n",
      "=======================================\n",
      "result of numpy package norm function : 1.25992 \n",
      "result of making norm : 1.25992 \n",
      "=======================================\n",
      "result of numpy package norm function : 1.18921 \n",
      "result of making norm : 1.18921 \n",
      "=======================================\n",
      "result of numpy package norm function : 1.14870 \n",
      "result of making norm : 1.14870 \n",
      "=======================================\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_x = np.array([1,1])\n",
    "p_list = []\n",
    "norm_list = []\n",
    "for i in range(1, 5+1):\n",
    "    p = i\n",
    "    norm_x = np.linalg.norm(test_x, ord=p)\n",
    "    making_norm = (sum(test_x**p))**(1/p)\n",
    "    p_list.append(i)\n",
    "    norm_list.append(making_norm)\n",
    "    print(\"result of numpy package norm function : %0.5f \"%norm_x) \n",
    "    print(\"result of making norm : %0.5f \"%making_norm)\n",
    "    print('=======================================')\n",
    "plt.figure()\n",
    "plt.scatter(p_list, norm_list)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "lined-central",
   "metadata": {},
   "source": [
    "--------------\n",
    "--------------\n",
    "--------------\n",
    "--------------"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "stuck-threat",
   "metadata": {},
   "source": [
    "### matrix norm\n",
    "\n",
    "- vector norm과 조금 다르다.\n",
    "- 주로 p=1, 혹은 inf일 때만 알면 된다.\n",
    "> - p=1인 경우에는 컬럼의 합이 가장 큰 값이 출력되고, p=∞인 경우에는 로우의 합이 가장 큰 값이 출력\n",
    "> - 인터넷에서 수식 찾아서 보면 이해간다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "indoor-stationery",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result inf norm of A : 18.0\n",
      "result one norm of A : 14.0\n"
     ]
    }
   ],
   "source": [
    "A=np.array([[1,2,3],[1,2,3],[4,6,8]])\n",
    "inf_norm_A=np.linalg.norm(A, ord=np.inf)\n",
    "print(\"result inf norm of A :\", inf_norm_A)\n",
    "one_norm_A=np.linalg.norm(A, ord=1)\n",
    "print(\"result one norm of A :\", one_norm_A)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "advisory-labor",
   "metadata": {},
   "source": [
    "## Dropout"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "checked-percentage",
   "metadata": {},
   "source": [
    "### 실습 (not overfitting)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fluid-invitation",
   "metadata": {},
   "source": [
    "- fashion mnist를 통해 확인해보기\n",
    "- epoch 5 정도면 결과가 나옴(데이터가 간단해서)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "existing-alias",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "=3\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "fashion_mnist = keras.datasets.fashion_mnist\n",
    "print('=3')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "numerical-anaheim",
   "metadata": {},
   "outputs": [],
   "source": [
    "(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()\n",
    "class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat',\n",
    "               'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']\n",
    "\n",
    "train_images = train_images / 255.0\n",
    "test_images = test_images / 255.0"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "italic-organizer",
   "metadata": {},
   "source": [
    "> - dropout rate : 0.9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "functioning-outreach",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "1875/1875 [==============================] - 3s 2ms/step - loss: 1.6385 - accuracy: 0.3795\n",
      "Epoch 2/5\n",
      "1875/1875 [==============================] - 3s 2ms/step - loss: 1.1949 - accuracy: 0.5232\n",
      "Epoch 3/5\n",
      "1875/1875 [==============================] - 3s 2ms/step - loss: 1.1381 - accuracy: 0.5417\n",
      "Epoch 4/5\n",
      "1875/1875 [==============================] - 3s 2ms/step - loss: 1.0960 - accuracy: 0.5573\n",
      "Epoch 5/5\n",
      "1875/1875 [==============================] - 3s 2ms/step - loss: 1.0847 - accuracy: 0.5548\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential([\n",
    "    keras.layers.Flatten(input_shape=(28, 28)),\n",
    "    keras.layers.Dense(128, activation='relu'),\n",
    "    # 여기에 dropout layer를 추가해보았습니다. 나머지 layer는 아래의 실습과 같습니다.\n",
    "    keras.layers.Dropout(0.9),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])\n",
    "\n",
    "model.compile(optimizer='adam',loss='sparse_categorical_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "history= model.fit(train_images, train_labels, epochs=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "prescribed-making",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "1875/1875 [==============================] - 3s 1ms/step - loss: 0.6259 - accuracy: 0.7796\n",
      "Epoch 2/5\n",
      "1875/1875 [==============================] - 3s 1ms/step - loss: 0.3840 - accuracy: 0.8609\n",
      "Epoch 3/5\n",
      "1875/1875 [==============================] - 3s 1ms/step - loss: 0.3370 - accuracy: 0.8772\n",
      "Epoch 4/5\n",
      "1875/1875 [==============================] - 3s 1ms/step - loss: 0.3091 - accuracy: 0.8860\n",
      "Epoch 5/5\n",
      "1875/1875 [==============================] - 3s 1ms/step - loss: 0.2896 - accuracy: 0.8912\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential([\n",
    "    keras.layers.Flatten(input_shape=(28, 28)),\n",
    "    # 이번에는 dropout layer가 없습니다. \n",
    "    keras.layers.Dense(128, activation='relu'),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])\n",
    "\n",
    "model.compile(optimizer='adam',loss='sparse_categorical_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "history = model.fit(train_images, train_labels, epochs=5)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "colonial-biotechnology",
   "metadata": {},
   "source": [
    "> - 이미지 데이터이지만 fully connect layer로도 잘 학습됨\n",
    "> - 반면 dropout이 학습에서 규제하는 모습을 보여줬다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "powerful-passing",
   "metadata": {},
   "source": [
    "------------------------------------\n",
    "------------------------------------\n",
    "------------------------------------\n",
    "------------------------------------"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "exceptional-honey",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n"
     ]
    },
    {
     "ename": "ValueError",
     "evalue": "in user code:\n\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py:805 train_function  *\n        return step_function(self, iterator)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py:795 step_function  **\n        outputs = model.distribute_strategy.run(run_step, args=(data,))\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/distribute/distribute_lib.py:1259 run\n        return self._extended.call_for_each_replica(fn, args=args, kwargs=kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/distribute/distribute_lib.py:2730 call_for_each_replica\n        return self._call_for_each_replica(fn, args, kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/distribute/distribute_lib.py:3417 _call_for_each_replica\n        return fn(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py:788 run_step  **\n        outputs = model.train_step(data)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py:756 train_step\n        y, y_pred, sample_weight, regularization_losses=self.losses)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/compile_utils.py:203 __call__\n        loss_value = loss_obj(y_t, y_p, sample_weight=sw)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/losses.py:152 __call__\n        losses = call_fn(y_true, y_pred)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/losses.py:256 call  **\n        return ag_fn(y_true, y_pred, **self._fn_kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py:201 wrapper\n        return target(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/losses.py:1569 sparse_categorical_crossentropy\n        y_true, y_pred, from_logits=from_logits, axis=axis)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py:201 wrapper\n        return target(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/backend.py:4941 sparse_categorical_crossentropy\n        labels=target, logits=output)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py:201 wrapper\n        return target(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/ops/nn_ops.py:4241 sparse_softmax_cross_entropy_with_logits_v2\n        labels=labels, logits=logits, name=name)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py:201 wrapper\n        return target(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/ops/nn_ops.py:4156 sparse_softmax_cross_entropy_with_logits\n        logits.get_shape()))\n\n    ValueError: Shape mismatch: The shape of labels (received (32, 1)) should equal the shape of logits except for the last dimension (received (32, 28, 10)).\n",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-43-a5acee76bf0c>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      9\u001b[0m              metrics=['accuracy'])\n\u001b[1;32m     10\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 11\u001b[0;31m \u001b[0mhistroy\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_images\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_labels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m5\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1098\u001b[0m                 _r=1):\n\u001b[1;32m   1099\u001b[0m               \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_train_batch_begin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1100\u001b[0;31m               \u001b[0mtmp_logs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0miterator\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1101\u001b[0m               \u001b[0;32mif\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshould_sync\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1102\u001b[0m                 \u001b[0mcontext\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masync_wait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    826\u001b[0m     \u001b[0mtracing_count\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexperimental_get_tracing_count\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    827\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mtrace\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTrace\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_name\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mtm\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 828\u001b[0;31m       \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    829\u001b[0m       \u001b[0mcompiler\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m\"xla\"\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_experimental_compile\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0;34m\"nonXla\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    830\u001b[0m       \u001b[0mnew_tracing_count\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexperimental_get_tracing_count\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py\u001b[0m in \u001b[0;36m_call\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    869\u001b[0m       \u001b[0;31m# This is the first call of __call__, so we have to initialize.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    870\u001b[0m       \u001b[0minitializers\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 871\u001b[0;31m       \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_initialize\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkwds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0madd_initializers_to\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minitializers\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    872\u001b[0m     \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    873\u001b[0m       \u001b[0;31m# At this point we know that the initialization is complete (or less\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py\u001b[0m in \u001b[0;36m_initialize\u001b[0;34m(self, args, kwds, add_initializers_to)\u001b[0m\n\u001b[1;32m    724\u001b[0m     self._concrete_stateful_fn = (\n\u001b[1;32m    725\u001b[0m         self._stateful_fn._get_concrete_function_internal_garbage_collected(  # pylint: disable=protected-access\n\u001b[0;32m--> 726\u001b[0;31m             *args, **kwds))\n\u001b[0m\u001b[1;32m    727\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    728\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0minvalid_creator_scope\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0munused_args\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0munused_kwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/eager/function.py\u001b[0m in \u001b[0;36m_get_concrete_function_internal_garbage_collected\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   2967\u001b[0m       \u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkwargs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2968\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_lock\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2969\u001b[0;31m       \u001b[0mgraph_function\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_maybe_define_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2970\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mgraph_function\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2971\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/eager/function.py\u001b[0m in \u001b[0;36m_maybe_define_function\u001b[0;34m(self, args, kwargs)\u001b[0m\n\u001b[1;32m   3359\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3360\u001b[0m           \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_function_cache\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmissed\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0madd\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcall_context_key\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 3361\u001b[0;31m           \u001b[0mgraph_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_create_graph_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   3362\u001b[0m           \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_function_cache\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprimary\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mcache_key\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mgraph_function\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3363\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/eager/function.py\u001b[0m in \u001b[0;36m_create_graph_function\u001b[0;34m(self, args, kwargs, override_flat_arg_shapes)\u001b[0m\n\u001b[1;32m   3204\u001b[0m             \u001b[0marg_names\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0marg_names\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3205\u001b[0m             \u001b[0moverride_flat_arg_shapes\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0moverride_flat_arg_shapes\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 3206\u001b[0;31m             capture_by_value=self._capture_by_value),\n\u001b[0m\u001b[1;32m   3207\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_function_attributes\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3208\u001b[0m         \u001b[0mfunction_spec\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfunction_spec\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/framework/func_graph.py\u001b[0m in \u001b[0;36mfunc_graph_from_py_func\u001b[0;34m(name, python_func, args, kwargs, signature, func_graph, autograph, autograph_options, add_control_dependencies, arg_names, op_return_value, collections, capture_by_value, override_flat_arg_shapes)\u001b[0m\n\u001b[1;32m    988\u001b[0m         \u001b[0m_\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moriginal_func\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_decorator\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munwrap\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpython_func\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    989\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 990\u001b[0;31m       \u001b[0mfunc_outputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpython_func\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mfunc_args\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfunc_kwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    991\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    992\u001b[0m       \u001b[0;31m# invariant: `func_outputs` contains only Tensors, CompositeTensors,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py\u001b[0m in \u001b[0;36mwrapped_fn\u001b[0;34m(*args, **kwds)\u001b[0m\n\u001b[1;32m    632\u001b[0m             \u001b[0mxla_context\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mExit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    633\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 634\u001b[0;31m           \u001b[0mout\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mweak_wrapped_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__wrapped__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    635\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mout\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    636\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tensorflow/python/framework/func_graph.py\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    975\u001b[0m           \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint:disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    976\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mhasattr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"ag_error_metadata\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 977\u001b[0;31m               \u001b[0;32mraise\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mag_error_metadata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mto_exception\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    978\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    979\u001b[0m               \u001b[0;32mraise\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: in user code:\n\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py:805 train_function  *\n        return step_function(self, iterator)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py:795 step_function  **\n        outputs = model.distribute_strategy.run(run_step, args=(data,))\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/distribute/distribute_lib.py:1259 run\n        return self._extended.call_for_each_replica(fn, args=args, kwargs=kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/distribute/distribute_lib.py:2730 call_for_each_replica\n        return self._call_for_each_replica(fn, args, kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/distribute/distribute_lib.py:3417 _call_for_each_replica\n        return fn(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py:788 run_step  **\n        outputs = model.train_step(data)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py:756 train_step\n        y, y_pred, sample_weight, regularization_losses=self.losses)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/engine/compile_utils.py:203 __call__\n        loss_value = loss_obj(y_t, y_p, sample_weight=sw)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/losses.py:152 __call__\n        losses = call_fn(y_true, y_pred)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/losses.py:256 call  **\n        return ag_fn(y_true, y_pred, **self._fn_kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py:201 wrapper\n        return target(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/losses.py:1569 sparse_categorical_crossentropy\n        y_true, y_pred, from_logits=from_logits, axis=axis)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py:201 wrapper\n        return target(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/keras/backend.py:4941 sparse_categorical_crossentropy\n        labels=target, logits=output)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py:201 wrapper\n        return target(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/ops/nn_ops.py:4241 sparse_softmax_cross_entropy_with_logits_v2\n        labels=labels, logits=logits, name=name)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py:201 wrapper\n        return target(*args, **kwargs)\n    /opt/conda/lib/python3.7/site-packages/tensorflow/python/ops/nn_ops.py:4156 sparse_softmax_cross_entropy_with_logits\n        logits.get_shape()))\n\n    ValueError: Shape mismatch: The shape of labels (received (32, 1)) should equal the shape of logits except for the last dimension (received (32, 28, 10)).\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential([\n",
    "    keras.layers.Dense(128, activation='relu', input_shape=(28,28)),\n",
    "    keras.layers.Dropout(0.9),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])\n",
    "\n",
    "model.compile(optimizer='adam',\n",
    "             loss='sparse_categorical_crossentropy',\n",
    "             metrics=['accuracy'])\n",
    "\n",
    "histroy = model.fit(train_images, train_labels, epochs=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "stylish-floating",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(60000, 28, 28)"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_images.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "sexual-chair",
   "metadata": {},
   "source": [
    "------------------------------------\n",
    "------------------------------------\n",
    "------------------------------------\n",
    "------------------------------------"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "capable-dimension",
   "metadata": {},
   "source": [
    "### 실습 (overfitting)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "herbal-theorem",
   "metadata": {},
   "source": [
    "#### 아이펠 설명\n",
    "\n",
    "overfitting이 되는지 확인해 보려면 train set과 validation set의 loss function을 그려보는 것이 가장 첫 번째 시도입니다. 우리는 위의 데이터를 이제 train:valid=99:1로 split 한 뒤에 loss function의 값을 그려보고, overfitting 이 나는 fully connected layer를 만들어보도록 합시다. overfitting이 되게 하기 위해 의도적으로 train set을 99%로 늘리고 validation set을 줄였습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "orange-lloyd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/200\n",
      "117/117 [==============================] - 1s 6ms/step - loss: 2.2072 - accuracy: 0.3482 - val_loss: 1.6631 - val_accuracy: 0.6100\n",
      "Epoch 2/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 1.5392 - accuracy: 0.5859 - val_loss: 1.1912 - val_accuracy: 0.6733\n",
      "Epoch 3/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 1.1478 - accuracy: 0.6692 - val_loss: 0.9687 - val_accuracy: 0.7217\n",
      "Epoch 4/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.9428 - accuracy: 0.7067 - val_loss: 0.8296 - val_accuracy: 0.7550\n",
      "Epoch 5/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.8194 - accuracy: 0.7329 - val_loss: 0.7495 - val_accuracy: 0.7733\n",
      "Epoch 6/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.7539 - accuracy: 0.7433 - val_loss: 0.6995 - val_accuracy: 0.7683\n",
      "Epoch 7/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.7004 - accuracy: 0.7535 - val_loss: 0.6623 - val_accuracy: 0.7783\n",
      "Epoch 8/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.6655 - accuracy: 0.7638 - val_loss: 0.6350 - val_accuracy: 0.7817\n",
      "Epoch 9/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.6354 - accuracy: 0.7723 - val_loss: 0.6189 - val_accuracy: 0.7883\n",
      "Epoch 10/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.6199 - accuracy: 0.7794 - val_loss: 0.5948 - val_accuracy: 0.7900\n",
      "Epoch 11/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5970 - accuracy: 0.7855 - val_loss: 0.5793 - val_accuracy: 0.7900\n",
      "Epoch 12/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5826 - accuracy: 0.7909 - val_loss: 0.5660 - val_accuracy: 0.7933\n",
      "Epoch 13/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5683 - accuracy: 0.7983 - val_loss: 0.5545 - val_accuracy: 0.8033\n",
      "Epoch 14/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5563 - accuracy: 0.8038 - val_loss: 0.5410 - val_accuracy: 0.8000\n",
      "Epoch 15/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.5444 - accuracy: 0.8062 - val_loss: 0.5335 - val_accuracy: 0.8050\n",
      "Epoch 16/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5312 - accuracy: 0.8152 - val_loss: 0.5253 - val_accuracy: 0.8017\n",
      "Epoch 17/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5249 - accuracy: 0.8151 - val_loss: 0.5161 - val_accuracy: 0.8133\n",
      "Epoch 18/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5166 - accuracy: 0.8200 - val_loss: 0.5108 - val_accuracy: 0.8100\n",
      "Epoch 19/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5107 - accuracy: 0.8206 - val_loss: 0.5039 - val_accuracy: 0.8133\n",
      "Epoch 20/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5042 - accuracy: 0.8226 - val_loss: 0.4959 - val_accuracy: 0.8200\n",
      "Epoch 21/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4923 - accuracy: 0.8282 - val_loss: 0.4910 - val_accuracy: 0.8233\n",
      "Epoch 22/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4867 - accuracy: 0.8290 - val_loss: 0.4837 - val_accuracy: 0.8217\n",
      "Epoch 23/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4847 - accuracy: 0.8308 - val_loss: 0.4810 - val_accuracy: 0.8217\n",
      "Epoch 24/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.4839 - accuracy: 0.8302 - val_loss: 0.4795 - val_accuracy: 0.8250\n",
      "Epoch 25/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4729 - accuracy: 0.8356 - val_loss: 0.4720 - val_accuracy: 0.8283\n",
      "Epoch 26/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4646 - accuracy: 0.8354 - val_loss: 0.4697 - val_accuracy: 0.8283\n",
      "Epoch 27/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4642 - accuracy: 0.8380 - val_loss: 0.4694 - val_accuracy: 0.8267\n",
      "Epoch 28/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.4564 - accuracy: 0.8390 - val_loss: 0.4621 - val_accuracy: 0.8283\n",
      "Epoch 29/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4568 - accuracy: 0.8401 - val_loss: 0.4591 - val_accuracy: 0.8317\n",
      "Epoch 30/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4553 - accuracy: 0.8409 - val_loss: 0.4555 - val_accuracy: 0.8350\n",
      "Epoch 31/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4472 - accuracy: 0.8424 - val_loss: 0.4522 - val_accuracy: 0.8400\n",
      "Epoch 32/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.4494 - accuracy: 0.8408 - val_loss: 0.4517 - val_accuracy: 0.8333\n",
      "Epoch 33/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4441 - accuracy: 0.8441 - val_loss: 0.4456 - val_accuracy: 0.8333\n",
      "Epoch 34/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4364 - accuracy: 0.8464 - val_loss: 0.4423 - val_accuracy: 0.8417\n",
      "Epoch 35/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4434 - accuracy: 0.8460 - val_loss: 0.4450 - val_accuracy: 0.8317\n",
      "Epoch 36/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4388 - accuracy: 0.8462 - val_loss: 0.4428 - val_accuracy: 0.8333\n",
      "Epoch 37/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.4385 - accuracy: 0.8476 - val_loss: 0.4357 - val_accuracy: 0.8383\n",
      "Epoch 38/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4325 - accuracy: 0.8489 - val_loss: 0.4342 - val_accuracy: 0.8367\n",
      "Epoch 39/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4317 - accuracy: 0.8489 - val_loss: 0.4296 - val_accuracy: 0.8400\n",
      "Epoch 40/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4201 - accuracy: 0.8523 - val_loss: 0.4284 - val_accuracy: 0.8383\n",
      "Epoch 41/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4196 - accuracy: 0.8548 - val_loss: 0.4289 - val_accuracy: 0.8283\n",
      "Epoch 42/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4236 - accuracy: 0.8516 - val_loss: 0.4286 - val_accuracy: 0.8417\n",
      "Epoch 43/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.4289 - accuracy: 0.8493 - val_loss: 0.4231 - val_accuracy: 0.8383\n",
      "Epoch 44/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4200 - accuracy: 0.8524 - val_loss: 0.4217 - val_accuracy: 0.8367\n",
      "Epoch 45/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4182 - accuracy: 0.8542 - val_loss: 0.4217 - val_accuracy: 0.8383\n",
      "Epoch 46/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4158 - accuracy: 0.8550 - val_loss: 0.4180 - val_accuracy: 0.8400\n",
      "Epoch 47/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4155 - accuracy: 0.8564 - val_loss: 0.4205 - val_accuracy: 0.8367\n",
      "Epoch 48/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4137 - accuracy: 0.8561 - val_loss: 0.4133 - val_accuracy: 0.8350\n",
      "Epoch 49/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4121 - accuracy: 0.8553 - val_loss: 0.4138 - val_accuracy: 0.8300\n",
      "Epoch 50/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.4113 - accuracy: 0.8572 - val_loss: 0.4134 - val_accuracy: 0.8350\n",
      "Epoch 51/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4074 - accuracy: 0.8573 - val_loss: 0.4113 - val_accuracy: 0.8283\n",
      "Epoch 52/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4100 - accuracy: 0.8590 - val_loss: 0.4074 - val_accuracy: 0.8350\n",
      "Epoch 53/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4060 - accuracy: 0.8574 - val_loss: 0.4091 - val_accuracy: 0.8383\n",
      "Epoch 54/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4058 - accuracy: 0.8592 - val_loss: 0.4064 - val_accuracy: 0.8383\n",
      "Epoch 55/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4081 - accuracy: 0.8568 - val_loss: 0.4070 - val_accuracy: 0.8333\n",
      "Epoch 56/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4037 - accuracy: 0.8576 - val_loss: 0.4067 - val_accuracy: 0.8317\n",
      "Epoch 57/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3995 - accuracy: 0.8618 - val_loss: 0.4060 - val_accuracy: 0.8350\n",
      "Epoch 58/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4004 - accuracy: 0.8589 - val_loss: 0.4008 - val_accuracy: 0.8317\n",
      "Epoch 59/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3922 - accuracy: 0.8617 - val_loss: 0.4007 - val_accuracy: 0.8367\n",
      "Epoch 60/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3992 - accuracy: 0.8596 - val_loss: 0.3992 - val_accuracy: 0.8367\n",
      "Epoch 61/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3922 - accuracy: 0.8639 - val_loss: 0.3971 - val_accuracy: 0.8333\n",
      "Epoch 62/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3961 - accuracy: 0.8577 - val_loss: 0.3963 - val_accuracy: 0.8317\n",
      "Epoch 63/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3955 - accuracy: 0.8610 - val_loss: 0.3934 - val_accuracy: 0.8317\n",
      "Epoch 64/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3853 - accuracy: 0.8678 - val_loss: 0.3984 - val_accuracy: 0.8417\n",
      "Epoch 65/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3838 - accuracy: 0.8652 - val_loss: 0.3936 - val_accuracy: 0.8417\n",
      "Epoch 66/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3819 - accuracy: 0.8672 - val_loss: 0.3958 - val_accuracy: 0.8350\n",
      "Epoch 67/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3864 - accuracy: 0.8654 - val_loss: 0.3920 - val_accuracy: 0.8367\n",
      "Epoch 68/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3831 - accuracy: 0.8644 - val_loss: 0.3897 - val_accuracy: 0.8333\n",
      "Epoch 69/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3894 - accuracy: 0.8636 - val_loss: 0.3903 - val_accuracy: 0.8350\n",
      "Epoch 70/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3833 - accuracy: 0.8656 - val_loss: 0.3834 - val_accuracy: 0.8333\n",
      "Epoch 71/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3781 - accuracy: 0.8671 - val_loss: 0.3897 - val_accuracy: 0.8333\n",
      "Epoch 72/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3771 - accuracy: 0.8661 - val_loss: 0.3853 - val_accuracy: 0.8367\n",
      "Epoch 73/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3873 - accuracy: 0.8618 - val_loss: 0.3848 - val_accuracy: 0.8367\n",
      "Epoch 74/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3747 - accuracy: 0.8685 - val_loss: 0.3824 - val_accuracy: 0.8350\n",
      "Epoch 75/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3780 - accuracy: 0.8667 - val_loss: 0.3869 - val_accuracy: 0.8417\n",
      "Epoch 76/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3734 - accuracy: 0.8692 - val_loss: 0.3902 - val_accuracy: 0.8400\n",
      "Epoch 77/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3760 - accuracy: 0.8677 - val_loss: 0.3830 - val_accuracy: 0.8383\n",
      "Epoch 78/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3763 - accuracy: 0.8693 - val_loss: 0.3818 - val_accuracy: 0.8400\n",
      "Epoch 79/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3740 - accuracy: 0.8681 - val_loss: 0.3812 - val_accuracy: 0.8433\n",
      "Epoch 80/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3756 - accuracy: 0.8666 - val_loss: 0.3793 - val_accuracy: 0.8383\n",
      "Epoch 81/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3722 - accuracy: 0.8692 - val_loss: 0.3810 - val_accuracy: 0.8367\n",
      "Epoch 82/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3678 - accuracy: 0.8704 - val_loss: 0.3752 - val_accuracy: 0.8367\n",
      "Epoch 83/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3762 - accuracy: 0.8676 - val_loss: 0.3765 - val_accuracy: 0.8367\n",
      "Epoch 84/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3695 - accuracy: 0.8688 - val_loss: 0.3795 - val_accuracy: 0.8433\n",
      "Epoch 85/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3650 - accuracy: 0.8716 - val_loss: 0.3765 - val_accuracy: 0.8400\n",
      "Epoch 86/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3656 - accuracy: 0.8712 - val_loss: 0.3762 - val_accuracy: 0.8417\n",
      "Epoch 87/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3643 - accuracy: 0.8705 - val_loss: 0.3768 - val_accuracy: 0.8467\n",
      "Epoch 88/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3642 - accuracy: 0.8721 - val_loss: 0.3737 - val_accuracy: 0.8450\n",
      "Epoch 89/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3597 - accuracy: 0.8733 - val_loss: 0.3719 - val_accuracy: 0.8483\n",
      "Epoch 90/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3625 - accuracy: 0.8723 - val_loss: 0.3748 - val_accuracy: 0.8417\n",
      "Epoch 91/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3596 - accuracy: 0.8747 - val_loss: 0.3747 - val_accuracy: 0.8483\n",
      "Epoch 92/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3624 - accuracy: 0.8711 - val_loss: 0.3749 - val_accuracy: 0.8467\n",
      "Epoch 93/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3579 - accuracy: 0.8735 - val_loss: 0.3750 - val_accuracy: 0.8450\n",
      "Epoch 94/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3537 - accuracy: 0.8757 - val_loss: 0.3690 - val_accuracy: 0.8483\n",
      "Epoch 95/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3563 - accuracy: 0.8738 - val_loss: 0.3710 - val_accuracy: 0.8533\n",
      "Epoch 96/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3578 - accuracy: 0.8729 - val_loss: 0.3671 - val_accuracy: 0.8467\n",
      "Epoch 97/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3606 - accuracy: 0.8732 - val_loss: 0.3727 - val_accuracy: 0.8483\n",
      "Epoch 98/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3468 - accuracy: 0.8773 - val_loss: 0.3668 - val_accuracy: 0.8533\n",
      "Epoch 99/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3606 - accuracy: 0.8725 - val_loss: 0.3695 - val_accuracy: 0.8567\n",
      "Epoch 100/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3545 - accuracy: 0.8760 - val_loss: 0.3661 - val_accuracy: 0.8467\n",
      "Epoch 101/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3449 - accuracy: 0.8782 - val_loss: 0.3640 - val_accuracy: 0.8517\n",
      "Epoch 102/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3470 - accuracy: 0.8777 - val_loss: 0.3640 - val_accuracy: 0.8450\n",
      "Epoch 103/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3536 - accuracy: 0.8748 - val_loss: 0.3620 - val_accuracy: 0.8533\n",
      "Epoch 104/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3516 - accuracy: 0.8762 - val_loss: 0.3627 - val_accuracy: 0.8567\n",
      "Epoch 105/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3477 - accuracy: 0.8770 - val_loss: 0.3561 - val_accuracy: 0.8433\n",
      "Epoch 106/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3430 - accuracy: 0.8771 - val_loss: 0.3602 - val_accuracy: 0.8483\n",
      "Epoch 107/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3432 - accuracy: 0.8781 - val_loss: 0.3638 - val_accuracy: 0.8550\n",
      "Epoch 108/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3463 - accuracy: 0.8786 - val_loss: 0.3588 - val_accuracy: 0.8483\n",
      "Epoch 109/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3462 - accuracy: 0.8767 - val_loss: 0.3557 - val_accuracy: 0.8467\n",
      "Epoch 110/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3428 - accuracy: 0.8785 - val_loss: 0.3567 - val_accuracy: 0.8467\n",
      "Epoch 111/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3417 - accuracy: 0.8785 - val_loss: 0.3593 - val_accuracy: 0.8483\n",
      "Epoch 112/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3416 - accuracy: 0.8796 - val_loss: 0.3585 - val_accuracy: 0.8500\n",
      "Epoch 113/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3402 - accuracy: 0.8793 - val_loss: 0.3574 - val_accuracy: 0.8533\n",
      "Epoch 114/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3325 - accuracy: 0.8822 - val_loss: 0.3534 - val_accuracy: 0.8583\n",
      "Epoch 115/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3422 - accuracy: 0.8781 - val_loss: 0.3535 - val_accuracy: 0.8500\n",
      "Epoch 116/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3433 - accuracy: 0.8790 - val_loss: 0.3542 - val_accuracy: 0.8533\n",
      "Epoch 117/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3402 - accuracy: 0.8802 - val_loss: 0.3536 - val_accuracy: 0.8517\n",
      "Epoch 118/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3388 - accuracy: 0.8797 - val_loss: 0.3519 - val_accuracy: 0.8517\n",
      "Epoch 119/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3327 - accuracy: 0.8826 - val_loss: 0.3510 - val_accuracy: 0.8500\n",
      "Epoch 120/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3334 - accuracy: 0.8815 - val_loss: 0.3604 - val_accuracy: 0.8567\n",
      "Epoch 121/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3387 - accuracy: 0.8780 - val_loss: 0.3521 - val_accuracy: 0.8550\n",
      "Epoch 122/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3344 - accuracy: 0.8831 - val_loss: 0.3545 - val_accuracy: 0.8550\n",
      "Epoch 123/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3311 - accuracy: 0.8838 - val_loss: 0.3542 - val_accuracy: 0.8467\n",
      "Epoch 124/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3372 - accuracy: 0.8814 - val_loss: 0.3523 - val_accuracy: 0.8567\n",
      "Epoch 125/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3321 - accuracy: 0.8813 - val_loss: 0.3498 - val_accuracy: 0.8533\n",
      "Epoch 126/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3339 - accuracy: 0.8822 - val_loss: 0.3493 - val_accuracy: 0.8517\n",
      "Epoch 127/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3315 - accuracy: 0.8841 - val_loss: 0.3517 - val_accuracy: 0.8550\n",
      "Epoch 128/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3290 - accuracy: 0.8825 - val_loss: 0.3493 - val_accuracy: 0.8567\n",
      "Epoch 129/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3318 - accuracy: 0.8823 - val_loss: 0.3477 - val_accuracy: 0.8517\n",
      "Epoch 130/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3250 - accuracy: 0.8837 - val_loss: 0.3487 - val_accuracy: 0.8483\n",
      "Epoch 131/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3238 - accuracy: 0.8844 - val_loss: 0.3448 - val_accuracy: 0.8567\n",
      "Epoch 132/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3242 - accuracy: 0.8856 - val_loss: 0.3472 - val_accuracy: 0.8483\n",
      "Epoch 133/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3328 - accuracy: 0.8815 - val_loss: 0.3513 - val_accuracy: 0.8533\n",
      "Epoch 134/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3257 - accuracy: 0.8833 - val_loss: 0.3475 - val_accuracy: 0.8483\n",
      "Epoch 135/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3263 - accuracy: 0.8842 - val_loss: 0.3441 - val_accuracy: 0.8633\n",
      "Epoch 136/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3186 - accuracy: 0.8879 - val_loss: 0.3451 - val_accuracy: 0.8567\n",
      "Epoch 137/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3247 - accuracy: 0.8854 - val_loss: 0.3481 - val_accuracy: 0.8500\n",
      "Epoch 138/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3207 - accuracy: 0.8859 - val_loss: 0.3443 - val_accuracy: 0.8517\n",
      "Epoch 139/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3258 - accuracy: 0.8848 - val_loss: 0.3446 - val_accuracy: 0.8600\n",
      "Epoch 140/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3227 - accuracy: 0.8846 - val_loss: 0.3471 - val_accuracy: 0.8517\n",
      "Epoch 141/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3211 - accuracy: 0.8863 - val_loss: 0.3438 - val_accuracy: 0.8567\n",
      "Epoch 142/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3256 - accuracy: 0.8840 - val_loss: 0.3470 - val_accuracy: 0.8600\n",
      "Epoch 143/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3208 - accuracy: 0.8864 - val_loss: 0.3506 - val_accuracy: 0.8500\n",
      "Epoch 144/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3221 - accuracy: 0.8871 - val_loss: 0.3418 - val_accuracy: 0.8533\n",
      "Epoch 145/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3238 - accuracy: 0.8840 - val_loss: 0.3414 - val_accuracy: 0.8567\n",
      "Epoch 146/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3177 - accuracy: 0.8854 - val_loss: 0.3458 - val_accuracy: 0.8583\n",
      "Epoch 147/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3201 - accuracy: 0.8853 - val_loss: 0.3441 - val_accuracy: 0.8633\n",
      "Epoch 148/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3234 - accuracy: 0.8855 - val_loss: 0.3443 - val_accuracy: 0.8533\n",
      "Epoch 149/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3100 - accuracy: 0.8882 - val_loss: 0.3456 - val_accuracy: 0.8633\n",
      "Epoch 150/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3127 - accuracy: 0.8895 - val_loss: 0.3480 - val_accuracy: 0.8633\n",
      "Epoch 151/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3166 - accuracy: 0.8866 - val_loss: 0.3407 - val_accuracy: 0.8600\n",
      "Epoch 152/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3110 - accuracy: 0.8894 - val_loss: 0.3430 - val_accuracy: 0.8600\n",
      "Epoch 153/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3159 - accuracy: 0.8872 - val_loss: 0.3432 - val_accuracy: 0.8633\n",
      "Epoch 154/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3167 - accuracy: 0.8858 - val_loss: 0.3370 - val_accuracy: 0.8633\n",
      "Epoch 155/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3128 - accuracy: 0.8893 - val_loss: 0.3431 - val_accuracy: 0.8550\n",
      "Epoch 156/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3102 - accuracy: 0.8888 - val_loss: 0.3368 - val_accuracy: 0.8617\n",
      "Epoch 157/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3176 - accuracy: 0.8862 - val_loss: 0.3412 - val_accuracy: 0.8583\n",
      "Epoch 158/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3105 - accuracy: 0.8895 - val_loss: 0.3354 - val_accuracy: 0.8600\n",
      "Epoch 159/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3164 - accuracy: 0.8864 - val_loss: 0.3374 - val_accuracy: 0.8600\n",
      "Epoch 160/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3077 - accuracy: 0.8899 - val_loss: 0.3443 - val_accuracy: 0.8567\n",
      "Epoch 161/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3084 - accuracy: 0.8910 - val_loss: 0.3416 - val_accuracy: 0.8650\n",
      "Epoch 162/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3116 - accuracy: 0.8897 - val_loss: 0.3426 - val_accuracy: 0.8633\n",
      "Epoch 163/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3092 - accuracy: 0.8880 - val_loss: 0.3393 - val_accuracy: 0.8633\n",
      "Epoch 164/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3088 - accuracy: 0.8891 - val_loss: 0.3411 - val_accuracy: 0.8633\n",
      "Epoch 165/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3114 - accuracy: 0.8889 - val_loss: 0.3374 - val_accuracy: 0.8667\n",
      "Epoch 166/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3109 - accuracy: 0.8877 - val_loss: 0.3400 - val_accuracy: 0.8583\n",
      "Epoch 167/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3098 - accuracy: 0.8896 - val_loss: 0.3359 - val_accuracy: 0.8617\n",
      "Epoch 168/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3098 - accuracy: 0.8905 - val_loss: 0.3353 - val_accuracy: 0.8617\n",
      "Epoch 169/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3079 - accuracy: 0.8891 - val_loss: 0.3332 - val_accuracy: 0.8633\n",
      "Epoch 170/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3078 - accuracy: 0.8905 - val_loss: 0.3407 - val_accuracy: 0.8600\n",
      "Epoch 171/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3064 - accuracy: 0.8911 - val_loss: 0.3383 - val_accuracy: 0.8667\n",
      "Epoch 172/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3009 - accuracy: 0.8926 - val_loss: 0.3348 - val_accuracy: 0.8650\n",
      "Epoch 173/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3029 - accuracy: 0.8916 - val_loss: 0.3296 - val_accuracy: 0.8583\n",
      "Epoch 174/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3029 - accuracy: 0.8919 - val_loss: 0.3407 - val_accuracy: 0.8667\n",
      "Epoch 175/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3066 - accuracy: 0.8908 - val_loss: 0.3363 - val_accuracy: 0.8633\n",
      "Epoch 176/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3043 - accuracy: 0.8914 - val_loss: 0.3338 - val_accuracy: 0.8683\n",
      "Epoch 177/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3030 - accuracy: 0.8920 - val_loss: 0.3375 - val_accuracy: 0.8633\n",
      "Epoch 178/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3001 - accuracy: 0.8942 - val_loss: 0.3340 - val_accuracy: 0.8617\n",
      "Epoch 179/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2959 - accuracy: 0.8942 - val_loss: 0.3331 - val_accuracy: 0.8650\n",
      "Epoch 180/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3039 - accuracy: 0.8913 - val_loss: 0.3339 - val_accuracy: 0.8550\n",
      "Epoch 181/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2974 - accuracy: 0.8933 - val_loss: 0.3336 - val_accuracy: 0.8667\n",
      "Epoch 182/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.3014 - accuracy: 0.8916 - val_loss: 0.3316 - val_accuracy: 0.8550\n",
      "Epoch 183/200\n",
      "117/117 [==============================] - 0s 4ms/step - loss: 0.2982 - accuracy: 0.8932 - val_loss: 0.3311 - val_accuracy: 0.8583\n",
      "Epoch 184/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2979 - accuracy: 0.8935 - val_loss: 0.3300 - val_accuracy: 0.8600\n",
      "Epoch 185/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3023 - accuracy: 0.8934 - val_loss: 0.3287 - val_accuracy: 0.8633\n",
      "Epoch 186/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2966 - accuracy: 0.8947 - val_loss: 0.3285 - val_accuracy: 0.8633\n",
      "Epoch 187/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3047 - accuracy: 0.8907 - val_loss: 0.3273 - val_accuracy: 0.8667\n",
      "Epoch 188/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2942 - accuracy: 0.8934 - val_loss: 0.3363 - val_accuracy: 0.8633\n",
      "Epoch 189/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2986 - accuracy: 0.8947 - val_loss: 0.3266 - val_accuracy: 0.8650\n",
      "Epoch 190/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2982 - accuracy: 0.8952 - val_loss: 0.3297 - val_accuracy: 0.8667\n",
      "Epoch 191/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2972 - accuracy: 0.8952 - val_loss: 0.3316 - val_accuracy: 0.8650\n",
      "Epoch 192/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2978 - accuracy: 0.8940 - val_loss: 0.3297 - val_accuracy: 0.8650\n",
      "Epoch 193/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2993 - accuracy: 0.8941 - val_loss: 0.3310 - val_accuracy: 0.8650\n",
      "Epoch 194/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2942 - accuracy: 0.8973 - val_loss: 0.3285 - val_accuracy: 0.8667\n",
      "Epoch 195/200\n",
      "117/117 [==============================] - 0s 2ms/step - loss: 0.2979 - accuracy: 0.8936 - val_loss: 0.3305 - val_accuracy: 0.8633\n",
      "Epoch 196/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2967 - accuracy: 0.8938 - val_loss: 0.3244 - val_accuracy: 0.8617\n",
      "Epoch 197/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2948 - accuracy: 0.8937 - val_loss: 0.3258 - val_accuracy: 0.8583\n",
      "Epoch 198/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2882 - accuracy: 0.8964 - val_loss: 0.3272 - val_accuracy: 0.8683\n",
      "Epoch 199/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2872 - accuracy: 0.8983 - val_loss: 0.3287 - val_accuracy: 0.8567\n",
      "Epoch 200/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.2995 - accuracy: 0.8916 - val_loss: 0.3269 - val_accuracy: 0.8633\n"
     ]
    }
   ],
   "source": [
    "X_train, X_valid, y_train, y_valid = train_test_split(train_images, train_labels, test_size=0.01, random_state=101)\n",
    "X_train = X_train / 255.0\n",
    "X_valid = X_valid / 255.0\n",
    "\n",
    "#Dense layer만으로 만들어 낸 classification 모델입니다.\n",
    "model = keras.Sequential([\n",
    "    keras.layers.Flatten(input_shape=(28, 28)),\n",
    "    keras.layers.Dense(256, activation='relu'),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])\n",
    "\n",
    "model.compile(optimizer='adam',loss='sparse_categorical_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "history= model.fit(X_train, y_train, epochs=200, batch_size=512, validation_data=(X_valid, y_valid))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "former-traffic",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# loss 값을 plot 해보겠습니다.\n",
    "y_vloss = history.history['val_loss']\n",
    "y_loss = history.history['loss']\n",
    "x_len = np.arange(len(y_loss))\n",
    "\n",
    "plt.plot(x_len, y_vloss, marker='.', c='red', label=\"Validation-set Loss\")\n",
    "plt.plot(x_len, y_loss, marker='.', c='blue', label=\"Train-set Loss\")\n",
    "plt.legend(loc='upper right')\n",
    "plt.grid()\n",
    "plt.title('Loss graph without dropout layer') \n",
    "plt.ylim(0,1)\n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('loss')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "exposed-means",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/Il7ecAAAACXBIWXMAAAsTAAALEwEAmpwYAABDfUlEQVR4nO2daXhURdaA35NOQlBWQVFZBB1BkX1RI4hBQVxwRUeRzw0VNxR3wXEUdRydcWMYUUEEBnVEx22QQUGYBFTiCGJYXQBFFgExrAGydFLfj7o3fdPpTjohnW7o8z5PP32XunVP1719TtWpqlNijEFRFEVJXJJiLYCiKIoSW9QQKIqiJDhqCBRFURIcNQSKoigJjhoCRVGUBEcNgaIoSoKjhkA5qBGR0SLyRg3n2UpE8kTEV0EaIyK/q8n7VhcRmSIif4q1HFVFRFo75Zgca1kOdtQQxBkikiUi20WkTqxlUUJjjFlnjKlnjCmG0md2Y23cOxqGLVbUZrkpFaOGII4QkdbA6YABLqzle8dlrauiWrey/8Trc48VYkk4vZhwPzjOuQb4EpgCXOs9ISItReR9EdkqIrki8qLn3E0i8q2I7BaRlSLSzTlexj3hdRGISIaIbBCRB0VkMzBZRBqLyAznHtud7Rae6w8Tkcki8otz/kPn+HIRucCTLkVEfhORrqF+pIg8ICKbnHxu9MrpyPiyiMwUkT1AXxE5X0S+EZFdIrJeREZ78nLdB8Oc/DaJyH1Bt0wVkalO+awQkR5h5HpMRP7u+Q17ROQZZ7+uiOQ7ZVDqshCRJ7HG+0XHXfSiJ8t+IrJKRHaIyDgRESevJBF5WER+FpFfHdkaep9LkFxrRaSfiJwDPARc4dxrSZjf0VVEFju/920gzXMu1HOvIyJjnPL7xdmuE5T+IeeZrhWRIZ78Gjryb3V+z8OuIpWg1ksVyi0kInK95z3/UURu9pyr8B0UkVNFZIHzLJaISIYnbZaIPCkiXwB7gWMrk+Wgwxijnzj5AKuB24DuQBHQzDnuA5YALwCHYv/YvZ1zlwMbgZ6AAL8DjnHOGeB3nvynAH9ytjMAP/AXoA5QF2gCDAIOAeoD/wI+9Fz/H+BtoDGQApzhHH8AeNuT7iJgWZjfeA6wGTjJuc8bXjkdGXcCvbAVlTRH1o7OfidgC3Cxk761c/1bTtl0BLYC/Zzzo4F84DynHJ8Cvgwj25mu3MBpwBrgf55zS4LumezsZwE3BuVlgBlAI6CVI9M5zrmhzrM+FqgHvA+87nkuG4LyWhv0e96o4B1KBX4G7nae0WXYd6mi5/44tgJyBHA4sAB4Iij98076M4A9QDvn/FTg39j3pTXwA3BDKFkjKbeg3xKc/nzgOOx7fgZWaXer7B0EmgO5zjuQBPR39g/3yLEO+04mAymx1gW1rntiLYB+nAcBvZ0/bFNn/zvgbmc73VEkySGumwWMCJNnZYagEEirQKYuwHZn+yigBGgcIt3RwG6ggbP/LvBAmDwnAU959n9HeUMwtZKyGgO84Gy7yuIEz/m/Aq8526OBOZ5z7YF9YfKtizUaTYCR2Nr3BqyyfgwYG3TPygxBb8/+O8BIZ3sucJvnXDvn2Sez/4agD/ALIJ5jCyp67liDd55nfwCw1pPeDxwa9Fv+iDWshUB7z7mbgaxQskZSbkG/pUz6EOc/xHn3qeAdBB7EMbRB/5trPXI8XhP/4wP1o66h+OFaYLYx5jdn/58E3EMtgZ+NMf4Q17XE/pGrw1ZjTL67IyKHiMh4p4m/C5gPNBLrp28JbDPGbA/OxBjzC/AFMEhEGgHnAm+GuefRwHrP/voQacocE5FTRCTTcT/sBG4BmlZwzc/OfVw2e7b3AmkSwjdujNkHLMLWNvsA87BKtJdzbF6Y3xSO4PvWc7aPdmT0ypsMNKti/qE4GthoHA3nyd9LmeceRh5v+W03xuwJcb4pttURfG3zaspeISJyroh8KSLbRGQHtobfFCp9B48BLnfcQjuca3tjKzcuod7DhEE7iuIAEakL/B7wOX5bsM3wRiLSGfuSthKR5BDGYD22uRyKvVj3i8uR2BquS3Do2XuxtdNTjDGbRaQL8A22Kb4eOExEGhljdoS41z+AG7HvVLYxZmMYmTYBLTz7LUOkCZbrn8CLwLnGmHwRGUN5Q9AS24oC64r5Jcz9K2Me1g3UFVjo7A8ATsYaxlBUNYTvL1jl5NIKW+veglWwpc/MMcKHV+Fem4DmIiIeY9CKspWF4DxceVZ40nvLr7GIHOoxBq2A5cBv2JbMMcBKzzn32e+h/PvnJeJyc/os3sP2o/3bGFMkto9KPMnCvYPrsS2Cmyq4RUKHYdYWQXxwMVCMdVt0cT4nAp9hX/yvsH/wp0XkUBFJE5FezrUTgftEpLtYficirpLJAa4SEZ/T0XhGJXLUB/YBO0TkMOBR94QxZhPwMfCS2E7lFBHp47n2Q6AbMALrNw7HO8D1InKiiByCdTFURn1sayRfRE4GrgqR5o9Oi+Yk4HpsX0Z1mIct85XGmEIc9wXwkzFma5hrtlC1Dsa3gLtFpI2I1AP+jPVv+7E+9jSxHeQpwMPYSoH3Xq0l/MiWbKxRudN5RpdijVhl8jwsIoeLSFPgEWzfjZfHRCRVRE4HBgL/Mnb47DvAkyJS33nv7vFcmwP0ETvvoiEwKijPqpRbKrYctgJ+ETkXODsozYeEfgffAC4QkQHOfyHN6QRvgQKoIYgXrgUmGzs+fbP7wdaCh2BrPRdg/enrsLX6KwCMMf8CnsTWmndj/wyHOfmOcK7b4eTzYSVyjMH6yX/Ddh5+EnT+amwN8DvgV+Au94TjVnkPaIPt/AyJMeZjYCyQie0w/dI5VVCBXLcBj4vIbqySeidEmnlOfnOBZ40xsyvIryIWYMvArf2vxPYbhGsNAPwNuEzsSKqxEdxjEvC6k+dPTv53ABhjdmJ/70RszXoPZVtx/3K+c0VkcXDGjvG6FLgO2IZ9T8I+D4c/YV1iS4FlwGLnmMtmYDu2lfAmcIsxxm193eHI+CPwOfY9nOTI8inWIC8FvsZ2nnuJuNyMMbuBO7HPfju2MjA9KE3Id9AYsx7befwQ1pCsB+5H9V8pUtaVqCjVR0QeAdoaY/6vCteciHUz1AnTB1LZ9a2xyjSlOtcrFeMMs3zDGHNA1J6r8w4qahGVGsJxJd0ATIgg7SVix643xg5j/EiVuLK/VOUdVMoSNUMgIpPETpZZHua8iMhYEVktIkvFmQSlHHiIyE3Y5vbHxpiKXCguN2NdS2uwfSO3RlE8JQGoxjuoeIiaa8jpSMzDjgnvEOL8eVj/4nnAKcDfjDGnREUYRVEUJSxRaxE4VnlbBUkuwhoJY4z5EjtU8qgK0iuKoihRIJbzCJpTdhLHBufYpuCEIjIMGAZQt27d7i1bhhp6XjklJSUkJcVnt0i8yqZyVQ2Vq+rEq2wHm1w//PDDb8aYw0OejOa0ZewU8eVhzs2g7BT8uUCPyvLs3r27qS6ZmZnVvjbaxKtsKlfVULmqTrzKdrDJBSwycRhiYiNlZ5W2IDAjUVEURaklYmkIpgPXOKOHTgV2Gjt7VVEURalFotZHICJvYSMXNhUbX/1RbIAqjDGvADOxI4ZWY2PiXB8tWRRFUZTwRM0QGGMGV3LeALdH6/6KoihKZMRfl7iiKIpSq6ghUBRFSXDUECiKoiQ4aggURVESHDUEiqIoCY4aAkVRlARHDYGiKEqCo4ZAURQlwVFDoCiKkuCoIVAURUlw1BAoiqIkOGoIFEVREhw1BIqiKAmOGgJFUZQERw2BoihKgqOGQFEUJcFRQ6AoipLgqCFQFEVJcNQQKIqiJDhqCBRFURIcNQSKoigJjhoCRVGUBEcNgaIoSoKjhkBRFCXBUUOgKIqS4KghUBRFSXDUECiKoiQ4aggURVESHDUEiqIoCY4aAkVRlARHDYGiKEqCo4ZAURQlwVFDoCiKkuCoIVAURUlw1BAoiqIkOFE1BCJyjoh8LyKrRWRkiPPHiMhcEVkqIlki0iKa8iiKoijliZohEBEfMA44F2gPDBaR9kHJngWmGmM6AY8DT0VLHkVRFCU00WwRnAysNsb8aIwpBKYBFwWlaQ/819nODHFeURQlocnOhqeest/RQowx0clY5DLgHGPMjc7+1cApxpjhnjT/BP5njPmbiFwKvAc0NcbkBuU1DBgG0KxZs+7Tpk2rlkx5eXnUq1evWtdGm3iVTeWqGipX1YlX2aor14oVDcjJaUSDBkWsWmWvP/74vAq3S0qE44/fTV5eSpnr6tb18+67LTFG8PlKOPfczfTp8xM9evirLFffvn2/Nsb0CHUuucq51Sz3AS+KyHXAfGAjUBycyBgzAZgA0KNHD5ORkVGtm2VlZVHda6NNvMqmclUNlavqxFK27GyYOtVuX3ONKw80aQIzZmykefPmdO0Kubn22OLFYAx0726vzc+Hzp1h7VoQgXr1YMwY8FddT1eK3+/jo4+aM3v2kWRm+khPr7m8o2kINgItPfstnGOlGGN+AS4FEJF6wCBjzI4oyqQoygFOdnZAWbsK+ptv7LmuXQPbwYp90SKroE891Sr0detg9mwodqqer75qv4tLq6JHRyRPNR0U1aaoSMjK4oAxBAuB40WkDdYAXAlc5U0gIk2BbcaYEmAUMCmK8iiKEkO8CtyruGfMOJ633y6rxENt+/1wyCHw0kuR1bgnTLC192Dv9+TJodMXl/NFSIS/rOYQKS+vCCQl2U9JCSQnG2q6ARU1Q2CM8YvIcGAW4AMmGWNWiMjjwCJjzHQgA3hKRAzWNXR7tORRFKVqhFPc4bYXL4a9e6FXL9i2zSrtZcsgORkaNoTnnw+nwCOreVeVkpL9zcHgGoNQCjocIuDzwT33wK5d9lhlRs7dDtXCyc2lVPFnZUGDBktIT++2vz+uDFHtIzDGzARmBh17xLP9LvBuNGVQFMVSkWIvKYHmzWH9equ409Lg738PVUuunNdfr+oV1a95V0VBhyIlBc4/325//DEUFdma9z33wLff/lKujyBSZZ6RUbOuG5f0dMjK2lXj+ca6s1hRlCpSkUL3+23nZU6OVZLu9pIlHfnmm+op9ugTqHlHgrfG3ahR6HLYvLm8Yg9VO7/mmoDCdsvVVeJZWavIyGheQ78xvlFDoCgxwKt0IDIXTEmJVWxTp1ZHoTepEbkjJZyvO5TLZMaMQM27JmvcwYq9MtLTo1OLPxBQQ6AoNYA7DHHjxuP54YfQSqxjR/j8c/j5Z/jqK1t7T0qyCjIea+rR8nUHK9u2baNT805kxV5V1BAoigfvuPLKfMMdOsBnn9lhiF995Srzo/noo8jvt/8dmpFiSEkRbrihaso8Wr5uJb5QQ6Ac1IRywbjKzT2XmmrHmP/6K8ybt7+18+gPOaxOTX3jxl8YNaq5KnUlJGoIlAOW4Nq7q/w6dYIlS2xNfc6c8i6YpCQ46yx7ruZr5JF1fPp8TmpjR+kMHRrdmnpW1irS0xOj41OpOmoIlLgluDbvKv22beH9908iOzvy2rs3XXGxnVEaCZEOT3SHIRYX/8LAgc0rVejeWa/qfqlhIuklDpXGe6yoiFavvw516oTOozr3CHdNRXnNnWv9jlF+SdQQKLVGOMXuVZBt28KXX8KWLbBggVXaIrbmXlYhHx4VGV2FfuSRlfcRhBqGWNUhh2oA9oNwyvzMM6Gw0CrxuXPLF3J2tm0SFhSUbY7deacdlpWcDEVFtDEG3norkId33O4dd9iXMzU1/D369rXN0dRUm/755+01KSlw3nn2JWvQIHA8La1sXm+/DVdeaZuw4X5LDaGGQIkKwWPdf/4ZPv00oNihdjpKvS6YpCT77RqVUL5277jyhKeq4y8jzQvKR3pz03iPV3TP7OzSmnsZxfrttzYSHFhj4A3K48qwbh3s2xdIM368fRHcac+FhYDj4Nu3z8q0ZQtcfnmgNuI2MUPd4x//gBkzrKFx8/jrXwOyFxbChx+W/0379sEf/whPPGHzGz/eHi8psde8+ipkZdGgQYNAOdYQagiUiAlXAfPW7LdutUr/tdfCK/qaiHyelGRITpbS/3+kLhh32w1Ypm6ZMLiKtrAQ6tYtWxutzK0SXKBvvQVDhljLm5xcqmgBmDiRk045xcaj+PTTwPHJk2HsWFizBi6+2B7zPrjVqwP5hFOsIrbGfcstNubFu+/al8+tHbgYEzL2RWlvz4QJMGlS6PgYqan2N2dn27Cj77xTPk1VmDvXjli44QYbn8PF57NlIkLn1FTo1q1GX1w1BEqFLF/egKwsW1l57rlAy7ZfP9i+Hf73v5pR7KEI7lD1Kv2FC39i6NBjI/4veNMdMIq/JmvkVSUrK6Bo8/MDtV7X9VJQEHBlgHWDFBWVd2HMnw833xyI/lZUVPY+fj9Nv/ii/P0LCuDWW21t4tlny062cKOwudvhXkCfD4YPL3/czcfns9e7Cr5ePdizJ9BcdPN1a+Qu7rmkJJg+3Srs226LrMPKfam9aUUCebpGyW0NHHss/PijDda0dSsYgxQVlW2F1ABqCBIYr/tm4UL7/qWn276pXbvs//z117uWu66oyE7fryoVKfaqdqi2bbuO9PRjqy5ETZOdDZmZVhGG+mMuWGB/gHs+UuXudX0E+45D3b9p0/IFFyyXe+/DDrM15Io6Mfv0CdzDGFi1KpCuoMAeKyiwzUH3GJR1lTz5pHV1VFJTCDvGym1SBjctjbGKtGFD60OfNKmsgXFDdboyVcTAgfDvf9s88/NtDd/vB58PYwzi9wfk9yrrM8+E//4XHn3UPuNQpKRYOZz8GDo08GxCTVa56y4rg7e81q2z127dWjr0zSQnq2tIqT5eN05Kig3nG1yJmTJl/+7h/lfC+d+h6pXcuKjBh1KWL71ka5zGBNwn3vSuq0DEKvMxY+yfvaCg8s6/554r6/pwFW6wO+aMM8rXsl97zX67HZ833shRbvxm13/ulen2262yqlvX7ufmwjHH2HQdO9oa75QpNvD+008HFJUxMHFiWZdJcrJVameeaQ2Ri1eJDhwILVrYgED/+Q+mqMgaA7enft06G8q0MvbsgWuvtZ9gxfrTT4EFBrwyeOM5p6bamohXtuuvh1atICODnMWL6bZ8uXXJuMrcfSaff26/g42AzwcXXGDzreiFD/XcO3a0v+PVVwN/TO8f1OnYXtKhA91q+E+hhuAgxav0u3SxLfS3367OZKlAfc3bInfzCTfKpiL/e1wo9qrgjgDxuj6KigJGAEprx638fvjhBxgxIqB0jbEK5L33AjW+goLwzftPP7VpXYyxyt3vL9sxWlBQ3ghA2WN+P7zyCscHu1DcGvD48QFFnp9vjUJxsVU6AGefbQ2B+xv+9S97vGVLG6rUvdbtiT/jDFsuwXIlJQVemjlzbG3aaY388tRTNG/ePNBB/MQTZQ2BzxdwxZx2mp3O7SrurCwYNSr0qJ033rBl5Eada9So/MxCsJ27hYXWMHg6qXcVFNjyuOaaQCfzK6+UL2OX5GQYNw6GDSt7vCo1nvR0+0caPjww6cUd3VBcDK1aseukkyLLrwqoITgICB6hs3mzHbRQE8vliRh8Pin3P4p0cMcBQUXTjwH+9reyro+pU2HWrLKKtaQEJk6kjd9va8/BhZ+SAiecEJjAUFJi1zfMzi5bwx8/PqBsXby+9VAdo8FKPoTfPCnYveHm67qT3H1X7qIiqxjbtrUth/x8q4jcmvCWLeV/X9u2oWfpJSeXdcF4fdzp6ay65x6ae10d/frZ1dpd5ey2Utw0Z50VOBfORZKebg12uOand7+idG5a14XmGg1v/4LX7VMTf4Zhw2zrwP1T33VX2d8bicuriqghOEBxdVdurtVT1VH6kYQqCNcpe9Ao/yZNyo4fd5Whz2drd+3a2aaUS3GxHUXiKjt3LKyjFAQCD8MdJVNUBP/3f9Y14x43JjAa5Z57bHyLqVPLKtGkJCtHqNqnF+9D/OwzWLHCHu/TB774omwna//+doSOa0xCGQiXwkKrhO68E555pux5r0wicO65MHNmQP6kpMAYfddFMmtW5QocKlfilSlubz41GXY0WC6IXme+VybXKLj3ycqq2XuhhuCAwtVd27ZVfYFs93/p7aCNZPhk3HTKVkZFnbDB5xYssC4Mv7+sAiwqCmz7/XbUSr165e/lNQL9+9ua/tixQFCAicGDbR7nnGOHLoZ6YH5/2THmXvr1s6NGvIbHdRV4McY21/7yFztU8ypnRdiFC+Hee+GFFzDFxUidOjB6tD03a1b5jslQFBbaBQ28rh0R2wJwa8Sur909n5RkZR89umo1by8VKedYhhUNvndtyFELv1cNQRzj9fODtw+p8ng2ZXz3DdaQm7OejEFNSB/WMZoiV0xVJgyFu95bG/MeP/PMgDsjeMy7O4u0Th1rQb1WNNhv7qWkJNBUcjsZgztGR4+2/u5QfPCBNTj79tmOTTef5GT7ICvqsPEqba8Pe8yYsiuvuIrYLZO1awPGrbDQGoh58/hp0iSOHTo0UC5z59pn4e0I9bo6vEp+0CDb0nBdIt5afjhfe7ARAI0LHceoIYhDsrOtm/m116C42FVOoRS/PSciAe/AD5vgl01cc0OKVfrejs7P6kDHSiYGRYsvvrBK0VV+kyfbUSWR3tdV6Pn5kJLC8QMGWGVpjHW7hJtNmpUVqPnu22cnF1VUC3Z7xL0xLdwa7qBBgVE/SUnw4ouB+9StC4WF1kS717odxF73i1tTXrbMdgi6wxODm2xeQxmuJh3q+WVkWJ++1wWTns66ggKODa7JpqcHOkJDdaJ68w52T3jzcalKjV+JK9QQxAkrVjQgOxs2bLADEwKtf9cAuK0AR/njx0cJ9/RZRKP2zcko/i/pfAcfObO+FifD10PLjqfOz7eKrFu30s7N0mF0lXV2zZtn/dlu2ooCcYUaNvSPf5StAYcbEhmcn7dJ5Cr0wkKO/uijgE86WLHv2GFdMlDW3QNlt90hUK7bw1XIN91kj7u1ZW8NN5RC9PiOV/32G+1efjmgiL21aW8+3rwiGWYV6fHK/OuV5VHR9v64c5T4xhhzQH26d+9uqktmZma1r40GCxYY8+c/G3PPPcaIlJjAmDj3U1LmIxSbZCkyD/Bn82dGmgWpZxgzfrwxKSnBFwY+qanhzwV/kpNtfq5gCxaYr1980ZjLLiuf54IFZX/E+PHGpKUZI2LTJCUZU7duIN1FF5W/n89n09eta6937lmar/d3+XyR/45Qn6SksvspKYF7jh9vZfD5ysrsKYdIyczMLH9dNfKpaeLt3fcSr7IdbHIBi0wYvaotghjgVnRfe807+CLY9WPw4ecCPuJIttCVxeS27UXG6omkN//ZNh0e/KMdc17RqBL33Kmn2rCeFeH326ny7kiVlBS6lJSU7+QsLIRHHrFBuLzjnb3pSkpsDX7qVOje3U5kGDAA2rSx2ytXBloIXpdNaqp1jyxbVvZ3ecMCJCVR4vcHhkRCoHZfUSS7lJSArC++WHa8d7iafnVquKFq2VpTVuIYNQS1hOs12bHDRp2taMSPYPBRxDhuZxgTAyd+cLZ/TbVhAl580Qb8KXOxBFZhcf3PjRvbmap9+5aPmeINyQllOzGLipBw/vQ5c+zHxTuSxr3GGNvD/eWXVs7+/e0olsceC3SEurjXhAsg5uWGG9i0cSPNZ8+2xsKdMPT3v1s3WLAxcMP4eseja0emopSihqAWcPs53ci3AQyu718osRO3Tv+KRln/JoMs0vky9GiVwkJrUbwKTwQuughOPtkqOmPssMXdu6FHDzsjMyur/FT8jIxAx2Vwy8IY205xhwuedx788osNRhQKn8/61zdvDijz4mI7/BDgD3+wcpx9th3q6PZdVFSLP/lkOProsiNkrrmGVQUFNB81qmwt/uKLy86sC/6dqugVJSRqCKKI2wqYMye0EfDh5yYm0pXF/EZT+pbMJz3LE7vEHWUyaJBd2MKtzbtT7r3p0tLggQfKjghyb5qVFZjBGq7TsWNHePBB27FZRkoQn8/WtocNs/n06RO+SdOqle10DjVO3R3RM2pUoEPTnTlZUBCYnOW2SNxafKjAaJ6ZqWV+hyp7RakyagiixIQJgVhe7kgfF6EYHyXlXT/BXhh3nLqrqL21eXfaebjp7VlZASVcUlJ52Nr0dFtLd6fve2fKGmNr1W66ceMCfQPJyWXHnLuK2h2n7u0I8Y53DzdzEkLPNVAlryhRQw1BFMjODhWeXBCKueiwzzl52ycB149DuSliycllx6kHK8Jw47pdMjKs4o1kSr+Ld+ihU1MvKSggKfh6byyUcFPtvePUK5tEFovZmoqilKKGoIb54gu4+mqvEbC18iT81KGQB7aNDBiAlBS7ElGDBpjnnrMds974LPsTR6Wq48lD5duxI2uDZ6SGu388hgNQFCUi1BDUIBMmBBZVAlPqArqH52jErrKtABFrBF5+GYCcVq3otmtXzXZq7q8SDjUjVVGUgw41BDWAG512wgRPNAEM/ZnDaB4jXf7nWfyCwHBGz8Ldu046qcZXHVIURYkENQT7iRvvzA11YzEkSwmjzWOkJ30FddICY9h11XRFUeIMNQT7yauveo2AAUpIppgXze3OPIBkawSCVy1SFEWJE9QQ7Afz5nnX+DWkUMgNTOIapgb6ArxDLxVFUeIQNQTVZMECuPJKb59ACTcwiZe5LZBIJPKhm4qiKDEiKZqZi8g5IvK9iKwWkZEhzrcSkUwR+UZElorIedGUp6bInrCMjNOL2bwZwJCEnzQKuOaY+XZIaFKSNQA331x2kRRFUZQ4JGotAhHxAeOA/sAGYKGITDfGrPQkexh4xxjzsoi0B2YCraMlU40wZw7Tb15EER0ASKKYfu7ooOMOgbfm6eIciqIcUETTNXQysNoY8yOAiEwDLgK8hsAADZzthsAvUZSnZhg/nizuBgKTxEbzmO0T+KKOTTNqVAwFVBRFqRpiKlq2b38yFrkMOMcYc6OzfzVwijFmuCfNUcBsoDFwKNDPGPN1iLyGAcMAmjVr1n3atGnVkikvL496oRYjrwLLBs/lzs1PYEcH+RnH7dzERAQoSUpi7dChrBsyJCayRQOVq2qoXFUnXmU72OTq27fv18aYHiFPhluxZn8/wGXARM/+1cCLQWnuAe51ttOxrYWkivKN5Qpln/3lc1OfHc6KYcb4KDR/9v3BmDp1yq9uVcuyRQuVq2qoXFUnXmU72ORif1coE5H3gdeAj40xFQSPL8NGoKVnv4VzzMsNwDmOQcoWkTSgKfBrhPeoNbInLOPMB3tSRCpgSJISUpMh48UroOP52i+gKMoBS6R9BC8B1wNjReRfwGRjzPeVXLMQOF5E2mANwJXAVUFp1gFnAVNE5EQgDdgaqfBRx40dUVDAf2afXraD+Lh1jJ56LOnpHW1aNQCKohygRGQIjDFzgDki0hAY7GyvB14F3jDGlFs01xjjF5HhwCzAB0wyxqwQkcexTZTpwL3AqyJyN7bj+DqnCRN7FiywC7A4YUSzuB6wRqAOhYy+f4/qfkVRDgoiHjUkIk2A/8P6+r8B3gR6A9cCGaGuMcbMxA4J9R57xLO9EuhVVaFrhRdeKDUC/+BqvuB07JyBYsb0eZ/0Yf8XW/kURVFqiEj7CD4A2gGvAxcYYzY5p94WkUXREi5mzJsHn3xSuvsXHnS2kjAIue1Pj41ciqIoUSDSFsFYY0xmqBMm3HCkA5XsbOjf3y6vmJLClA7P8O03J+FLKgFTQmpqEhnXHBNrKRVFUWqMSENMtBeRRu6OiDQWkdsqSH9gkp1tV5Zx1tjNLj6ZG3PuAMCXnMRNN/uYm+nTvgFFUQ4qIjUENxljdrg7xpjtwE1RkShWZGdD376wZIndT0riXbmcYmNXEi4uhlatdHCQoigHH5G6hnwiIu6IHieOUGr0xKpFsrPtHIB16+xC72CDxvXrx6Jfr4McKY0hp0FEFUU5GInUEHyC7Rge7+zf7Bw7sPEuL5aaamNKi0CdOsw8+wXm39cQsHZhzBhtDSiKcnASqWvoQSATuNX5zAUeiJZQtUZWFhQU2G23NTBgAMydy9/ntC9NpmvLKIpyMBPphLIS4GXnc/CQkWGr+858AQDmzcMYWL7cntK1ZRRFOdiJdB7B8cBTQHtsGAgAjDHHRkmu2iE9HY4+2rYKfnXCGxUW8tLTu9iwAa6/Ho4/XkMIKYpycBOpa2gytjXgB/oCU4E3oiVUrZGXBxs2wIUXQt264POR7evNiP+cDcC0aWoEFEU5+InUENQ1xszFrl/wszFmNHB+9MSqJZYssR0AF15ol5R84glmXvUGxSV2yGhhoe1GUBRFOZiJdNRQgYgkAaucQHIbgfhbsaGqfPON/e7aFVq0gPR0dt9lD/l82jegKEpiEGmLYARwCHAn0B0bfO7aaAlVa8yaBYceaucQYAOOvvUWtG4NTzyh684ripIYVNoicCaPXWGMuQ/IAyce84FOdjb85z/WNdSvH9lj/seZd3akoABSUrRvQFGUxKHSFoExphgbbvrgIivLGgGAwkKy3sstnVJQUqJ9A4qiJA6R9hF8IyLTgX8Be9yDxpj3oyJVbXDKKfbbmSjQ/dwjYLbOG1AUJfGItI8gDcgFzgQucD4DoyVUrdCihf2+7DKYO5fZG+xM4ksv1b4BRVESi0hnFh8c/QJeNm6037fcQjbpPP+83Z05E+69N3ZiKYqi1DaRziyejF1TuAzGmKE1LlFt4RqC5s2ZPqVMdwFZWdoiUBQlcYi0j2CGZzsNuAT4pebFqUU8hsA1Ajp3QFGURCRS19B73n0ReQv4PCoS1RYbN0KDBlCvHt9+C0ceCXfcYdem0daAoiiJRKQtgmCOB46oSUFqnY0boUULsrLg449tlImHHoq1UIqiKLVPRKOGRGS3iOxyP8BH2DUKDlw2biT70H6cc45donjGDDvHTFEUJdGI1DVUP9qC1DobN5J1xOml69H4/dpJrChKYhJpi+ASEWno2W8kIhdHTapoU1wMmzaR0TGXJKcEtJNYUZREJdIJZY8aY3a6O8aYHcCjUZGoNvj1VyguJv2UElq1ghNO0ElkiqIkLpEaglDpqtvRHHs+/hiAwu17WL8eLrlEjYCiKIlLpIZgkYg8LyLHOZ/nga+jKVjUyM6GW28F4PvRb+H3Q8eOMZZJURQlhkRqCO4ACoG3gWlAPnB7tISKKllZdpgQsMx/IqCGQFGUxCbSUUN7gJFRlqV2yMiwU4j9fpb5OpNsSmjbNlJ7qCiKcvAR6aihT0WkkWe/sYjMippU0SQ9HQYNgtRU5p94M02aJvH1genkUhRFqREirQo3dUYKAWCM2c6BPLO4Th2yDzuf7OUN2LIFzjpLJ5MpipK4RGoISkSklbsjIq0JEY30gGHbNjKlb7mIo4qiKIlIpENA/wB8LiLzAAFOB4ZFTapok5tLzyPXwyZdkUxRFCWiFoEx5hOgB/A98BZwL7AvinJFl9xcjjnSxpa4/HKdTKYoSmIT6cI0NwIjgBZADnAqkI1durKi684B/gb4gInGmKeDzr8A9HV2DwGOMMY0ilz8arJtG5tPtJ6um25SI6AoSmITaR/BCKAn8LMxpi/QFdhR0QUi4gPGAecC7YHBItLem8YYc7cxposxpgvwd+D9KklfHUpKrCHwNQfsOgSKoiiJTKSGIN8Ykw8gInWMMd8B7Sq55mRgtTHmR2NMIXYi2kUVpB+MdTtFl507oaSEzSV20JMaAkVREp1IO4s3OPMIPgQ+FZHtwM+VXNMcWO/NAzglVEIROQZoA/w3zPlhOJ3TzZo1I6uaQ3zy8vL4cuZMTgWWbwGfr4SlS+eXRiCNJXl5edX+XdFE5aoaKlfViVfZEkouY0yVPsAZwIVAaiXpLsP2C7j7VwMvhkn7IPD3SO7fvXt3U10yMzON+d//jAEztP8607x5tbOqcTIzM2MtQkhUrqqhclWdeJXtYJMLWGTC6NUqRxA1xsyLMOlGoKVnv4VzLBRXUluxi3JzAdicV49mzWrljoqiKHFNNJ0iC4HjRaSNiKRilf304EQicgLQGDsKKfps2wbA5l11tX9AURSFKBoCY4wfGA7MAr4F3jHGrBCRx0XkQk/SK4FpTtMl+rgtgtwUNQSKoihEeXEZY8xMYGbQsUeC9kdHU4ZybNtGCUls2ZqkhkBRFIXouobik9xctjVsQ3GxqCFQFEUhQQ3B5vrHA2hnsaIoColoCLZt479J/QD47bcYy6IoihIHJJwhyP6uMfevuwOAe+/VdQgURVESyhA0WLGCrJ/bUIQPgKJCo+sQKIqS8ER11FC80SgnhwyWk0wxRfhITfKTkZESa7EURVFiSkK1CHZ07kw6XzKYtxCKmTPuOw1BrShKwpNQhmDPcccB0OiYhjSsbzhtWMcYS6QoihJ7Eso15NuzB4C85idQrzihfrqiKEpYEqpFkLx3LwB5JYdQr16MhVEURYkTEsoQlLYI/GlqCBRFURwSyhAku4agqI4aAkVRFIeEMgSlLYLCFDUEiqIoDgllCEr7CArUECiKorgklCEobRHs86khUBRFcUgoQ1DaItgjaggURVEcEssQ7NmDqXsIeXlqCBRFUVwSyhD49uwhv8ERlJSghkBRFMUhoQxB8t695NWzy5KpIVAURbEklCHw7d3L7kPssmRqCBRFUSwJZQiS8/LIq3s4APXrx1gYRVGUOCGhDIFv717y0poC2iJQFEVxSShDkLx3L3mphwFqCBRFUVwSyhD49uxRQ6AoihJE4hgCY2yLILkRoIZAURTFJXEMwZ49SEkJeUkNADUEiqIoLoljCHbtAiBP7HAhNQSKoiiWBDQE9RCBunVjLI+iKEqckHiGoORQ6tUDkRjLoyiKEickniEorqtuIUVRFA+JYwh27gQgz6/LVCqKonhJHEOweDEAeVvz1RAoiqJ4SAxDkJ0NzzwDQN7i76lvdsVYIEVRlPghMQxBVhYUFwNOZ/G+X2Mrj6IoShwRVUMgIueIyPcislpERoZJ83sRWSkiK0Tkn1ERJCMD6tShJCmJXzmc9f6jyM6Oyp0URVEOOKJmCETEB4wDzgXaA4NFpH1QmuOBUUAvY8xJwF1RESY9HebO5f3z/8Q6jmH5j4dy1lmoMVAURSG6LYKTgdXGmB+NMYXANOCioDQ3AeOMMdsBjDHR89mkpzOr2RUYBGOgsNB6jBRFURKd5Cjm3RxY79nfAJwSlKYtgIh8AfiA0caYT4IzEpFhwDCAZs2akVVNDd6uXTLQBhFITi6hQYMlZGXFR8dxXl5etX9XNFG5qobKVXXiVbaEkssYE5UPcBkw0bN/NfBiUJoZwAdACtAGazgaVZRv9+7dTXXJzMw0Pp8xGRnGLFhQ7WyiQmZmZqxFCInKVTVUrqoTr7IdbHIBi0wYvRpN19BGoKVnv4VzzMsGYLoxpsgY8xPwA3B8tAQqKbGDh/r2td0GiqIoSnT7CBYCx4tIGxFJBa4Epgel+RDIABCRplhX0Y/REsjvtz83NTVad1AURTnwiJohMMb4geHALOBb4B1jzAoReVxELnSSzQJyRWQlkAncb4zJjZZMhYU20lydOtG6g6IoyoFHNDuLMcbMBGYGHXvEs22Ae5xP1NEWgaIoSnmiagjijaIiawi0RaDEkqKiIjZs2EB+fn6t3bNhw4Z8++23tXa/qhCvsh2ocqWlpdGiRQtSUlIizjPBDIF1DWmLQIklGzZsoH79+rRu3RqppYUxdu/eTf369WvlXlUlXmU7EOUyxpCbm8uGDRto06ZNxHkmRqwhB20RKPFAfn4+TZo0qTUjoCQOIkKTJk2q3NpMMEOgLQIlPlAjoESL6rxbCWYItEWgKIoSTEIZAh01pCjQt29fZs2aVebYmDFjuPXWW8Nek5GRwaJFiwA477zz2LFjR7k0o0eP5tlnn63w3h9++CErV64s3X/kkUfIzMysgvQ1Q1ZWFgsWLKj1+8YrCWUIdB6BcsCSnQ1PPVUjIXMHDx7MtGnTyhybNm0agwcPjuj6mTNn0qhRo2rdO9gQPP744/Tt27daee0P8WIIip11UmJNQhkCbREoccddd9n1Mir6dO0KvXvDQw/Z765dK05/110V3vKyyy7jP//5D4WFhQCsXbuWX375hdNPP51bb72VHj16cNJJJ/Hoo4+GvL5169b89ttvADz55JO0bduW3r178/3335emefXVV+nZsyedO3dm0KBB7N27lwULFjB9+nTuv/9+unTpwpo1a7juuuv48MMPAZg7dy5du3alY8eODB06lIKCgtL7Pfroo3Tr1o2OHTvy3XffhZRr5MiRtG/fnk6dOnHfffcBsHXrVgYNGkTPnj3p2bMnX3zxBWvXruWVV17hhRdeoEuXLnz22Wdl8vnqq69IT0+nd+/enHbaaaW/q7i4mPvuu48OHTrQqVMn/v73vwOwcOFCTjvtNDp37szJJ5/M7t27mTJlCsOHDy/Nc+DAgaWB4urVq8e9995L586dyc7O5vHHH6dnz5506NCBYcOGuXHYWL16Nf369aNz585069aNNWvWcM011zBjxozSfIcMGcK///3vCp93JCSUIdA+AuWAZOdOGygL7PfOnfuV3WGHHcbJJ5/Mxx9/DNjWwO9//3tEhCeffJJFixaxdOlS5s2bx9KlS8Pm8/XXXzNt2jRycnKYOXMmCxcuLD136aWXsnDhQpYsWcKJJ57Ia6+9xmmnncaFF17IM888Q05ODscdd1xp+vz8fK677jrefvttli1bht/v5+WXXy4937RpUxYvXsytt94a0v2Um5vLBx98wIoVK1i6dCkPP/wwACNGjODuu+9m4cKFvPfee9x44420bt2aW265hbvvvpucnBxOP/30MnmdcMIJfPbZZ3z++ec8/vjjPPTQQwBMmDCBtWvXkpOTw9KlSxkyZAiFhYVcccUV/O1vf2PJkiXMmTOHunXrVlj+e/bs4ZRTTmHJkiX07t2b4cOHs3DhQpYvX86+fftKFf2QIUO4/fbbWbJkCQsWLOCoo47ihhtu4M033wRg586dLFiwgPPPP7/C+0WCziNQlFgyZkzlabKz4ayz7CIaqanw5pv7HTXRdQ9ddNFFTJs2jddeew2Ad955hwkTJuD3+9m0aRMrV66kU6dOIfP47LPPuOSSSzjkkEMAuPDCC0vPLV++nIcffpgdO3aQl5fHgAEDKpTn+++/p02bNrRt2xaAa6+9lnHjxnGX07q59NJLAejevTvvv/9+uesbNmxIWloaN9xwAwMHDmTgwIEAzJkzp4wrateuXeTl5VUoy86dO7n22mv5/vvv8fl8FBUVleZ1yy23kJxs1eZhhx3GsmXLOOqoo+jZsycADRo0qDBvAJ/Px6BBg0r3MzMz+etf/8revXvZtm0bJ510EhkZGWzcuJFLLrkEsJPEAM444wxuueUWtm7dynvvvcegQYNK5dkfEswQaItAOQBxVtgjK8u6fmogdO5FF13E3XffzeLFi9m7dy/du3fnp59+4tlnn2XhwoU0btyY6667rtqzn12XT+fOnZkyZcp+x8+v4/xpfT4ffr8fgAEDBrBlyxZ69OjBxIkT+eqrr5g7dy7vvvsuL774Iv/9738pKSnhyy+/LFWkkfDHP/6Rvn37MnXqVHJzc8nIyKiyvMnJyZS4rTgoU45paWn4fL7S47fddhuLFi2iZcuWjB49utIyHzx4MG+88QbTpk1j8uTJVZYtFAnmGtIWgXKAkp4Oo0bVWPz0evXq0bdvX4YOHVraSbxr1y4OPfRQGjZsyJYtW0pdR+Ho06cPH374Ifv27WP37t189NFHped2797NUUcdRVFRUakrA6B+/frs3r27XF7t2rVj7dq1rF69GoDXX3+dM844o8L7z5o1i5ycHCZOnEheXh47d+7kvPPO44UXXmDJkiUAnH322aW+fICcnJwK5QDbImjevDkAU6ZMKT3ev39/xo8fX2qItm3bRrt27di0aVOpW2z37t34/X5at25NTk4OJSUlrF+/nq+++irkvVyl37RpU/Ly8nj33XdL5WvRokVp/0lBQQF79+4FrMtojNOSbN++fbk8q0OCGQJtESiKy+DBg1myZEmpIejcuTNdu3blhBNO4KqrrqJXr14VXt+tWzeuuOIKOnfuzLnnnlvqHgF44oknOOWUU+jVqxcnnHBC6fErr7ySZ555hq5du7JmzZrS42lpaUyePJnLL7+cjh07kpSUxC233BLxb9m9ezcDBw6kU6dO9O7dm+effx6AsWPHsmjRIjp16kT79u155ZVXALjgggv44IMPQnYWP/DAA4waNYrevXuXKn2AG2+8kVatWtGpUyc6d+7MP//5T1JTU3n77be544476Ny5M/379yc/P59evXrRpk0b2rdvz5133km3bt1Cyt2oUSNuuukmOnTowIABA8qU4euvv87YsWPp1KkTp512Gps3bwbgiCOO4MQTT+T666+PuHwqJdyKNfH62Z8Vym6/fZUBY7Zvr3YWUeNgWw0p2hzIcq1cuTL6ggSxa9euWr9npMSrbPEq1+bNm82xxx5rduzYETZNqHeMGK1QFne4riFtESiKciAyZ84cevbsyR133EHDhg1rLN+E7CzWPgJFUQ5E+vXrx4oVK2o8KmqCtQiS8PnA6bBXFEVRSDhDINoaUBRFCSLBDEGS9g8oiqIEkVCGwO/XFoGiKEowCWUICgu1RaAoubm5dOnShS5dunDkkUfSvHnz0n03EF04Fi1axJ133hk12Xbs2MFLL70UtfyV0CTUqCG/P0lbBMoBSXZ2zUWYaNKkSekM29GjR1OvXr3SaJ0Afr8/bPyaHj160KNHj/0ToAJcQ3DbbbdF7R6R4J1IlggklCEoKhJtEShxxV13gaOTw7JzJyxdagOPJiVBp05Q0RDyLl0ii2Xn5brrriMtLY1vvvmGXr16ceWVVzJixAjy8/OpW7cukydPpl27dmRlZfHss88yY8YMRo8ezbp16/jxxx9Zt24dd911V8jWwrx58xgxYgRgl1GcP38+9evX55lnnuGdd95h3759DBo0iMcee4yRI0eyZs0aunTpQv/+/XnmmWfK5HXxxRezfv168vPzGTFiBMOGDQPgk08+4aGHHqK4uJimTZsyd+5c8vLyuOOOO1i0aBEiwqOPPsqgQYOoV69eaeC5d999lxkzZjBlypRyZXDBBRfw0EMPlSuD4uJiHnzwQT755BOSkpK46aabOOmkkxg7dmxpSIhPP/2Ul156iQ8++KBqDyJGJJgh0BaBcuARKgp1Dc4lKmXDhg0sWLAAn8/Hrl27+Oyzz0hOTmbOnDk89NBDvPfee+Wu+e6778jMzGT37t20a9eOW2+9lZSUlDJpnn32WcaNG0evXr3Iy8sjLS2N2bNns2rVKr766it27drFkCFDmD9/Pk8//TTLly8vbbEEM2nSJA477DD27dtHz549GTRoECUlJdx0003Mnz+fNm3asG3bNsCGuWjYsCHLli0DYPv27VUqg40bN4YsA2846uTkZLZt20bjxo257bbb2Lp1K4cffjiTJ09m6NChVXwCsSPhDIG2CJR4IkZRqENy+eWXl0bFdEMxr1q1ChEpDcUczPnnn0+dOnWoU6cORxxxBFu2bKFFixZl0vTq1Yt77rmHIUOGcOmll9KiRQtmz57N7Nmz6dq1KyUlJezdu5dVq1bRqlWrCmUcO3ZsaS17/fr1rFq1iq1bt9KnTx/atGkD2PDQYGfheldia9y4cZXKYNeuXQwfPrxcGYQKRw1w9dVX88Ybb3D99deTnZ3N1KlTK71fvJBQhsDvF+rVi7UUilI1ohCFOiSHHnpo6bYbivmDDz5g7dq1YUMx1/HUrNwQ0ePGjePVV18F7LKWI0eO5Pzzz2fmzJn06tWLWbNmYYxh1KhR3Hzzzezevbt0puzatWvDypeVlcWcOXPIzs7mkEMOISMjo1phskWkdDv4em8Z/OlPf4qoDFyuv/56LrjgAtLS0rj88strZJ2A2kJHDSnKAUANR6GulHChmCPh9ttvJycnh5ycHI4++mjWrFlDx44defDBB+nZsyffffcdAwYMYNKkSaW++o0bN/Lrr79WGh66cePGHHLIIXz33Xd8+eWXAJx66qnMnz+fn376CaDUNdS/f3/GjRtXer3rGmrWrBnffvstJSUlFfrwd+3aFXE4aoCjjz6ao48+mj/96U81Gxm0FkgoQ6DzCBQlMtxQzF27dt3vETRjxowpXec3JSWFc889l7PPPpurrrqK9PR0Tj31VC677DJ2795NkyZN6NWrFx06dOD+++8vk88555yD3+/nxBNPZOTIkZx66qkAHH744UyYMIFLL72Uzp07c8UVVwDw8MMPs337djp06EDnzp3JzMwE4Omnn2bgwIGcdtppHHXUUWHlHjFiRMgyCBWO2mXIkCG0bNmSE088cb/KrNYJF5Y0Xj/7E4b62GN3m0suqfblUeVADqscCw5kuTQMdVniVbbqyHX77bebiRMnRkGaAJHIVdUw1AeOE6sG0FFDiqJEi+7du3PooYfy3HPPxVqUKpNghkDnESiKEh2+/vrrWItQbRKsj0BbBEp8YFvqilLzVOfdSihDoKOGlHggLS2N3NxcNQZKjWOMITc3l7S0tCpdl1CuIR01pMQDLVq0YMOGDWzdurXW7pmfn19l5VBbxKtsB6pcaWlp5Sb1VUZCGQKdWazEAykpKaWzYGuLrKwsunbtWqv3jJR4lS2R5Iqqa0hEzhGR70VktYiMDHH+OhHZKiI5zufGaMlijI4aUhRFCUXUWgQi4gPGAf2BDcBCEZlujFkZlPRtY8zwaMnh4oZK0RaBoihKWaLZIjgZWG2M+dEYUwhMAy6K4v0qxF1vQ1sEiqIoZYlmH0FzYL1nfwNwSoh0g0SkD/ADcLcxZn1wAhEZBgxzdvNE5PtqytT0/vv5LWjmerzQFPgt1kKEQOWqGipX1YlX2Q42uY4JdyLWncUfAW8ZYwpE5GbgH8CZwYmMMROACft7MxFZZIyJ3vJK+0G8yqZyVQ2Vq+rEq2yJJFc0XUMbgZae/RbOsVKMMbnGmAJndyLQPYryKIqiKCGIpiFYCBwvIm1EJBW4EpjuTSAi3tB/FwLfRlEeRVEUJQRRcw0ZY/wiMhyYBfiAScaYFSLyODYK3nTgThG5EPAD24DroiWPw367l6JIvMqmclUNlavqxKtsCSOX6DR3RVGUxCahYg0piqIo5VFDoCiKkuAkjCGoLNxFLcrRUkQyRWSliKwQkRHO8dEistETbuO8GMi2VkSWOfdf5Bw7TEQ+FZFVznfjWpapnadMckRkl4jcFavyEpFJIvKriCz3HAtZRmIZ67xzS0WkWy3L9YyIfOfc+wMRaeQcby0i+zxl90otyxX22YnIKKe8vheRAdGSqwLZ3vbItVZEcpzjtVJmFeiH6L5j4ZYuO5g+2M7qNcCxQCqwBGgfI1mOAro52/WxE+naA6OB+2JcTmuBpkHH/gqMdLZHAn+J8XPcjJ0YE5PyAvoA3YDllZURcB7wMSDAqcD/almus4FkZ/svHrlae9PFoLxCPjvnf7AEqAO0cf6zvtqULej8c8AjtVlmFeiHqL5jidIiiJtwF8aYTcaYxc72buyQ2eaxkCVCLsJO9MP5vjh2onAWsMYY83OsBDDGzMeOcPMSrowuAqYay5dAo6Ah01GVyxgz2xjjrrr+JXYuT60SprzCcREwzRhTYIz5CViN/e/WumwiIsDvgbeidf8wMoXTD1F9xxLFEIQKdxFz5SsirYGuwP+cQ8Od5t2k2nbBOBhgtoh8LTasB0AzY8wmZ3sz0CwGcrlcSdk/ZqzLyyVcGcXTezcUW3N0aSMi34jIPBE5PQbyhHp28VRepwNbjDGrPMdqtcyC9ENU37FEMQRxh4jUA94D7jLG7AJeBo4DugCbsM3S2qa3MaYbcC5wu9gYUKUY2xaNyXhjsZMSLwT+5RyKh/IqRyzLKBwi8gfsXJ03nUObgFbGmK7APcA/RaRBLYoUl88uiMGUrXTUapmF0A+lROMdSxRDUGm4i9pERFKwD/lNY8z7AMaYLcaYYmNMCfAqUWwSh8MYs9H5/hX4wJFhi9vUdL5/rW25HM4FFhtjtjgyxry8PIQro5i/dyJyHTAQGOIoEBzXS66z/TXWF9+2tmSq4NnFvLwARCQZuBR42z1Wm2UWSj8Q5XcsUQxBpeEuagvH9/ga8K0x5nnPca9f7xJgefC1UZbrUBGp725jOxqXY8vpWifZtcC/a1MuD2VqaLEuryDCldF04BpnZMepwE5P8z7qiMg5wAPAhcaYvZ7jh4tdLwQRORY4HvixFuUK9+ymA1eKSB0RaePI9VVtyeWhH/CdMWaDe6C2yiycfiDa71i0e8Hj5YPtXf8Ba8n/EEM5emObdUuBHOdzHvA6sMw5Ph04qpblOhY7YmMJsMItI6AJMBdYBcwBDotBmR0K5AINPcdiUl5YY7QJKML6Y28IV0bYkRzjnHduGdCjluVajfUfu+/ZK07aQc4zzgEWAxfUslxhnx3wB6e8vgfOre1n6RyfAtwSlLZWyqwC/RDVd0xDTCiKoiQ4ieIaUhRFUcKghkBRFCXBUUOgKIqS4KghUBRFSXDUECiKoiQ4aggUpRYRkQwRmRFrORTFixoCRVGUBEcNgaKEQET+T0S+cmLPjxcRn4jkicgLTpz4uSJyuJO2i4h8KYG4/26s+N+JyBwRWSIii0XkOCf7eiLyrti1At50ZpMqSsxQQ6AoQYjIicAVQC9jTBegGBiCneG8yBhzEjAPeNS5ZCrwoDGmE3Z2p3v8TWCcMaYzcBp2FivYiJJ3YePMHwv0ivJPUpQKSY61AIoSh5wFdAcWOpX1utggXyUEApG9AbwvIg2BRsaYec7xfwD/cuI2NTfGfABgjMkHcPL7yjhxbMSugNUa+Dzqv0pRwqCGQFHKI8A/jDGjyhwU+WNQuurGZynwbBej/0MlxqhrSFHKMxe4TESOgNL1Yo/B/l8uc9JcBXxujNkJbPcsVHI1MM/Y1aU2iMjFTh51ROSQ2vwRihIpWhNRlCCMMStF5GHsam1J2OiUtwN7gJOdc79i+xHAhgV+xVH0PwLXO8evBsaLyONOHpfX4s9QlIjR6KOKEiEikmeMqRdrORSlplHXkKIoSoKjLQJFUZQER1sEiqIoCY4aAkVRlARHDYGiKEqCo4ZAURQlwVFDoCiKkuD8PxW9ogFuF/sMAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# accuracy 값을 plot 해보겠습니다.\n",
    "y_vacc = history.history['val_accuracy']\n",
    "y_acc = history.history['accuracy']\n",
    "x_len = np.arange(len(y_acc))\n",
    "\n",
    "plt.plot(x_len, y_vacc, marker='.', c='red', label=\"Validation-set accuracy\")\n",
    "plt.plot(x_len, y_acc, marker='.', c='blue', label=\"Train-set accuracy\")\n",
    "plt.legend(loc='lower right')\n",
    "plt.grid()\n",
    "plt.ylim(0.5,1) \n",
    "plt.title('Accuracy graph without dropout layer') \n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('accuracy')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "allied-damages",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/200\n",
      "117/117 [==============================] - 1s 4ms/step - loss: 2.2140 - accuracy: 0.4025 - val_loss: 1.6939 - val_accuracy: 0.5817\n",
      "Epoch 2/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 1.5796 - accuracy: 0.5388 - val_loss: 1.2251 - val_accuracy: 0.6700\n",
      "Epoch 3/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 1.2064 - accuracy: 0.6161 - val_loss: 1.0097 - val_accuracy: 0.7250\n",
      "Epoch 4/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 1.0263 - accuracy: 0.6594 - val_loss: 0.8788 - val_accuracy: 0.7350\n",
      "Epoch 5/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.9093 - accuracy: 0.6937 - val_loss: 0.7910 - val_accuracy: 0.7600\n",
      "Epoch 6/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.8295 - accuracy: 0.7114 - val_loss: 0.7373 - val_accuracy: 0.7517\n",
      "Epoch 7/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.7766 - accuracy: 0.7252 - val_loss: 0.6940 - val_accuracy: 0.7733\n",
      "Epoch 8/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.7425 - accuracy: 0.7367 - val_loss: 0.6666 - val_accuracy: 0.7800\n",
      "Epoch 9/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.7091 - accuracy: 0.7462 - val_loss: 0.6434 - val_accuracy: 0.7917\n",
      "Epoch 10/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.6945 - accuracy: 0.7485 - val_loss: 0.6264 - val_accuracy: 0.7883\n",
      "Epoch 11/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.6637 - accuracy: 0.7605 - val_loss: 0.6096 - val_accuracy: 0.7967\n",
      "Epoch 12/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.6517 - accuracy: 0.7653 - val_loss: 0.5933 - val_accuracy: 0.7917\n",
      "Epoch 13/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.6385 - accuracy: 0.7686 - val_loss: 0.5790 - val_accuracy: 0.7950\n",
      "Epoch 14/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.6200 - accuracy: 0.7785 - val_loss: 0.5703 - val_accuracy: 0.7917\n",
      "Epoch 15/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.6089 - accuracy: 0.7834 - val_loss: 0.5574 - val_accuracy: 0.7933\n",
      "Epoch 16/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5993 - accuracy: 0.7864 - val_loss: 0.5481 - val_accuracy: 0.7967\n",
      "Epoch 17/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5849 - accuracy: 0.7906 - val_loss: 0.5408 - val_accuracy: 0.7950\n",
      "Epoch 18/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5759 - accuracy: 0.7959 - val_loss: 0.5321 - val_accuracy: 0.8017\n",
      "Epoch 19/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5690 - accuracy: 0.7961 - val_loss: 0.5224 - val_accuracy: 0.8050\n",
      "Epoch 20/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5622 - accuracy: 0.8013 - val_loss: 0.5143 - val_accuracy: 0.8067\n",
      "Epoch 21/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5508 - accuracy: 0.8045 - val_loss: 0.5104 - val_accuracy: 0.8167\n",
      "Epoch 22/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5422 - accuracy: 0.8068 - val_loss: 0.5031 - val_accuracy: 0.8150\n",
      "Epoch 23/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5366 - accuracy: 0.8102 - val_loss: 0.4967 - val_accuracy: 0.8150\n",
      "Epoch 24/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5301 - accuracy: 0.8110 - val_loss: 0.4927 - val_accuracy: 0.8100\n",
      "Epoch 25/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5227 - accuracy: 0.8159 - val_loss: 0.4878 - val_accuracy: 0.8167\n",
      "Epoch 26/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5162 - accuracy: 0.8187 - val_loss: 0.4829 - val_accuracy: 0.8167\n",
      "Epoch 27/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5095 - accuracy: 0.8191 - val_loss: 0.4780 - val_accuracy: 0.8250\n",
      "Epoch 28/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5113 - accuracy: 0.8179 - val_loss: 0.4735 - val_accuracy: 0.8200\n",
      "Epoch 29/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5028 - accuracy: 0.8219 - val_loss: 0.4677 - val_accuracy: 0.8183\n",
      "Epoch 30/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.5006 - accuracy: 0.8248 - val_loss: 0.4655 - val_accuracy: 0.8217\n",
      "Epoch 31/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4936 - accuracy: 0.8252 - val_loss: 0.4596 - val_accuracy: 0.8217\n",
      "Epoch 32/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4918 - accuracy: 0.8268 - val_loss: 0.4574 - val_accuracy: 0.8250\n",
      "Epoch 33/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4808 - accuracy: 0.8317 - val_loss: 0.4531 - val_accuracy: 0.8267\n",
      "Epoch 34/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4866 - accuracy: 0.8283 - val_loss: 0.4496 - val_accuracy: 0.8233\n",
      "Epoch 35/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4786 - accuracy: 0.8304 - val_loss: 0.4501 - val_accuracy: 0.8283\n",
      "Epoch 36/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4755 - accuracy: 0.8325 - val_loss: 0.4441 - val_accuracy: 0.8283\n",
      "Epoch 37/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4731 - accuracy: 0.8327 - val_loss: 0.4413 - val_accuracy: 0.8300\n",
      "Epoch 38/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4714 - accuracy: 0.8356 - val_loss: 0.4381 - val_accuracy: 0.8283\n",
      "Epoch 39/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4659 - accuracy: 0.8349 - val_loss: 0.4334 - val_accuracy: 0.8317\n",
      "Epoch 40/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4670 - accuracy: 0.8378 - val_loss: 0.4339 - val_accuracy: 0.8333\n",
      "Epoch 41/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4577 - accuracy: 0.8387 - val_loss: 0.4301 - val_accuracy: 0.8300\n",
      "Epoch 42/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4552 - accuracy: 0.8396 - val_loss: 0.4282 - val_accuracy: 0.8333\n",
      "Epoch 43/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4588 - accuracy: 0.8367 - val_loss: 0.4254 - val_accuracy: 0.8350\n",
      "Epoch 44/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4480 - accuracy: 0.8429 - val_loss: 0.4240 - val_accuracy: 0.8300\n",
      "Epoch 45/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4523 - accuracy: 0.8389 - val_loss: 0.4217 - val_accuracy: 0.8350\n",
      "Epoch 46/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4448 - accuracy: 0.8437 - val_loss: 0.4186 - val_accuracy: 0.8367\n",
      "Epoch 47/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4488 - accuracy: 0.8415 - val_loss: 0.4174 - val_accuracy: 0.8383\n",
      "Epoch 48/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4415 - accuracy: 0.8460 - val_loss: 0.4160 - val_accuracy: 0.8300\n",
      "Epoch 49/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4361 - accuracy: 0.8449 - val_loss: 0.4141 - val_accuracy: 0.8400\n",
      "Epoch 50/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4399 - accuracy: 0.8451 - val_loss: 0.4127 - val_accuracy: 0.8350\n",
      "Epoch 51/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4351 - accuracy: 0.8467 - val_loss: 0.4128 - val_accuracy: 0.8350\n",
      "Epoch 52/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4369 - accuracy: 0.8448 - val_loss: 0.4102 - val_accuracy: 0.8417\n",
      "Epoch 53/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4378 - accuracy: 0.8432 - val_loss: 0.4085 - val_accuracy: 0.8383\n",
      "Epoch 54/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4305 - accuracy: 0.8496 - val_loss: 0.4084 - val_accuracy: 0.8350\n",
      "Epoch 55/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4262 - accuracy: 0.8501 - val_loss: 0.4023 - val_accuracy: 0.8433\n",
      "Epoch 56/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4301 - accuracy: 0.8477 - val_loss: 0.4035 - val_accuracy: 0.8417\n",
      "Epoch 57/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4229 - accuracy: 0.8510 - val_loss: 0.4053 - val_accuracy: 0.8433\n",
      "Epoch 58/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4241 - accuracy: 0.8503 - val_loss: 0.4026 - val_accuracy: 0.8467\n",
      "Epoch 59/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4188 - accuracy: 0.8501 - val_loss: 0.3992 - val_accuracy: 0.8433\n",
      "Epoch 60/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4233 - accuracy: 0.8514 - val_loss: 0.3975 - val_accuracy: 0.8433\n",
      "Epoch 61/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4231 - accuracy: 0.8508 - val_loss: 0.3994 - val_accuracy: 0.8450\n",
      "Epoch 62/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4159 - accuracy: 0.8549 - val_loss: 0.3918 - val_accuracy: 0.8450\n",
      "Epoch 63/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4177 - accuracy: 0.8524 - val_loss: 0.3964 - val_accuracy: 0.8467\n",
      "Epoch 64/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4217 - accuracy: 0.8508 - val_loss: 0.3930 - val_accuracy: 0.8433\n",
      "Epoch 65/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4137 - accuracy: 0.8543 - val_loss: 0.3919 - val_accuracy: 0.8417\n",
      "Epoch 66/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4074 - accuracy: 0.8568 - val_loss: 0.3935 - val_accuracy: 0.8433\n",
      "Epoch 67/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4099 - accuracy: 0.8565 - val_loss: 0.3897 - val_accuracy: 0.8450\n",
      "Epoch 68/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4089 - accuracy: 0.8552 - val_loss: 0.3864 - val_accuracy: 0.8467\n",
      "Epoch 69/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4124 - accuracy: 0.8525 - val_loss: 0.3877 - val_accuracy: 0.8500\n",
      "Epoch 70/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4055 - accuracy: 0.8568 - val_loss: 0.3824 - val_accuracy: 0.8450\n",
      "Epoch 71/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3963 - accuracy: 0.8597 - val_loss: 0.3827 - val_accuracy: 0.8467\n",
      "Epoch 72/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4058 - accuracy: 0.8553 - val_loss: 0.3817 - val_accuracy: 0.8483\n",
      "Epoch 73/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3981 - accuracy: 0.8596 - val_loss: 0.3809 - val_accuracy: 0.8483\n",
      "Epoch 74/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4038 - accuracy: 0.8592 - val_loss: 0.3779 - val_accuracy: 0.8467\n",
      "Epoch 75/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3973 - accuracy: 0.8617 - val_loss: 0.3818 - val_accuracy: 0.8450\n",
      "Epoch 76/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4002 - accuracy: 0.8591 - val_loss: 0.3774 - val_accuracy: 0.8483\n",
      "Epoch 77/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.4001 - accuracy: 0.8587 - val_loss: 0.3749 - val_accuracy: 0.8467\n",
      "Epoch 78/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3994 - accuracy: 0.8593 - val_loss: 0.3804 - val_accuracy: 0.8467\n",
      "Epoch 79/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3947 - accuracy: 0.8596 - val_loss: 0.3762 - val_accuracy: 0.8450\n",
      "Epoch 80/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3942 - accuracy: 0.8593 - val_loss: 0.3766 - val_accuracy: 0.8450\n",
      "Epoch 81/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3981 - accuracy: 0.8606 - val_loss: 0.3750 - val_accuracy: 0.8450\n",
      "Epoch 82/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3977 - accuracy: 0.8583 - val_loss: 0.3719 - val_accuracy: 0.8450\n",
      "Epoch 83/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3996 - accuracy: 0.8593 - val_loss: 0.3726 - val_accuracy: 0.8450\n",
      "Epoch 84/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3856 - accuracy: 0.8627 - val_loss: 0.3729 - val_accuracy: 0.8433\n",
      "Epoch 85/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3887 - accuracy: 0.8620 - val_loss: 0.3716 - val_accuracy: 0.8517\n",
      "Epoch 86/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3856 - accuracy: 0.8658 - val_loss: 0.3696 - val_accuracy: 0.8500\n",
      "Epoch 87/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3890 - accuracy: 0.8648 - val_loss: 0.3660 - val_accuracy: 0.8467\n",
      "Epoch 88/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3905 - accuracy: 0.8616 - val_loss: 0.3682 - val_accuracy: 0.8483\n",
      "Epoch 89/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3858 - accuracy: 0.8657 - val_loss: 0.3668 - val_accuracy: 0.8500\n",
      "Epoch 90/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3865 - accuracy: 0.8643 - val_loss: 0.3662 - val_accuracy: 0.8467\n",
      "Epoch 91/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3795 - accuracy: 0.8635 - val_loss: 0.3674 - val_accuracy: 0.8483\n",
      "Epoch 92/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3842 - accuracy: 0.8652 - val_loss: 0.3624 - val_accuracy: 0.8433\n",
      "Epoch 93/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3813 - accuracy: 0.8634 - val_loss: 0.3641 - val_accuracy: 0.8450\n",
      "Epoch 94/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3796 - accuracy: 0.8659 - val_loss: 0.3654 - val_accuracy: 0.8533\n",
      "Epoch 95/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3832 - accuracy: 0.8647 - val_loss: 0.3611 - val_accuracy: 0.8467\n",
      "Epoch 96/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3784 - accuracy: 0.8664 - val_loss: 0.3640 - val_accuracy: 0.8450\n",
      "Epoch 97/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3819 - accuracy: 0.8655 - val_loss: 0.3609 - val_accuracy: 0.8467\n",
      "Epoch 98/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3802 - accuracy: 0.8657 - val_loss: 0.3625 - val_accuracy: 0.8550\n",
      "Epoch 99/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3773 - accuracy: 0.8654 - val_loss: 0.3587 - val_accuracy: 0.8517\n",
      "Epoch 100/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3744 - accuracy: 0.8688 - val_loss: 0.3594 - val_accuracy: 0.8483\n",
      "Epoch 101/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3752 - accuracy: 0.8657 - val_loss: 0.3616 - val_accuracy: 0.8517\n",
      "Epoch 102/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3761 - accuracy: 0.8679 - val_loss: 0.3586 - val_accuracy: 0.8600\n",
      "Epoch 103/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3726 - accuracy: 0.8682 - val_loss: 0.3573 - val_accuracy: 0.8567\n",
      "Epoch 104/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3787 - accuracy: 0.8670 - val_loss: 0.3607 - val_accuracy: 0.8517\n",
      "Epoch 105/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3742 - accuracy: 0.8677 - val_loss: 0.3580 - val_accuracy: 0.8533\n",
      "Epoch 106/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3717 - accuracy: 0.8696 - val_loss: 0.3541 - val_accuracy: 0.8533\n",
      "Epoch 107/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3737 - accuracy: 0.8676 - val_loss: 0.3535 - val_accuracy: 0.8483\n",
      "Epoch 108/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3744 - accuracy: 0.8673 - val_loss: 0.3529 - val_accuracy: 0.8517\n",
      "Epoch 109/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3705 - accuracy: 0.8685 - val_loss: 0.3556 - val_accuracy: 0.8467\n",
      "Epoch 110/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3714 - accuracy: 0.8699 - val_loss: 0.3516 - val_accuracy: 0.8517\n",
      "Epoch 111/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3682 - accuracy: 0.8706 - val_loss: 0.3538 - val_accuracy: 0.8517\n",
      "Epoch 112/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3670 - accuracy: 0.8722 - val_loss: 0.3486 - val_accuracy: 0.8500\n",
      "Epoch 113/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3681 - accuracy: 0.8695 - val_loss: 0.3495 - val_accuracy: 0.8517\n",
      "Epoch 114/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3673 - accuracy: 0.8713 - val_loss: 0.3512 - val_accuracy: 0.8567\n",
      "Epoch 115/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3567 - accuracy: 0.8733 - val_loss: 0.3505 - val_accuracy: 0.8533\n",
      "Epoch 116/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3641 - accuracy: 0.8727 - val_loss: 0.3500 - val_accuracy: 0.8517\n",
      "Epoch 117/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3629 - accuracy: 0.8720 - val_loss: 0.3502 - val_accuracy: 0.8517\n",
      "Epoch 118/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3628 - accuracy: 0.8721 - val_loss: 0.3463 - val_accuracy: 0.8583\n",
      "Epoch 119/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3629 - accuracy: 0.8713 - val_loss: 0.3475 - val_accuracy: 0.8600\n",
      "Epoch 120/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3580 - accuracy: 0.8722 - val_loss: 0.3451 - val_accuracy: 0.8483\n",
      "Epoch 121/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3673 - accuracy: 0.8706 - val_loss: 0.3474 - val_accuracy: 0.8583\n",
      "Epoch 122/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3613 - accuracy: 0.8699 - val_loss: 0.3473 - val_accuracy: 0.8517\n",
      "Epoch 123/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3613 - accuracy: 0.8714 - val_loss: 0.3438 - val_accuracy: 0.8533\n",
      "Epoch 124/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3599 - accuracy: 0.8707 - val_loss: 0.3458 - val_accuracy: 0.8500\n",
      "Epoch 125/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3535 - accuracy: 0.8759 - val_loss: 0.3472 - val_accuracy: 0.8567\n",
      "Epoch 126/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3619 - accuracy: 0.8714 - val_loss: 0.3466 - val_accuracy: 0.8650\n",
      "Epoch 127/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3586 - accuracy: 0.8713 - val_loss: 0.3419 - val_accuracy: 0.8567\n",
      "Epoch 128/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3541 - accuracy: 0.8771 - val_loss: 0.3402 - val_accuracy: 0.8567\n",
      "Epoch 129/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3503 - accuracy: 0.8756 - val_loss: 0.3412 - val_accuracy: 0.8517\n",
      "Epoch 130/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3531 - accuracy: 0.8732 - val_loss: 0.3436 - val_accuracy: 0.8600\n",
      "Epoch 131/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3576 - accuracy: 0.8738 - val_loss: 0.3390 - val_accuracy: 0.8600\n",
      "Epoch 132/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3538 - accuracy: 0.8751 - val_loss: 0.3395 - val_accuracy: 0.8567\n",
      "Epoch 133/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3514 - accuracy: 0.8734 - val_loss: 0.3428 - val_accuracy: 0.8583\n",
      "Epoch 134/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3530 - accuracy: 0.8742 - val_loss: 0.3426 - val_accuracy: 0.8600\n",
      "Epoch 135/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3517 - accuracy: 0.8755 - val_loss: 0.3395 - val_accuracy: 0.8550\n",
      "Epoch 136/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3476 - accuracy: 0.8752 - val_loss: 0.3386 - val_accuracy: 0.8550\n",
      "Epoch 137/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3471 - accuracy: 0.8762 - val_loss: 0.3388 - val_accuracy: 0.8550\n",
      "Epoch 138/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3567 - accuracy: 0.8755 - val_loss: 0.3365 - val_accuracy: 0.8567\n",
      "Epoch 139/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3447 - accuracy: 0.8783 - val_loss: 0.3404 - val_accuracy: 0.8533\n",
      "Epoch 140/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3502 - accuracy: 0.8764 - val_loss: 0.3413 - val_accuracy: 0.8650\n",
      "Epoch 141/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3480 - accuracy: 0.8772 - val_loss: 0.3350 - val_accuracy: 0.8617\n",
      "Epoch 142/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3447 - accuracy: 0.8778 - val_loss: 0.3375 - val_accuracy: 0.8583\n",
      "Epoch 143/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3495 - accuracy: 0.8749 - val_loss: 0.3346 - val_accuracy: 0.8567\n",
      "Epoch 144/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3549 - accuracy: 0.8753 - val_loss: 0.3369 - val_accuracy: 0.8567\n",
      "Epoch 145/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3489 - accuracy: 0.8761 - val_loss: 0.3376 - val_accuracy: 0.8617\n",
      "Epoch 146/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3416 - accuracy: 0.8788 - val_loss: 0.3361 - val_accuracy: 0.8617\n",
      "Epoch 147/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3379 - accuracy: 0.8816 - val_loss: 0.3335 - val_accuracy: 0.8550\n",
      "Epoch 148/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3477 - accuracy: 0.8770 - val_loss: 0.3373 - val_accuracy: 0.8617\n",
      "Epoch 149/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3445 - accuracy: 0.8770 - val_loss: 0.3318 - val_accuracy: 0.8617\n",
      "Epoch 150/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3397 - accuracy: 0.8781 - val_loss: 0.3343 - val_accuracy: 0.8600\n",
      "Epoch 151/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3407 - accuracy: 0.8801 - val_loss: 0.3333 - val_accuracy: 0.8567\n",
      "Epoch 152/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3410 - accuracy: 0.8791 - val_loss: 0.3303 - val_accuracy: 0.8633\n",
      "Epoch 153/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3427 - accuracy: 0.8786 - val_loss: 0.3348 - val_accuracy: 0.8667\n",
      "Epoch 154/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3393 - accuracy: 0.8805 - val_loss: 0.3333 - val_accuracy: 0.8567\n",
      "Epoch 155/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3404 - accuracy: 0.8797 - val_loss: 0.3320 - val_accuracy: 0.8633\n",
      "Epoch 156/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3365 - accuracy: 0.8806 - val_loss: 0.3341 - val_accuracy: 0.8583\n",
      "Epoch 157/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3444 - accuracy: 0.8756 - val_loss: 0.3329 - val_accuracy: 0.8633\n",
      "Epoch 158/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3390 - accuracy: 0.8794 - val_loss: 0.3333 - val_accuracy: 0.8667\n",
      "Epoch 159/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3423 - accuracy: 0.8797 - val_loss: 0.3330 - val_accuracy: 0.8667\n",
      "Epoch 160/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3321 - accuracy: 0.8811 - val_loss: 0.3337 - val_accuracy: 0.8633\n",
      "Epoch 161/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3406 - accuracy: 0.8798 - val_loss: 0.3316 - val_accuracy: 0.8633\n",
      "Epoch 162/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3369 - accuracy: 0.8816 - val_loss: 0.3305 - val_accuracy: 0.8683\n",
      "Epoch 163/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3378 - accuracy: 0.8810 - val_loss: 0.3278 - val_accuracy: 0.8717\n",
      "Epoch 164/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3330 - accuracy: 0.8807 - val_loss: 0.3273 - val_accuracy: 0.8667\n",
      "Epoch 165/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3325 - accuracy: 0.8813 - val_loss: 0.3282 - val_accuracy: 0.8667\n",
      "Epoch 166/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3344 - accuracy: 0.8814 - val_loss: 0.3283 - val_accuracy: 0.8700\n",
      "Epoch 167/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3358 - accuracy: 0.8805 - val_loss: 0.3272 - val_accuracy: 0.8667\n",
      "Epoch 168/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3307 - accuracy: 0.8823 - val_loss: 0.3286 - val_accuracy: 0.8667\n",
      "Epoch 169/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3301 - accuracy: 0.8821 - val_loss: 0.3288 - val_accuracy: 0.8683\n",
      "Epoch 170/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3291 - accuracy: 0.8829 - val_loss: 0.3268 - val_accuracy: 0.8650\n",
      "Epoch 171/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3288 - accuracy: 0.8814 - val_loss: 0.3253 - val_accuracy: 0.8683\n",
      "Epoch 172/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3345 - accuracy: 0.8813 - val_loss: 0.3281 - val_accuracy: 0.8700\n",
      "Epoch 173/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3342 - accuracy: 0.8819 - val_loss: 0.3254 - val_accuracy: 0.8650\n",
      "Epoch 174/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3334 - accuracy: 0.8812 - val_loss: 0.3265 - val_accuracy: 0.8700\n",
      "Epoch 175/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3375 - accuracy: 0.8801 - val_loss: 0.3236 - val_accuracy: 0.8650\n",
      "Epoch 176/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3271 - accuracy: 0.8845 - val_loss: 0.3251 - val_accuracy: 0.8683\n",
      "Epoch 177/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3315 - accuracy: 0.8824 - val_loss: 0.3230 - val_accuracy: 0.8667\n",
      "Epoch 178/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3286 - accuracy: 0.8815 - val_loss: 0.3251 - val_accuracy: 0.8683\n",
      "Epoch 179/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3265 - accuracy: 0.8840 - val_loss: 0.3253 - val_accuracy: 0.8633\n",
      "Epoch 180/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3277 - accuracy: 0.8853 - val_loss: 0.3243 - val_accuracy: 0.8650\n",
      "Epoch 181/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3244 - accuracy: 0.8847 - val_loss: 0.3217 - val_accuracy: 0.8700\n",
      "Epoch 182/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3268 - accuracy: 0.8853 - val_loss: 0.3294 - val_accuracy: 0.8683\n",
      "Epoch 183/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3232 - accuracy: 0.8844 - val_loss: 0.3250 - val_accuracy: 0.8700\n",
      "Epoch 184/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3238 - accuracy: 0.8837 - val_loss: 0.3218 - val_accuracy: 0.8700\n",
      "Epoch 185/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3268 - accuracy: 0.8839 - val_loss: 0.3210 - val_accuracy: 0.8733\n",
      "Epoch 186/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3229 - accuracy: 0.8865 - val_loss: 0.3181 - val_accuracy: 0.8650\n",
      "Epoch 187/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3243 - accuracy: 0.8843 - val_loss: 0.3224 - val_accuracy: 0.8667\n",
      "Epoch 188/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3257 - accuracy: 0.8840 - val_loss: 0.3211 - val_accuracy: 0.8733\n",
      "Epoch 189/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3310 - accuracy: 0.8822 - val_loss: 0.3216 - val_accuracy: 0.8700\n",
      "Epoch 190/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3217 - accuracy: 0.8860 - val_loss: 0.3207 - val_accuracy: 0.8667\n",
      "Epoch 191/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3214 - accuracy: 0.8875 - val_loss: 0.3229 - val_accuracy: 0.8717\n",
      "Epoch 192/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3206 - accuracy: 0.8873 - val_loss: 0.3230 - val_accuracy: 0.8617\n",
      "Epoch 193/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3277 - accuracy: 0.8857 - val_loss: 0.3192 - val_accuracy: 0.8683\n",
      "Epoch 194/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3176 - accuracy: 0.8856 - val_loss: 0.3193 - val_accuracy: 0.8700\n",
      "Epoch 195/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3209 - accuracy: 0.8869 - val_loss: 0.3217 - val_accuracy: 0.8667\n",
      "Epoch 196/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3133 - accuracy: 0.8894 - val_loss: 0.3193 - val_accuracy: 0.8733\n",
      "Epoch 197/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3210 - accuracy: 0.8848 - val_loss: 0.3185 - val_accuracy: 0.8783\n",
      "Epoch 198/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3272 - accuracy: 0.8835 - val_loss: 0.3180 - val_accuracy: 0.8700\n",
      "Epoch 199/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3196 - accuracy: 0.8857 - val_loss: 0.3197 - val_accuracy: 0.8667\n",
      "Epoch 200/200\n",
      "117/117 [==============================] - 0s 3ms/step - loss: 0.3185 - accuracy: 0.8875 - val_loss: 0.3204 - val_accuracy: 0.8667\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential([\n",
    "    keras.layers.Flatten(input_shape=(28, 28)),\n",
    "    keras.layers.Dense(256, activation='relu'),\n",
    "    # 여기에 dropout layer를 추가해보았습니다. 나머지 layer는 위의 실습과 같습니다. \n",
    "    keras.layers.Dropout(0.5),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])\n",
    "\n",
    "model.compile(optimizer='adam',loss='sparse_categorical_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "history= model.fit(X_train, y_train, epochs=200, batch_size=512, validation_data=(X_valid, y_valid))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "starting-chinese",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# loss 값을 plot 해보겠습니다. \n",
    "y_vloss = history.history['val_loss']\n",
    "y_loss = history.history['loss']\n",
    "x_len = np.arange(len(y_loss))\n",
    "\n",
    "plt.plot(x_len, y_vloss, marker='.', c='red', label=\"Validation-set Loss\")\n",
    "plt.plot(x_len, y_loss, marker='.', c='blue', label=\"Train-set Loss\")\n",
    "plt.legend(loc='upper right')\n",
    "plt.grid()\n",
    "plt.ylim(0,1)\n",
    "plt.title('Loss graph with dropout layer') \n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('loss')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "revised-gasoline",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# accuracy 값을 plot 해보겠습니다. \n",
    "y_vacc = history.history['val_accuracy']\n",
    "y_acc = history.history['accuracy']\n",
    "x_len = np.arange(len(y_acc))\n",
    "\n",
    "plt.plot(x_len, y_vacc, marker='.', c='red', label=\"Validation-set accuracy\")\n",
    "plt.plot(x_len, y_acc, marker='.', c='blue', label=\"Train-set accuracy\")\n",
    "plt.legend(loc='lower right')\n",
    "plt.grid()\n",
    "plt.ylim(0.5,1) \n",
    "plt.title('Accuracy graph with dropout layer') \n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('accuracy')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "beautiful-symphony",
   "metadata": {},
   "source": [
    "## Batch Normalization"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "matched-roman",
   "metadata": {},
   "source": [
    "### 실습"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "julian-decline",
   "metadata": {},
   "source": [
    "아무것도 하지 않은 fully connected layer와 Batch Normalization layer를 추가한 두 실험을 비교하고자 합니다. 중점적으로 봐야 할 내용은 **정확도 비교와 속도의 차이**입니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "daily-spyware",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "=3\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "fashion_mnist = keras.datasets.fashion_mnist\n",
    "print('=3')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "theoretical-james",
   "metadata": {},
   "outputs": [],
   "source": [
    "(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()\n",
    "class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat',\n",
    "               'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']\n",
    "\n",
    "train_images = train_images / 255.0\n",
    "test_images = test_images / 255.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "another-northeast",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "21/21 [==============================] - 1s 24ms/step - loss: 1.7286 - accuracy: 0.4312 - val_loss: 0.7761 - val_accuracy: 0.7326\n",
      "Epoch 2/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.7125 - accuracy: 0.7537 - val_loss: 0.6095 - val_accuracy: 0.7976\n",
      "Epoch 3/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.5810 - accuracy: 0.8057 - val_loss: 0.5451 - val_accuracy: 0.8167\n",
      "Epoch 4/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.5270 - accuracy: 0.8230 - val_loss: 0.5105 - val_accuracy: 0.8260\n",
      "Epoch 5/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.4897 - accuracy: 0.8349 - val_loss: 0.4864 - val_accuracy: 0.8369\n",
      "Epoch 6/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.4682 - accuracy: 0.8403 - val_loss: 0.4719 - val_accuracy: 0.8397\n",
      "Epoch 7/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.4585 - accuracy: 0.8460 - val_loss: 0.4558 - val_accuracy: 0.8442\n",
      "Epoch 8/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.4333 - accuracy: 0.8545 - val_loss: 0.4466 - val_accuracy: 0.8506\n",
      "Epoch 9/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.4321 - accuracy: 0.8516 - val_loss: 0.4358 - val_accuracy: 0.8525\n",
      "Epoch 10/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.4155 - accuracy: 0.8592 - val_loss: 0.4298 - val_accuracy: 0.8521\n",
      "Epoch 11/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.4100 - accuracy: 0.8598 - val_loss: 0.4208 - val_accuracy: 0.8557\n",
      "Epoch 12/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3977 - accuracy: 0.8621 - val_loss: 0.4167 - val_accuracy: 0.8574\n",
      "Epoch 13/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3954 - accuracy: 0.8631 - val_loss: 0.4113 - val_accuracy: 0.8588\n",
      "Epoch 14/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3877 - accuracy: 0.8675 - val_loss: 0.4042 - val_accuracy: 0.8596\n",
      "Epoch 15/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3735 - accuracy: 0.8712 - val_loss: 0.4001 - val_accuracy: 0.8610\n",
      "Epoch 16/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3727 - accuracy: 0.8719 - val_loss: 0.3956 - val_accuracy: 0.8634\n",
      "Epoch 17/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3652 - accuracy: 0.8731 - val_loss: 0.3928 - val_accuracy: 0.8634\n",
      "Epoch 18/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3661 - accuracy: 0.8715 - val_loss: 0.3929 - val_accuracy: 0.8609\n",
      "Epoch 19/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3536 - accuracy: 0.8780 - val_loss: 0.3864 - val_accuracy: 0.8654\n",
      "Epoch 20/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3493 - accuracy: 0.8772 - val_loss: 0.3836 - val_accuracy: 0.8659\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_valid, y_train, y_valid = train_test_split(train_images, train_labels, test_size=0.3, random_state=101)\n",
    "\n",
    "model = keras.Sequential([\n",
    "    keras.layers.Flatten(input_shape=(28, 28)),\n",
    "    keras.layers.Dense(128, activation='relu'),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])\n",
    "\n",
    "model.compile(optimizer='adam',loss='sparse_categorical_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "history= model.fit(X_train, y_train, epochs=20, batch_size=2048, validation_data=(X_valid, y_valid))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "dated-blade",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# loss 값을 plot 해보겠습니다. \n",
    "y_vloss = history.history['val_loss']\n",
    "y_loss = history.history['loss']\n",
    "x_len = np.arange(len(y_loss))\n",
    "\n",
    "plt.plot(x_len, y_vloss, marker='.', c='red', label=\"Validation-set Loss\")\n",
    "plt.plot(x_len, y_loss, marker='.', c='blue', label=\"Train-set Loss\")\n",
    "plt.legend(loc='upper right')\n",
    "plt.grid()\n",
    "plt.ylim(0,1)\n",
    "plt.title('Loss graph without batch normalization') \n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('loss')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "smoking-prospect",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# accuracy 값을 plot 해보겠습니다. \n",
    "y_vacc = history.history['val_accuracy']\n",
    "y_acc = history.history['accuracy']\n",
    "x_len = np.arange(len(y_acc))\n",
    "\n",
    "plt.plot(x_len, y_vacc, marker='.', c='red', label=\"Validation-set accuracy\")\n",
    "plt.plot(x_len, y_acc, marker='.', c='blue', label=\"Train-set accuracy\")\n",
    "plt.legend(loc='lower right')\n",
    "plt.grid()\n",
    "plt.ylim(0.5,1)\n",
    "plt.title('Accuracy graph without batch normalization') \n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('accuracy')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "present-jumping",
   "metadata": {},
   "source": [
    "#### BatchNoramlization layer를 추가한 실습"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "classified-energy",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "21/21 [==============================] - 1s 17ms/step - loss: 1.3686 - accuracy: 0.5438 - val_loss: 1.0523 - val_accuracy: 0.6618\n",
      "Epoch 2/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.5358 - accuracy: 0.8156 - val_loss: 0.8783 - val_accuracy: 0.7222\n",
      "Epoch 3/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.4577 - accuracy: 0.8430 - val_loss: 0.7684 - val_accuracy: 0.7744\n",
      "Epoch 4/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.4112 - accuracy: 0.8590 - val_loss: 0.6781 - val_accuracy: 0.8162\n",
      "Epoch 5/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3843 - accuracy: 0.8680 - val_loss: 0.6145 - val_accuracy: 0.8416\n",
      "Epoch 6/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.3605 - accuracy: 0.8766 - val_loss: 0.5815 - val_accuracy: 0.8513\n",
      "Epoch 7/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.3440 - accuracy: 0.8820 - val_loss: 0.5418 - val_accuracy: 0.8596\n",
      "Epoch 8/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.3306 - accuracy: 0.8843 - val_loss: 0.5015 - val_accuracy: 0.8659\n",
      "Epoch 9/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.3185 - accuracy: 0.8900 - val_loss: 0.4837 - val_accuracy: 0.8641\n",
      "Epoch 10/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.3017 - accuracy: 0.8958 - val_loss: 0.4732 - val_accuracy: 0.8633\n",
      "Epoch 11/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.2919 - accuracy: 0.8988 - val_loss: 0.4403 - val_accuracy: 0.8692\n",
      "Epoch 12/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.2848 - accuracy: 0.8992 - val_loss: 0.4115 - val_accuracy: 0.8726\n",
      "Epoch 13/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.2705 - accuracy: 0.9047 - val_loss: 0.4152 - val_accuracy: 0.8616\n",
      "Epoch 14/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.2634 - accuracy: 0.9081 - val_loss: 0.3884 - val_accuracy: 0.8721\n",
      "Epoch 15/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.2575 - accuracy: 0.9108 - val_loss: 0.3748 - val_accuracy: 0.8779\n",
      "Epoch 16/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.2480 - accuracy: 0.9138 - val_loss: 0.3694 - val_accuracy: 0.8751\n",
      "Epoch 17/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.2407 - accuracy: 0.9144 - val_loss: 0.3637 - val_accuracy: 0.8761\n",
      "Epoch 18/20\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.2305 - accuracy: 0.9190 - val_loss: 0.3621 - val_accuracy: 0.8740\n",
      "Epoch 19/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.2286 - accuracy: 0.9204 - val_loss: 0.3461 - val_accuracy: 0.8798\n",
      "Epoch 20/20\n",
      "21/21 [==============================] - 0s 8ms/step - loss: 0.2192 - accuracy: 0.9250 - val_loss: 0.3579 - val_accuracy: 0.8737\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential([\n",
    "    keras.layers.Flatten(input_shape=(28, 28)),\n",
    "    keras.layers.Dense(128, activation='relu'),\n",
    "    #여기에 batchnormalization layer를 추가해보았습니다. 나머지 layer는 위의 실습과 같습니다.\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])\n",
    "\n",
    "model.compile(optimizer='adam',loss='sparse_categorical_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "history= model.fit(X_train, y_train, epochs=20, batch_size=2048, validation_data=(X_valid, y_valid))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "blocked-authority",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# loss 값을 plot 해보겠습니다. \n",
    "y_vloss = history.history['val_loss']\n",
    "y_loss = history.history['loss']\n",
    "x_len = np.arange(len(y_loss))\n",
    "\n",
    "plt.plot(x_len, y_vloss, marker='.', c='red', label=\"Validation-set Loss\")\n",
    "plt.plot(x_len, y_loss, marker='.', c='blue', label=\"Train-set Loss\")\n",
    "plt.legend(loc='upper right')\n",
    "plt.grid()\n",
    "plt.ylim(0,1)\n",
    "plt.title('Loss graph with batch normalization') \n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('loss')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "ideal-conclusion",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/Il7ecAAAACXBIWXMAAAsTAAALEwEAmpwYAABAfElEQVR4nO3deXhU1fnA8e+bsAQIsioiiyAqsgYI+6JBURERVNzQqoiAiijS2orUBa1bq3VrsWpRcKGiPxSKFIWABLCEQlB2UEAj+xa2REggyfn9ce4kk8lMMlnuTGDez/PMM3Pnbu/cmTnvvefce64YY1BKKRW5osIdgFJKqfDSRKCUUhFOE4FSSkU4TQRKKRXhNBEopVSE00SglFIRThOBOm2ISIKI7HRhuRtEJKGI8UkiMiLIZbkSY4B1pYpIv1Csy00iMlFEPnZeNxWRDBGJLud19BGRH8pzmWcSTQRh5BQwh0WkarhjiWTGmDbGmCQoWCiFmog0ExEjIpXCsf6KwBiz3RgTa4zJKctynO14oddylxpjWpY9wjOTJoIwEZFmQB/AAINCvO6wFDSRXMCdSfR7PPNoIgifu4DlwFTgbu8RItJERL4QkQMikiYif/caN1JENolIuohsFJFOzvsF9oBEZKqIPOe8ThCRnSLymIjsBaaISLSITBCRbc6yVjnrnSQif/WJZ7aIjPP3IUTkKhH5QUSOishbIrLYU40iIsNE5L8i8pqIpAETRaSFiHzjfK6DIjJNRGp7LS9VRB53PtthEZkiIjE+6/ydiOwXkT0ick+AuPqKyDqv4UQRWek1vFRErvdaZz8R6Q9MAG51qifWeC3yfOezpIvIfBGp72+9Xsuf4Hy+VBG5w+v9a0XkexE5JiI7RGSi12xLnOcjzvp7OPP4/c4dHURkrbP9P/XdVl7rHSYi34rIK852/VlErvEaf57zPR8Ska0iMtJr3EQRmSEiH4vIMWCYczT7nIgsc2L9UkTqOd/nMRFZ6ezseJbxhvN5jzm/tT4B4sw7KhKRHs6yPY9MEUl1pusqIskicsT5HfxdRKo44zzbcY0z363iU2UnIq2cz3BEbNXgIK9xU53/wX+cbf4/EWnhL94zhjFGH2F4AFuB0UA8cApo4LwfDawBXgNqADFAb2fczcAuoAsgwIXA+c44A1zotfypwHPO6wQgG/gzUBWoBvweWAe0dJYVB9QDugK7gShn3vrAcU98Pp+hPnAMuBGoBIx1PssIZ/wwZ70POeOrOTFf6cRxNrbwe91rmanAeqAJUBf4r5/P8SxQGRjgxFbHT2zVgEwnxsrAPmfb1XTGnQDqea2zn/N6IvCxz7KSgG3Axc68ScBLAb5XT4yvOp/xMuBXoKXX+HbYnbD2TlzXO+OaOd9jJa/lFfWdpwIrgPOcbbUJuD9AXMOc72Yk9jf2gPM9izN+CfAW9vfWATgAXO61TU4B1ztxe7bBVqAFUAvYCPwI9HO+6w+BKV7r/w3291UJ+B2wF4jx3eb+toHzfmVgMfCiMxwPdHeW18z57I94Te/7f0gAdnotays26VcBLgfSvb6jqUAa9r9QCZgGTA93meFqeRTuACLxAfR2/lj1neHNwDjndQ/nT1jJz3zzgLEBlllcIjjp+eM57/0ADA6wrE3Alc7rMcDcANPdBSR7DQuwg4KJYHsx2+J64Huv4VS8CjNsYb/N63OcoGBBuR/oHmDZS7FJqjswH/gM6A/0Bdb6rLO4RPCE1/Bo4OsA60zAJoIaXu99BjwZYPrXgdec180onAiK+s5Tgd94Df8FeDvAtMOArV7D1Z11nYtNujlATa/xLwJTvbbJEj/b5I9ew38FvvIavg5YXcT3fhiI893m/raB8/4/gDk4Oyh+lvcIMLOI/0MC+YmgDzYRRXmN/wSY6PXfmezzG9xc1O/4dH9o1VB43A3MN8YcdIb/RX71UBPgF2NMtp/5mmD3TEvjgDEmM8hlfYDdg8N5/ijAdOdhC34AjP3X+J4xs8N7QEQaiMh0EdnlVDN8jN1rDzTPL856PNJ8ts1xIDZAfIuxBcClzusk7B76Zc5wSewNcp0Ah40xv3oN530GEekmIovEVvsdBe6n8Of3Vtx3XpK48qY1xhx3XsY6sR0yxqT7xNzIa7jA9+jY5/X6hJ/hvFhE5FGneuuoiBzBHkUUWb3mNe992O/xdmNMrvPexSIyR0T2Or+jF4JdHs7v1rMsh+/nLcl2Pe1pIggxEakG3AJc5vyI9wLjgDgRicP+4ZqK/wa5HdhDcX+OY/fyPM71Ge/bzWxRy/oYGOzE0wqYFWC6PUBjz4CIiPdwgPW+4LzXzhhzFjbRiM80TbxeN8VWYZSGbyJYTPGJoDy6460jIjW8hr0/w7+A2UATY0wt4G3yP7+/dRf1PZWX3UBdEanp9V5TbJWUR6m3i9Me8Afs776OMaY2cJTC33ugef+EPXo95jXqH9gj6Yuc39GEYJbn2A00ERHv8s/380YUTQShdz32MLw1ti62A7awXYqtalmBLWBfEpEaIhIjIr2ceScDj4pIvFgXisj5zrjVwO1iG4H7Ywu7okwG/iQiFznLai8i9QCMMTuBldgjgc+NMScCLOM/QDsRud5JXA9SOAH5qglkAEdFpBG2rcLXgyLSWETqAn8EPi1mmYEsw7aBdAVWGGM2AOcD3chvmPW1D2jmU0iUxjMiUsUpyAYC/+e8XxO7950pIl2B273mOQDkAhd4vVfUd14ujDE7sNvqRef31h64F7tDUB5qYqvLDgCVROQp4KziZhKRJthqtbuMMT/6WeYxIENELsG2eXjbR8Ht6O1/2B2nP4hIZbHXkFwHTA/q05yBNBGE3t3YRrTtxpi9ngfwd+AO7F7NddhGwe3YqpZbAYwx/wc8j92rTMfuqdd1ljvWme+Is5xZxcTxKvZPNh/7h3oP2wjo8QG2UTNQtRBO1dbN2LrpNGxySwGyiljvM0An7B7hf4Av/EzzLyeun7DVIs8V81kCxfcr8B2wwRhz0nk7GVv1tj/AbJ4CO01EvivNerHVCoexe57TsG0em51xo4FnRSQdeAr7HXjiPY79fv/rnM3SvZjvvDwNxdbP7wZmAk8bYxaU07LnAV9jG5N/wTbi+6tq8nUF0ACY4XXm0AZn3KPYJJoO/JPCOwsTgQ+c7XiL9wjnt3AdcA1wENtIfpfXdxRxPGcMKFWAiFyK3SM83wT5I3H2oncCdxhjFpVyvanYxubyKoSUUsXQIwJViIhUxh5hTC4uCYjI1SJSW+zV0Z562uUhCFMpVU5cSwQi8r7Yi37WBxgvIvKm2ItX1krBi2RUmIhIK2z1UkPsqY3F6YGtvjmIPdy+vog2BaVUBeRa1ZBTtZABfGiMaetn/ADshUYDsI13bxhjurkSjFJKqYBcOyIwxiwBDhUxyWBskjDGmOVAbRFp6FY8Siml/Atn51GNKHjmwE7nvT2+E4rIKGAUQLVq1eKbNGniO0lQcnNziYqquM0iGl/ZaHxlV9Fj1PhK78cffzxojDnb70g3L1vGno62PsC4OTh96DjDC4HOxS0zPj7elNaiRYtKPW8oaHxlo/GVXUWPUeMrPSDFVMAuJnZR8ArSxkTwlX1KKRUu4UwEs4G7nLOHugNHjTGFqoWUUkq5y7U2AhH5BNvPS32nH/Cnsd2/Yox5G5iLPWNoK/Zyb7/9yiullHKXa4nAGDO0mPEG2zeNUkqpMKqYzdtKKaVCRhOBUkpFOE0ESikV4TQRKKVUhNNEoJRSEU4TgVJKRThNBEopFeE0ESilVITTRKCUUhFOE4FSSkU4TQRKKRXhNBEopVSE00SglFIRThOBUkpFOE0ESikV4TQRKKVUhNNEoJRSEU4TgVJKRThNBEopFeE0ESilVITTRKCUUhFOE4FSSkU4TQRKKRXhNBEopVSE00SglFIRThOBUkqdBpKT4cUX7XN5q1T+i1RKKVUefv0V9u6FxEQYOxays6FqVVi4EHr0KL/1aCJQSqkQSE6GpCS49FK46CLYs8cW8oGe9+6F9PTCyzl50i5HE4FSSoWYpyBPSIBu3WwhfeRI/uPoUUhObsDatQXfP3IEtm+H774DYwIvv2ZNaNgQzj0XOnWyz57hw4dh/Hh7RFClio2hPGkiUEpFBO+C3LM3ffKkLWTT0uDQofxn79dpafDzz8UX5FarvFexsVC7tn2kp+fPKwIDB8Jdd+UX9OeeCzVqFL3kbt0Kx19eNBEopU4LyckwbVpTqla1BaExkJlp98SPHSv6ecsW+M9/ICfHFsQNGtj6d39VLx7R0VC3LtSrB8ePFyzIExLg2mvzC3rPY/Pm5fTv351ataCSV+manAxXXGETT5Uq8PjjJS/Me/Qo/wTgoYlAKRUS/vbIPU6ehAMH4OBB++z7evNmWLIEcnOb8957thrlxAk4dar49Vavbgv1nBw7bIzdE7/ssvyCvm7dwq/POssW+p7YvQvy55/3Xyinp2dSr17h93v0sA28bu3Rl5UmAqWUK06dgl27YOdOWwg+/7yt446Kgu7d7WtPgX/smP9liNjCWQRycwFsydyqFVx+uS2sa9UK/FyzJlSuXLggnzSpZIVxeRTkbu7Rl5UmAqVUULz36Dt3ht27YccOW9D7e963z3+dek4O/PQTtG0LLVrA2WdD/fr22fd1nTp2b95TkGdl5VK1ahSvvaYFeXnSRKBUhCiqaiY319al79xZjeXL7V56Wlr+84YNMGeOZ6/cv7POgsaNoUkTaN/ePnuGDx6EESPy98g//7x0Bfn776cyfPgFWpCXM1cTgYj0B94AooHJxpiXfMafD7wPnA0cAn5jjNnpZkxKnemMgYyM/Pr1AwdsEnj55fyqmZ497XSegj4tzVPIdyu0vEqVICYmPwmIQN++cPvt+QV948Y2ERSlefOy75FnZW2nR48LSj6zKpJriUBEooFJwJXATmCliMw2xmz0muwV4ENjzAcicjnwInCnWzEpFS6+Z7wEKyfHnt3y66+weLEtSJs1s/Xm3gW97yMrq+hl/vgjtG5tq2fq17fLq18f9u/fRJ8+rfKG69e39ezLlxesY3/uuYp11osqGzePCLoCW40xPwGIyHRgMOCdCFoDv3VeLwJmuRiPUiGXkQH//jcMHw6nTjXngw/gzjtt3XdGhi3gMzIKPrzfO3Gi6OXXqJFfn96woa2S8Qx7P3bssOetewrymTP9F8pJSftISGhV6P2KftaLKhsxxV8hUboFi9wE9DfGjHCG7wS6GWPGeE3zL+B/xpg3RORG4HOgvjEmzWdZo4BRAA0aNIifPn16qWLKyMggNja2VPOGgsZXNm7Gt2HDWaxeXZsOHY7Qpk3+KS4ZGdHs2xfD3r0xhZ737o3h2LHKfpcXE5NDtWr24f3a/3Au69adxbJl9TFGiIoy3HLLdoYN+4WqVYuotA/yM3iL5O+4PFTk+Pr27bvKGNPZ70hjjCsP4CZsu4Bn+E7g7z7TnAd8AXyPbUvYCdQuarnx8fGmtBYtWlTqeUNB4yub8owvJ8eY/fuNWb/emDffNKZKFWNEjKlUyZg+fYzp0MGY2rWNsTXt+Y9q1Yy55BJj+vc35r77jHnxRWOeecaYqlWNiYrKNdWqGfPttyWPZ9kyu+zoaPu8bFm5fdQCIuk7dkNFjg9IMQHKVTerhnYBTbyGGzvveSeh3cCNACISCwwxxhxxMSYVoZKTbdVGu3a2cXPfPvvYv9//64MH8y9A8padDT/8AF262AbXZs3s4/zz7fPZZ+dfhOTtyivh/fd/LtMZL1o1o9ziZiJYCVwkIs2xCeA24HbvCUSkPnDIGJMLPI49g0ipUsnNteew//ST7RvG87x6NaxbF3i+atVslwMNGtjCvGvX/OFzzrFJ4dFH7QVSVarArFmlaygt6xkv2tiq3OJaIjDGZIvIGGAe9vTR940xG0TkWewhymwgAXhRRAywBHjQrXjU6c1zDnynTnav27ew/+knSE29tECXAyL2tMbo6Pz3oqLsaY+jR9tCvkED2zlYcTp10r1xdeZy9ToCY8xcYK7Pe095vZ4BzHAzBnV6yc62e/XbttnCfds2WLHCFsL+zmuoW9eenx4XB/HxO7nssqZccIF9r2lTexMP3+4FRo/WUx+V8qZXFquQ8L6qtV27/ELeu8C3e/U2GXhUrmwvVPLu+fH2221VTfPmtj8Zj6Skn0hIaFpo3Vq/rlTRNBEoV2RmwtattmE1MREmT/bf+Ar2nPoWLWz1y803wwUX2OELLrBVOytWFNyjf/BB6NChZPHoHn0FUFQfF2c6Y2DePFi1yvaWV8E+vyYCFRR/V8YaY2+r98MPhR+pqf77pRGBq66yF1h5Cvs6dYpet+7Rn+ZOnbKdC919d/4ttmbMgKuvLthpv9tCkIgkO9v2me392LQJ1q+3NzUA+yfo18/es7J1a/to0cIe/oaJJgJVrP/+1/5us7KaM3Wq7Wfm0CHbTYH3jT2qV4eLL7anVv7mN9CypX0cPgyDBuXv0T/9tNbRn5HS0+1ewKZN+QXgpk320NC7vi8z096iS8TuBXj6sijusW0bzaZNs/d+bNvWLicryz78vfZ+78cfYepUe1haqRL88Y8QH2/rFj2Ps86yD++zC7x5J5LWrf0W+H22bi146NuoEVxyib3k+3//y7/kZOVKe6jsUbmy/fN4EoPncdFFtqHLd/3l/GfQRKAKyMmx/+VVq+yt+Vatsr/fkycBhOxsW1XTrRsMG5Zf2LdsaX/zUVH+l6t79OUg3FUrycmwaJFt5ImNLVjgb95sW/k9oqPhwgvtjQNuuMEWvn/5iz06qFQJxoyxnRgdPJj/+OUXSEmxr+0PrpBmAB9/XLbPceoUTJwYeHxsbP4NDTyPU6dsZ0/+6jcrV7YFdps27OjcmfOvvtoW/i1b5vfE53vGwty5Njls3gwbN+Y/vvvOHi15GsWio+2yGzSwe2S5uTYxLFxYrr8BTQQR7NQp+x/2FPrffWfPufccwVarZuvir7sOZs+GnBxD1arCV1/pHn2JJSfTdNo0CvU6Z4w9ZPJcyeb92Ls3//X27faKN7AFz2efwfXXux93To7N/O++Cx9+WLi+LzbWFnp9+9rnVq3sc4sWtsDzds01wSUyT/ep3kliypT8AjIqCoYMsQ1KVavaR0xM0a+//95WRZ08abfflCm2XtL7fpaBXh85Ys9k8CQBEbus0aPtZ23ePK+K6+ekJM73d2f5QPWb8fH24e3ECbs35p0gli7NP6o6edIuRxOBKqmlS+HTT+1O2JEjttBfu9YePYPtvKxjRxg50jbaxsfbHRpPFW5yctmujI1Y6em20B49muYnT9oCqGtX+2f2XM7sb+83Ojr/QocGDWwBfOCALQhPnbJ72R062CqWgQNtfVygw7GS2r/fNmzOnQvz59t6QJH8vdSoKHs4+Mwz9jDQ36XU/gS7NyBif6g1a9pCFuwNgefMITcri6iqVWHcuJIVhH36lO2w1HeP/qmn3Nsb8uyBeZ8R4bt+f8mmDDQRnKFycmDNGnsk/8UXsGxZ/rjq1W1ZNHq0LfA7dbJHn4GqRqEC9AW/bJktlK64wv6pS8rtapUDB/LrxDduzH/tVV0iYL+YX36xFz60a5df0DdoAOeem/+6bt2CBbt3QVC5sm1tX7cOXnjB9gl9zjn2burXXWf7syhJx2c5ObbO+quv6PTZZ3Zv1Bi7zIEDYcAAWxDfcEN+QTRihD2lK1ScPerU99/nguHDQ3+LsXCfseDy+jURnCFyc+2JCYsW2cfixXbPH2w7m2eHLjratpNNmBDWcIvmuSXW+vX2edkyW2cFdi/Uc3FBbKzda4yNhdhY2mRl2T4ivN+vWdPueb/6qj20rlwZ/vpXe/hTpYodrlIl/+FvWCS/frx1a5tJfQv8gwfz469RI7+6pFUrO/8zz5B78qTdm50xo3R7k/4KgrQ0+Ppr+PJLm/GnTLFx9+1rC/HrrrMdIfk6eDB/r3/ePLscEUzr1nYbDxhgt5F3Mgp3Q0+PHmzPyuKCcB2Shrt+08X1ayI4TRljyx9PwZ+UZP/LYKs+hwyxZUFCgq1e9j6q7Ns3DAH72yM/dswWpuvX5xf669fbunEPT2OdJ5OJ2PnbtbP1yOnpec/V9u2ze+Ce9zz1Xt6ysmxDZUlER/tvJKxbN78xtFUr+2jd2u4p+1bTXHZZ2fZmwX9BUK8e3HGHfZw6ZRsUv/zS3lfyoYfso21bmxDq1rVJY/du20hpjO2vY8AAW39/1VV8v24dCYGqHcJdECrXaCI4TSQn2x3JSpVszUJSkt3RBduVwsCBtoDv29cOe2vUKMw7c8uW5WeiqChbn71rl81QHtWr20K0f39bcLVpY58bNSp8e6yXXvL7IVKSkgoWYtnZNiEsXgy33ZZ/xspf/2pP1Tt5suDj1Cn/7yUl2WV4GiqHD4fnnw/c1ag/odibrVzZfsEJCfYz/vijTQhffgl//nN+Q6/nM9x3n60bLK+2BXXa0kRQwR06ZKuAX389v62ufn170oKn4G/evPjyKKQ7c7/+auucly2zj2++yd87z821vcRdfnnBAr9Zs8AFUmnrRytVsnXbgwfbGEqbCa+6qmAiGj7c1p9XdBdfDL/9rX089ZRNXrm59sfSooVNyEqhiaBCys215dZ779lbCnrffzY62p4wUWHq+I2x90H0FPqe+nxPVUqrVrYQnTfPfrAqVWxddqjPPz2dGwrLwzXXwCuvuHbWiTq9aSKoQPbtq8ozz9j2vl9+sRddjhoFnTvD/feHv46/6bRpdq89JqZgwb/Lud9Q9er2SrPx4+1dW7p3t/XSzvyndUF6utePnwnJTLlGE0GYZWXZG5289x4sWNAdsN05vPSSvV4oJsZOd9FFYfoPZ2XZ4MaOpXl2tu09zuP8821/KT172kf79oH7jjndC9IzgX4HKgBNBGGydq0tXz/+2LYDNG0Kd931CxMnNqNZs8LTh+w/nJtrq3YWLoQFC+yVaCdOAM558CIwdCi8/DKcd14IAlJKuU0TQYgkJ8NXX9kyddEi261DlSp2r//ee201+tKlqTTzlwXcZAxs2WIL/oULbXCHDtlxrVvbC4caNbLnwXuu6hwzRpOAUmcQTQQhMG+ePb3T01VIixbwxhv21O969UIUhHcdfbNm+QX/woW2sRfsXd0HDbJ1U5dfDg0b5s9/6aVlPw9eKVUhaSJw0alT8I9/2LZTTxKIjrZnHz78cAgDSU62Bbvn9CPPeah169r3J0ywhyQXXhj4PNRwX9WplHKNJgIXGAP//jf84Q+21qVzZ9stjOeeHCE968cYe9qg91W2/fvbc8o7dNCLiZRSaClQzlatsgX9DTfYvf85c2wvvosWwZ/+VO7diBdt/34byBdf2AI/Otr2bPjUU7anOU0CSin0iKDc7Nxpa1g++she+Ttpku3S2XP3uZCfuffvf9sAjh2zHa517QpLlug55EqpQjQRlFF6ur3x0l//as+8fOwxePxx209aWBw7BmPH2tvydexoM1ObNnZcr15hCkopVZFpIiilnBx4/3148knb+dttt8GLL+L3GoCQWbzY3iB8xw7b1/RTTxW+S5RSSvnQSuJSmD/f7myPGmVPBV2+HD75JIxJIDMTHn3UNk5Urgzffmt7qtMkoJQKgh4RBCk52d7qcflyezP35s3h//7P9vsfbE/Ervj+e7jzTtuX/wMP2Ct+a9QIY0BKqdONJoIgJCfbNlbPrWXHjLFnZFatGsagsrNt48TEibZ1+quv7GmhSilVQpoIgpCUlJ8EoqNt7wphTQJbt8Jdd9kMdcst8NZbIbxEWSl1ptFEEIRLL7XPImHsyt1zz9wjR+y5qVWqwLRptgO4sNZNKaVOd5oIglC9un2+6SZ7U5iQn4bv6SLCc3Vwly72IrHGjUMciFLqTKSJIAiJifb5jTcK9sMWMvPn5yeBqCjbZakmAaVUOdHTR4Mwfz60axemJJCdnZ+JoqJs40RYblGmlDpT6RFBMU6csKflP/hgGFZujO2m9L//tTcgr19fu4hQSpU7TQTFWLrU9t585ZVhWPmf/2z7sX7sMXvvSqWUcoGrVUMi0l9EfhCRrSIy3s/4piKySES+F5G1IjLAzXhKIzHRnqDjOXMoZKZNs50WDR0KL7wQ4pUrpSKJa4lARKKBScA1QGtgqIi09pnsCeAzY0xH4DbgLbfiKa3586F37/wzh0Lim2/gnntsNdCUKdpdtFLKVW6WMF2BrcaYn4wxJ4HpwGCfaQxwlvO6FrDbxXhKbN8+e5P5kFYLrVtn7yFw8cUwc2aYr1xTSkUCMZ7bFpb3gkVuAvobY0Y4w3cC3YwxY7ymaQjMB+oANYB+xphVfpY1ChgF0KBBg/jp06eXKqaMjAxiY2ODnj4x8RxeeKE1b7+dQsuWGaVaZ0lkp6bS5/e/B+C7SZPIOucc19dZEiXdfqGm8ZVdRY9R4yu9vn37rjLGdPY70hjjygO4CZjsNXwn8HefaX4L/M553QPYCEQVtdz4+HhTWosWLSrR9HffbUy9esbk5JR6lcE7csSkN29uTM2axqxZE4IVllxJt1+oaXxlV9Fj1PhKD0gxAcrVoKqGROQLEblWREpSlbQLaOI13Nh5z9u9wGdOQkoGYoD6JViHa4yxDcX9+oWgiv7kSbjxRqpv326vGG7f3uUVKqVUvmCLuLeA24EtIvKSiLQMYp6VwEUi0lxEqmAbg2f7TLMduAJARFphE8GBIGNy1caNsHt3CNoHjIHhw+Gbb/jhD3+wmUcppUIoqERgjFlgjLkD6ASkAgtEZJmI3CMilQPMkw2MAeYBm7BnB20QkWdFZJAz2e+AkSKyBvgEGOYcwoSd52Je1xPBH/9oTxV9/nn2XXWVyytTSqnCgr6gTETqAb/B1vV/D0wDegN3Awn+5jHGzAXm+rz3lNfrjUCFvJFuYqI9cadpUxdX8vbb9v6Wo0bZawYWL3ZxZUop5V9QiUBEZgItgY+A64wxe5xRn4pIilvBhUtWlr0HwfDhLq7kyy9tvxXXXmu7ldaupJVSYRLsEcGbxphF/kaYQKcjncaSk+H4cRerhVasgFtvhU6d7P0vK2lPH0qp8Am2sbi1iNT2DIhIHREZ7U5I4ZeYaO9E5soNaLZtg4ED4dxzYc4cvb+wUirsgk0EI40xRzwDxpjDwEhXIqoAEhOhe3c466zipw1acjI88QRcdhnk5sLXX0ODBuW4AqWUKp1g6ySiRUQ8Z/Q4/QhVcS+s8Dl0CFJS4Omny3GhyclwxRW2T2uAd96xLdFKKVUBBJsIvsY2DL/jDN/nvHfGWbjQntpfrmdyJiUVvMNYWlo5Llwppcom2ETwGLbwf8AZTgQmuxJRmCUmQq1a9rbA5UbEZhcR24mcK40PSilVOkElAmNMLvAP53HG8nQr0bdvOZ7Is3s3vPIKtGwJv/mNrSLSO4wppSqQYK8juAh4EXtfgRjP+8aYC1yKKyy2bYPUVHA6AC273Fx7X4Hjx2HWLLjkknJasFJKlZ9gzxqagj0ayAb6Ah8CH7sVVLh4upUot/aBv/3N3tnm1Vc1CSilKqxgE0E1Y8xC7P0LfjHGTASudS+s8Jg/H5o1gxYtymFh69fbew1fdx3cd185LFAppdwRbE14ltMF9RYRGYPtTrpi3n2hlLKz7R0ib721HHp7yMqCO+6wrc6TJ2v3EUqpCi3YI4KxQHXgYSAe2/nc3W4FFQ4rV8KxY+XUrcSECfYel1OmQAW7y5hSSvkq9ojAuXjsVmPMo0AGcI/rUYVBYqLdcb/88jIuaMEC2yYwejQMGFAusSmllJuKPSIwxuRgu5s+o82fD507Q716ZVhIWhrcfbdtGH755XKLTSml3BRsG8H3IjIb+D/gV8+bxpgvXIkqxI4dg+XLbdtuqRljG4UPHLBdTFevXm7xKaWUm4JNBDFAGuBdcWKAMyIRJCVBTk4Z2wc++AA+/xxeesl2L62UUqeJYK8sPiPbBTwSE+0OfKkv+N22DR56yPYs+uij5RqbUkq5Ldgri6dgjwAKMMa4eQ+vkElMtN3/VK1aipmzs+HOO+0NDD780D4rpdRpJNiqoTler2OAG4Dd5R9O6G3fDj/8APffX8oFvPCC7Wb6k09cvsGxUkq5I9iqoc+9h0XkE+BbVyIKMU+3EqVqH1i+HJ591l48dttt5RqXUkqFSrAXlPm6CDgjrpRKTITzzoPWrUs4Y3q67U20cWN783mllDpNBdtGkE7BNoK92HsUnNZyc+31XwMHlqIXiEcegZ9/tqcc1arlQnRKKRUawVYN1XQ7kHD4/nt7DViJq4W++ALef992JdGnjyuxKaVUqARVNSQiN4hILa/h2iJyvWtRhYinfaBfvxLMtHs3jBwJ8fHlfGNjpZQKj2DbCJ42xhz1DBhjjgCnfSmYmAjt20ODBkHOkJsLw4bZ+w9PmwZVqrgZnlJKhUSwicDfdOV1M8ewOH4cvv22BNVCyckwaJDNHq++am89qZRSZ4BgC/MUEXkV8Jwe8yCwyp2QQmPJEjh5Msi7kSUn225JMzMhKgratXM9PqWUCpVgjwgeAk4CnwLTgUxsMjhtJSbaK4mDautdtMgmAbCnFy1e7GpsSikVSsGeNfQrMN7lWEIqMRF694Zq1YKYOCPDPkdF2XaBhAQ3Q1NKqZAK9qyhRBGp7TVcR0TmuRaVy/buhXXrgmwfOHgQ3nkH4uLgT3+ChQvL0DudUkpVPMG2EdR3zhQCwBhzWERO2yuLFyywz0G1Dzz+OBw9aquD2rZ1NS6llAqHYNsIckUkr0c1EWmGn95ITxfz58PZZ9ud/CIlJ9ubz48bp0lAKXXGCvaI4I/AtyKyGBCgDzDKtahcZIw9IrjiClvlH1B2NjzwADRqpBeOKaXOaME2Fn8tIp2xhf/3wCzghItxuWbDBtizJ4j2gUmTYM0amDEDYmNDEptSSoVDsJ3OjQDGAo2B1UB3IJmCt670N19/4A0gGphsjHnJZ/xrQF9nsDpwjjGmdvDhl1xQ3U7v3g1PPgn9+8ONN7oZjlJKhV2wbQRjgS7AL8aYvkBH4EhRM4hINPYCtGuA1sBQESnQ2bMxZpwxpoMxpgPwN0JwD+T58+GSS6BJkyIm+t3v7NVmf/tbKbolVUqp00uwiSDTGJMJICJVjTGbgeL6WOgKbDXG/GSMOYm9EG1wEdMPBT4JMp5SOXlSWLy4mKOBBQtg+nR7ttCFF7oZjlJKVQjBNhbvdK4jmAUkishh4Jdi5mkE7PBeBtDN34Qicj7QHPgmwPhROI3TDRo0ICkpKciwC0pJqcqJE3DuuetISkorvJ6TJ+kyYgRy3nms7NGD3FKup7QyMjJK/dlCQeMrm4oeH1T8GDU+lxhjSvQALgMGAVWKme4mbLuAZ/hO4O8Bpn0M+Fsw64+PjzeldfvtqaZSJWOOHQswwXPPGQPGfPVVqddRFosWLQrLeoOl8ZVNRY/PmIofo8ZXekCKCVCulrgHUWNMsB3t7AK8a+IbO+/5cxsh6LsoJaUuPXpATX+32fn5Z3juObjpJttIrJRSEaK09ywOxkrgIhFpLiJVsIX9bN+JROQSoA72LCTXfP01/PhjrP/eo42Bhx6C6Gh47TU3w1BKqQrHtURgjMkGxgDzgE3AZ8aYDSLyrIgM8pr0NmC6c+jiiuRkGDwYQPjoIztcwOzZ8J//wDPP2JvRK6VUBHH15jLGmLnAXJ/3nvIZnuhmDGDvL3/qlH2dnW2H8/qN+/VXePhh24XEww+7HYpSSlU4p/VdxoKVkAAxMZCVlUuVKlEFe5F+7jnYvh2WLoXKlcMUoVJKhY+bbQQVRo8etvfo4cNTC/YivXEjvPKKvQ9x797hDFEppcImIo4IwBb+WVnb6dHjAvuGMfDgg/YUor/8JbzBKaVUGEVMIijkX/+yjQVvv237pFZKqQgVEVVDhRw5YvsT6toVRowIdzRKKRVWkXlE8OSTcOCAPWU0Ojrc0SilVFhF3hHBqlXw1lswejTEx4c7GqWUCrvISgQ5OfauY2efbW9Er5RSKoKqhpKTafP007ByJXz8MdSuHe6IlFKqQoiMRJCcDJdfTv3MTHuj4ubNwx2RUkpVGJFRNZSUBFlZCNg7ji0OtgNVpZQ680VGInD6mDBRUVClCgX7mFBKqcgWGYnA6WPi5+HDKdjHhFJKqchoIwDo0YPtWVlcoElAKaUKiIwjAqWUUgFpIlBKqQiniUAppSKcJgKllIpwmgiUUirCaSJQSqkIp4lAKaUinCYCpZSKcJoIlFIqwmkiUEqpCKeJQCmlIpwmAqWUinCaCJRSKsJpIlBKqQiniUAppSKcJgKllIpwmgiUUirCaSJQSqkIp4lAKaUinCYCpZSKcK4mAhHpLyI/iMhWERkfYJpbRGSjiGwQkX+5GY9SSqnCKrm1YBGJBiYBVwI7gZUiMtsYs9FrmouAx4FexpjDInKOW/EopZTyz80jgq7AVmPMT8aYk8B0YLDPNCOBScaYwwDGmP0uxqOUUsoPMca4s2CRm4D+xpgRzvCdQDdjzBivaWYBPwK9gGhgojHmaz/LGgWMAmjQoEH89OnTSxVTRkYGsbGxpZo3FDS+stH4yq6ix6jxlV7fvn1XGWM6+x1pjHHlAdwETPYavhP4u880c4CZQGWgObADqF3UcuPj401pLVq0qNTzhoLGVzYaX9lV9Bg1vtIDUkyActXNqqFdQBOv4cbOe952ArONMaeMMT9jjw4ucjEmpZRSPtxMBCuBi0SkuYhUAW4DZvtMMwtIABCR+sDFwE8uxqSUUsqHa4nAGJMNjAHmAZuAz4wxG0TkWREZ5Ew2D0gTkY3AIuD3xpg0t2JSSilVmGunjwIYY+YCc33ee8rrtQF+6zyUUkqFgauJQClV2KlTp9i5cyeZmZnhDqWQWrVqsWnTpnCHEZDGV7yYmBgaN25M5cqVg55HE4FSIbZz505q1qxJs2bNEJFwh1NAeno6NWvWDHcYAWl8RTPGkJaWxs6dO2nevHnQ82lfQ0qFWGZmJvXq1atwSUCd/kSEevXqlfhoUxOBUmGgSUC5pTS/LU0ESikV4TQRKBVh+vbty7x58wq89/rrr/PAAw8EnCchIYGUlBQABgwYwJEjRwpNM3HiRF555ZUi1z1r1iw2bszrd5KnnnqKBQsWlCD68pGUlMSyZctCvt6KShOBUqeD5GR48UX7XEZDhw7Ft7+u6dOnM3To0KDmnzt3LrVr1y7Vun0TwbPPPku/fv1KtayyqCiJICcnJ9whAJoIlAqvRx6BhISiHx07Qu/eMGGCfe7YsejpH3mkyFXedNNN/Oc//+HkyZMApKamsnv3bvr06cO4cePo3Lkzbdq04emnn/Y7f7NmzTh48CAAzz//PBdffDG9e/fmhx9+yJvmn//8J126dCEuLo4hQ4Zw/Phxli1bxuzZs/n9739Phw4d2LZtG8OGDWPGjBkALFy4kI4dO9KuXTuGDx9OVlZW3vqefvppOnXqRPfu3dm8ebPfuMaPH0/r1q1p3749jz76KAAHDhxgyJAhdOnShS5duvDf//6X1NRU3n77bV577TU6dOjA0qVLCyxnxYoV9OjRg44dO9KzZ8+8z5WTk8Ojjz5K27Ztad++PX/7298AWLlyJT179iQuLo6EhATS09OZOnUqY8bk9a/JwIEDSUpKAiA2Npbf/e53xMXFkZyczLPPPkuXLl1o27Yto0aN8vTDxtatW+nXrx9xcXF06tSJbdu2cddddzFr1qy85d5xxx38+9//LvL7DoYmAqUquqNHITfXvs7NtcNlULduXbp27cpXX30F2KOBW265BRHhySefJCUlhbVr17J48WLWrl0bcDmrVq1i+vTprF69mrlz57Jy5cq8cTfeeCMrV65kzZo1tGrVivfee4+ePXsyaNAgXn75ZVavXk2LFi3yps/MzGTYsGF8+umnrFu3juzsbP7xj3/kja9fvz7fffcd9957r9/qp7S0NGbOnMmGDRtYu3YtTzzxBABjx45l3LhxrFy5ks8//5wRI0bQrFkz7r//fsaNG8fq1avp06dPgWVdcsklLF26lO+//55nn32WCRMmAPDuu++SmprK6tWrWbt2LXfccQcnT57k1ltv5Y033mDNmjXMnj2batWqFbn9f/31V7p168aaNWvo3bs3Y8aMYeXKlaxfv54TJ04wZ84cwBbyDz74IGvWrGHZsmU0bNiQe++9l6lTpwJw9OhRli1bxrXXXlvk+oKh1xEoFU6vv178NMnJcMUVcPIkVKkC06ZBjx5lWq2nemjw4MFMnz6d9957D4CZM2fy4Ycfkp2dzZ49e9i4cSPt27f3u4ylS5dyww03UL16dQAGDRqUN279+vU88cQTHDlyhIyMDK6++uoi4/nhhx9o3rw5F198MQB33303kyZN4hHn6ObGG28EoEOHDsydO7fQ/LVq1SImJoZ7772XgQMHMnDgQAAWLFhQoCrq2LFjZGRkFBnL0aNHufvuu9myZQsiwqlTp/KWdf/991Opki0269aty7p162jYsCFdunQB4KyzzsobH0h0dDRDhgzJG160aBF/+ctfOH78OIcOHaJNmzYkJCSwa9cubrjhBsBeJAZw2WWXMXr0aA4cOMDnn3/OkCFDil1fMDQRKFXR9egBCxdCUpKt+iljEgAYPHgw48aN47vvvuP48ePEx8fz888/8+abb7Jq1Srq1KnDsGHDSn3187Bhw5g1axZxcXFMnTo1r1qktKpWrQrYQjQ7OxuAq6++mn379tG5c2cmT57MihUrWLhwITNmzODvf/8733zzDbm5uSxfvjyvIA3Gk08+Sd++fZk5cyapqakkJCSUON5KlSqR6zmKgwLbMSYmhujo6Lz3R48eTUpKCk2aNGHixInFbvO77rqLjz/+mOnTpzNlypQSx+aPVg0pdTro0QMef7xckgDYeuq+ffsyfPjwvEbiY8eOUaNGDWrVqsW+ffvyqo4CufTSS5k1axYnTpwgPT2dL7/8Mm9ceno6DRs25NSpU0ybNi3v/Zo1a5Kenl5oWS1btiQ1NZWtW7cC8NFHH3HZZZcVuf558+axevVqJk+eTEZGBkePHmXAgAG89tprrFmzBoCrrroqry4fYPXq1UXGAfaIoFGjRgB51TAAV155Je+8805eIjp06BAtW7Zkz549edVi6enpZGdn06xZM1avXk1ubi47duxgxYoVftflKfTr169PRkZGXntJzZo1ady4cV57QFZWFsePHwdskn3dOZJs3bp1kdsoWJoIlIpQQ4cOZc2aNXmJIC4ujvbt23PJJZdw++2306tXryLn79SpE7feeitxcXFcc801edUjAH/605/o1q0bvXr14pJLLsl7/7bbbuPll1+mY8eObNu2Le/9mJgYpkyZws0330y7du2Iiori/vvvD/qzpKenM3DgQNq3b0/v3r159dVXAXjzzTdJSUmhffv2tG7dmrfffhuA6667jpkzZ/ptLP7DH/7A448/TseOHfMKfYARI0bQtGlT2rdvT1xcHP/617+oUqUKn376KQ899BBxcXEMHjyYzMxMevXqRfPmzWndujUPP/wwnTp18ht37dq1GTlyJG3btuXqq68usA0/+ugj3nzzTdq3b0/Pnj3Zu3cvAA0aNKBVq1bcc889QW+fYgW6Y01FfegdysJH4ysbT3wbN24MbyBFOHbsWLhDKJLGZ8yvv/5qLrjgAnPkyJGA0/j7jRGmO5QppZQqRwsWLKBVq1Y89NBD1KpVq9yWq43FSil1mujXrx+//PJLuS9XjwiUUirCaSJQSqkIp4lAKaUinCYCpZSKcJoIlIowaWlpdOjQgQ4dOnDuuefSqFGjvGFPR3SBpKSk8PDDD7sW25EjR3jrrbdcW77yT88aUuo0kJxcfj1M1KtXL+8K24kTJxIbG5vXW6fnythA/dd07tyZzp07ly2AIngSwejRo11bRzCK2gZnosj5pEpVQI88Ak6ZHNDRo7B2re14NCoK2reHok4h79AhuL7svA0bNoyYmBhSUlK49NJLue222xg7diyZmZlUq1aNKVOm0LJlS5KSknjllVeYM2cOEydOZPv27fz0009s376dRx55xO/RwuLFixk7dixgb6O4ZMkSatasycsvv8xnn31GVlYWN9xwA8888wzjx49n27ZtdOjQgSuvvJKXX365wLKGDh3Knj17yMzMZOzYsYwaNQqAr7/+mgkTJpCTk0P9+vVZuHAhGRkZPPTQQ6SkpCAiPP300wwZMoTY2Ni8judmzJjBnDlzmDp1at42+P777+nVq1fAbZCTk8Njjz3G119/TVRUFCNHjqRNmza8+eabfPTRRwAkJiby1ltvMXPmzJJ9EWGiiUCpCs5fL9TleC1Rnp07d7JgwQJq167NsWPHWLp0KZUqVWLBggVMmDCBzz//vNA8mzdvZtGiRaSnp9OyZUseeOABKleuXGCaV155hUmTJtGrVy8yMjKIiYlh/vz5bNmyhRUrVmCMYdCgQSxZsoSXXnqJ9evX5x2x+Jo0aRLnn38+J06coEuXLgwZMoTc3FxGjhzJkiVLaN68OYcOHQJsNxe1atVi3bp1ABw+fDiobbBs2TKio6MDbgPv7qgrVarEoUOHqFOnDqNHj+bgwYPUrFmTKVOmMHz48BJ+A+GjiUCpMApTL9R+3XzzzXm9YgbqitnXtddeS9WqValatSrnnHMO+/bto3HjxgWm6dWrF7/97W+54447uPHGG2ncuDHz589n/vz5dOzYEYCMjAy2bNlC06ZNi4zx7bffzuuGeseOHWzZsoUDBw5w6aWX0rx5c8B2Dw32KlzvO7HVqVOnXLaBv+6oAe68806mT5/OAw88QHJyMh9++GGx66soNBEoVcG50Au1XzVq1Mh7HWxXzJ7uoSG/i+hJkybxz3/+E7C3tRw/fjzXXnstc+fOpVevXsybNw9jDI8//jj33XdfgeWlpqYGjC8pKYmkpCSSk5OpXr06CQkJpeomW0TyXvvOX5pt4HHPPfdw7bXXUrt2bW6++ebTqo1BzxpS6jRQzr1QFytQV8zBePDBB1m9ejWrV6/mvPPOY9u2bbRr147HHnuMLl26sHnzZq6++mref//9vLr6Xbt2sX///mK7h65duzbVq1dn8+bNLF++HIDu3buzZMkSfv75Z4C8qqErr7ySSZMm5c3vqRpq0KABmzZtIjc3t8g6/JJ0Rw1w3nnnce655/Lcc8+Vb8+gIaCJQClVSKCumEvj9ddfz7vPb+XKlbnmmmu46qqruP322+nRowft2rXjpptuIj09nXr16tGrVy/atm3L73//+wLL6d+/P9nZ2bRq1Yrx48fTvXt3AM4++2zeffddbrzxRuLi4rj11lsBeOKJJzh8+DBt27YlLi6ORYsWAfDSSy8xcOBAevbsScOGDUu8Dfx1R+1xyy230KRJE1q1alWmbRZygbolragP7YY6fDS+stFuqMuuosc3cuRIM3ny5HCHUeJuqE+fSiyllKrA4uPjiYmJKXBHtNOFJgKllCoHq1atIj09vUAD+ulC2wiUCgN7pK5U+SvNb0sTgVIhFhMTQ1pamiYDVe6MMaSlpRETE1Oi+bRqSKkQa9y4MTt37uTAgQPhDqWQzMzMEhcioaTxFS8mJqbQRX3F0USgVIhVrlw57yrYiiYpKSnvat+KSONzh6tVQyLSX0R+EJGtIjLez/hhInJARFY7jxFuxqOUUqow144IRCQamARcCewEVorIbGPMRp9JPzXGjHErDqWUUkVz84igK7DVGPOTMeYkMB0Y7OL6lFJKlYKbbQSNgB1ewzuBbn6mGyIilwI/AuOMMTt8JxCRUcAoZzBDRH4oZUz1gYOlnDcUNL6y0fjKrqLHqPGV3vmBRoS7sfhL4BNjTJaI3Ad8AFzuO5Ex5l3g3bKuTERSjDHu3V6pjDS+stH4yq6ix6jxucPNqqFdQBOv4cbOe3mMMWnGmCxncDIQ72I8Siml/HAzEawELhKR5iJSBbgNmO09gYh4d/03CNjkYjxKKaX8cK1qyBiTLSJjgHlANPC+MWaDiDyL7QVvNvCwiAwCsoFDwDC34nGUuXrJZRpf2Wh8ZVfRY9T4XCB6mbtSSkU27WtIKaUinCYCpZSKcGdkIgiia4uqIvKpM/5/ItIshLE1EZFFIrJRRDaIyFg/0ySIyFGvrjeeClV8zvpTRWSds+4UP+NFRN50tt9aEekUwthaem2X1SJyTEQe8Zkm5NtPRN4Xkf0ist7rvboikigiW5znOgHmvduZZouI3B2i2F4Wkc3O9zdTRGoHmLfI34LLMU4UkV1e3+OAAPMW+X93Mb5PvWJLFZHVAeYNyTYsk0C3LjtdH9iG6W3ABUAVYA3Q2mea0cDbzuvbsN1chCq+hkAn53VN7IV0vvElAHPCuA1TgfpFjB8AfAUI0B34Xxi/673A+eHefsClQCdgvdd7fwHGO6/HA3/2M19d4CfnuY7zuk4IYrsKqOS8/rO/2IL5Lbgc40Tg0SB+A0X+392Kz2f8X4GnwrkNy/I4E48IgunaYjD24jWAGcAVIiKhCM4Ys8cY853zOh17ymyjUKy7HA0GPjTWcqC2z6nAoXIFsM0Y80sY1l2AMWYJ9sw3b96/sw+A6/3MejWQaIw5ZIw5DCQC/d2OzRgz3xjjuSP7cux1PmETYPsFIyRd2RQVn1N23AJ8Ut7rDZUzMRH469rCt6DNm8b5MxwF6oUkOi9OlVRH4H9+RvcQkTUi8pWItAltZBhgvoiscrr38BXMNg6F2wj85wvn9vNoYIzZ47zeCzTwM01F2JbDsUd4/hT3W3DbGKf66v0AVWsVYfv1AfYZY7YEGB/ubVisMzERnBZEJBb4HHjEGHPMZ/R32OqOOOBvwKwQh9fbGNMJuAZ4UGxfUBWKc5HiIOD//IwO9/YrxNg6ggp3rraI/BF7Hc+0AJOE87fwD6AF0AHYg61+qYiGUvTRQIX/P52JiaDYri28pxGRSkAtIC0k0dl1VsYmgWnGmC98xxtjjhljMpzXc4HKIlI/VPEZY3Y5z/uBmdjDb2/BbGO3XQN8Z4zZ5zsi3NvPyz5PlZnzvN/PNGHbliIyDBgI3OEkqkKC+C24xhizzxiTY4zJBf4ZYN1h/S065ceNwKeBpgnnNgzWmZgIiu3awhn2nJ1xE/BNoD9CeXPqE98DNhljXg0wzbmeNgsR6Yr9nkKSqESkhojU9LzGNiqu95lsNnCXc/ZQd+CoVxVIqATcCwvn9vPh/Tu7G/i3n2nmAVeJSB2n6uMq5z1XiUh/4A/AIGPM8QDTBPNbcDNG73anGwKsO5j/u5v6AZuNMTv9jQz3NgxauFur3Xhgz2r5EXs2wR+d957F/ugBYrBVCluBFcAFIYytN7aKYC2w2nkMAO4H7nemGQNswJ4BsRzoGcL4LnDWu8aJwbP9vOMT7E2HtgHrgM4h/n5rYAv2Wl7vhXX7YZPSHuAUtp76Xmy700JgC7AAqOtM2xmY7DXvcOe3uBW4J0SxbcXWrXt+g56z6M4D5hb1Wwjh9vvI+X2txRbuDX1jdIYL/d9DEZ/z/lTP785r2rBsw7I8tIsJpZSKcGdi1ZBSSqkS0ESglFIRThOBUkpFOE0ESikV4TQRKKVUhNNEoFQIOT2jzgl3HEp500SglFIRThOBUn6IyG9EZIXTh/w7IhItIhki8prY+0gsFJGznWk7iMhyr7796zjvXygiC5zO774TkRbO4mNFZIZzP4Bpoer5VqlANBEo5UNEWgG3Ar2MMR2AHOAO7BXNKcaYNsBi4Glnlg+Bx4wx7bFXwnrenwZMMrbzu57YK1PB9jj7CNAae+VpL5c/klJFqhTuAJSqgK4A4oGVzs56NWyHcbnkdy72MfCFiNQCahtjFjvvfwD8n9O/TCNjzEwAY0wmgLO8Fcbpm8a5q1Uz4FvXP5VSAWgiUKowAT4wxjxe4E2RJ32mK23/LFler3PQ/6EKM60aUqqwhcBNInIO5N17+Hzs/+UmZ5rbgW+NMUeBwyLSx3n/TmCxsXef2yki1zvLqCoi1UP5IZQKlu6JKOXDGLNRRJ7A3lUqCtvj5IPAr0BXZ9x+bDsC2C6m33YK+p+Ae5z37wTeEZFnnWXcHMKPoVTQtPdRpYIkIhnGmNhwx6FUedOqIaWUinB6RKCUUhFOjwiUUirCaSJQSqkIp4lAKaUinCYCpZSKcJoIlFIqwv0/uxEXuwnOXjwAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# accuracy 값을 plot 해보겠습니다. \n",
    "y_vacc = history.history['val_accuracy']\n",
    "y_acc = history.history['accuracy']\n",
    "x_len = np.arange(len(y_acc))\n",
    "\n",
    "plt.plot(x_len, y_vacc, marker='.', c='red', label=\"Validation-set accuracy\")\n",
    "plt.plot(x_len, y_acc, marker='.', c='blue', label=\"Train-set accuracy\")\n",
    "plt.legend(loc='lower right')\n",
    "plt.grid()\n",
    "plt.ylim(0.5,1) \n",
    "plt.title('Accurcy graph with batch normalization') \n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('accuracy')\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}