HW2MNIST.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.utils import np_utils  #匯入keras.utils\n",
    "import numpy as np #匯入numpy模組\n",
    "np.random.seed(10) #設定seed讓每次需要隨機產生的資料\n",
    "from keras.datasets import mnist #匯入mnist模組"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "(x_train_image,y_train_label),\\\n",
    "(x_test_image,y_test_label)=mnist.load_data() #讀取mnist資料集"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#將原本28X28數字影像，用reshape轉換為784個float數字\n",
    "x_Train = x_train_image.reshape(60000,784).astype('float32')\n",
    "x_Test = x_test_image.reshape(10000,784).astype('float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "#將features(影像特徵值)標準化，可以提高模型預測的準確度，並加快收斂\n",
    "x_Train_normalize = x_Train/255\n",
    "x_Test_normalize = x_Test/255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#使用np_utils.to_categorical將訓練資料與測試資料的label進行One-Hot encoding轉換\n",
    "y_Train_OneHot = np_utils.to_categorical(y_train_label)\n",
    "y_Test_OneHot = np_utils.to_categorical(y_test_label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#建立模型前須匯入需要用到之模組\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential() #建立一個線性堆疊模型，之後只要使用model.add()方法，即可將神經網路層加入模型中"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#通過使用model.add方法，加入Dense神經網路層。Dense神經網路層的特色為所有上一層與下一層的神經元都為完全連結狀態\n",
    "#units為定義隱藏層神經元數量，input_dim為設定輸入層的神經元個數，kernel_initializer='normal'為使用常態分佈的亂數初始化weight(權重)及bias(偏差值)，activation為定義激活函數為relu\n",
    "model.add(Dense(units=256,input_dim=784,kernel_initializer='normal',activation='relu'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "#建立輸出層 共有10個神經元，對應0~9 10個數字，使用softmax激活函數進行轉換，softmax可以將神經元的輸出轉換為預測每一個數字的機率\n",
    "#units為定義輸出層神經元個數，kernel_initializer='normal'為使用常態分佈的亂數初始化weight(權重)及bias(偏差值)，activation定義激活函數為softmax\n",
    "model.add(Dense(units=10,kernel_initializer='normal',activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 256)               200960    \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                2570      \n",
      "=================================================================\n",
      "Total params: 203,530\n",
      "Trainable params: 203,530\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "print(model.summary()) #此指令可察看模型摘要"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy']) #訓練模型前，使用compile方法對模型進行設定\n",
    "#loss 為設定損失函數，在深度學習通常使用cross_entropy交叉墒，訓練效果比較好\n",
    "#optimizer 為設定訓練時的最優化方法，在深度學習使用adam最優畫方法，可以讓訓練更快收斂並提高準確率\n",
    "#metrics 為設定評估模型的方式為accuracy準確率"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 48000 samples, validate on 12000 samples\n",
      "Epoch 1/10\n",
      " - 1s - loss: 0.4382 - acc: 0.8828 - val_loss: 0.2182 - val_acc: 0.9404\n",
      "Epoch 2/10\n",
      " - 1s - loss: 0.1908 - acc: 0.9456 - val_loss: 0.1558 - val_acc: 0.9558\n",
      "Epoch 3/10\n",
      " - 1s - loss: 0.1354 - acc: 0.9618 - val_loss: 0.1259 - val_acc: 0.9645\n",
      "Epoch 4/10\n",
      " - 1s - loss: 0.1026 - acc: 0.9702 - val_loss: 0.1121 - val_acc: 0.9680\n",
      "Epoch 5/10\n",
      " - 1s - loss: 0.0810 - acc: 0.9772 - val_loss: 0.0979 - val_acc: 0.9721\n",
      "Epoch 6/10\n",
      " - 1s - loss: 0.0659 - acc: 0.9819 - val_loss: 0.0932 - val_acc: 0.9723\n",
      "Epoch 7/10\n",
      " - 1s - loss: 0.0542 - acc: 0.9851 - val_loss: 0.0916 - val_acc: 0.9739\n",
      "Epoch 8/10\n",
      " - 1s - loss: 0.0458 - acc: 0.9876 - val_loss: 0.0826 - val_acc: 0.9761\n",
      "Epoch 9/10\n",
      " - 1s - loss: 0.0379 - acc: 0.9904 - val_loss: 0.0820 - val_acc: 0.9760\n",
      "Epoch 10/10\n",
      " - 1s - loss: 0.0315 - acc: 0.9916 - val_loss: 0.0801 - val_acc: 0.9761\n"
     ]
    }
   ],
   "source": [
    "train_history=model.fit(x=x_Train_normalize,y=y_Train_OneHot,validation_split=0.2,epochs=10,batch_size=200,verbose=2) #使用model.fit進行訓練，過程會儲存在train_history變數\n",
    "#x=x_Train_normalize 為features(影像特徵值)\n",
    "#y=y_Train_OneHot 為label(影像真實值)\n",
    "#validation_split=0.2 訓練前keras會自動將資料分成80%訓練資料及20%驗證資料\n",
    "#epochs=10 為設定執行10次訓練週期\n",
    "#batch_size=200 設定為每一批次200筆資料\n",
    "#verbose=2 顯示訓練過程"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "#使用matplotlib.pyplot顯示訓練過程\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline \n",
    "#設定matplotlib在jupyter note網頁內顯示圖形，如果沒使用此指令，會另開視窗顯示圖形\n",
    "def show_train_history(train_history,train,validation): #train_history為之前訓練過程所產生的train_history，train為訓練資料的執行結果，validation為驗證資料的執行結果\n",
    "    plt.plot(train_history.history[train])\n",
    "    plt.plot(train_history.history[validation])\n",
    "    plt.title('Train History') #顯示圖的標題\n",
    "    plt.ylabel(train) #顯示y軸標籤\n",
    "    plt.xlabel('Epoch') #顯示x軸標籤為Epoch\n",
    "    plt.legend(['train','validation'],loc = 'upper left') #設定圖利為顯示'train','validation'，位置在左上角\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_train_history(train_history,'acc','val_acc') #畫出accuracy準確率執行結果"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzt3Xd4XOWZ9/Hvrd6LpbGQVZCbjCuyEaZbEMBrSGiJA4ZAgIVQElLf7IaUd5OQsizZlyVZDMRJICFAHGJCYgjNIS5AMFguuOKKsWXZsiRbsmSr637/OEfSSJZV52hU7s91zTVz6twzYP3mOc85zxFVxRhjjOlKSLALMMYYM/hZWBhjjOmWhYUxxphuWVgYY4zploWFMcaYbllYGGOM6ZaFhTGnICKhIlItItke7X+ciFR7sW9jAs3Cwgwb7h/2lkeziNT4TX+ut/tT1SZVjVPVfX2oZYKInHQRk4g8IyI/cPe/R1XjerCvO0VkRW9rMCaQwoJdgDGB4v+HV0T2Aneq6t9Ptb6IhKlq40DUFkwj5XMab1nLwowYIvJjEfmjiPxBRKqAm0XkPBFZLSIVInJQRH4hIuHu+mEioiKS404/4y5/VUSqRORdERnbj3ratT5E5A4R2evue4+ILBCR6cCjwEVuC6nMXTfJrafU3ebbIiLusjtFZJVb6xHgx+7nm+z3XukickJEUvpavxlZLCzMSHMd8ByQCPwRaAS+CqQCFwDzgLu72P4m4P8Co4B9wI8CUZSIJAAPA5erarxby0ZV3QTcB7zlHhJLdTd5DIgBxgGfAO4APu+3y/OBbYAP+CHwPHBzh8/xuqqWB6J+M/xZWJiR5m1VfUlVm1W1RlXXqOp7qtqoqnuARUBBF9svUdVCVW0AngXyunoz9xd96wO4vovVFZgmIlGqelBVt55in+Hufu5X1Sq37v8BbvFbbZ+qPu72u9QAvwNuaml9uOv+vqvajfFnYWFGmv3+EyJyhoj8TUQOicgx4AGcVsapHPJ7fQLosoNaVZP8Hzi/8Dtb7xhwI/Al4JCIvCwiuafY7WggFPjYb97HQIbfdLvPqarv4LSiLhSRaUA28LeuajfGn4WFGWk6nqH0S2AzMEFVE4D/AOSkrQaAqr6qqpcB6cAutzY4uebDQBNwut+8bOCA/+46eYuncQ5F3QI8r6p1gajbjAwWFmakiwcqgeNuB3BX/RWecTucrxKRGKAeOI4TCAAlQGZLx7t7CGwJ8FMRiXM72b8OPNPN2/wemI/TX/G0Bx/DDGMWFmak+z/ArUAVzi/5PwapjlDg34CDQDlOB/V97rJlwE6gRERaDoN9ESdUPgJW4vRJdBkAqroX2ATUq+o/A1y/GebEbn5kzMghIk8De1T1B8GuxQwtdlGeMSOEiIwDrgGmB7sWM/TYYShjRgAR+U/gA+CnfRm+xBg7DGWMMaZb1rIwxhjTrWHTZ5Gamqo5OTnBLsMYY4aUtWvXlqmqr7v1PA0LEZkH/BzntMBfq+qDp1hvPvAn4GxVLXQHbtsGbHdXWa2q93T1Xjk5ORQWFgaqdGOMGRFE5OPu1/IwLEQkFFgIXA4UAWtEZGnH8W5EJB74CvBeh13sVtUux90xxhgzMLzss5gN7HJv8FIPLMY5ba+jHwEPAbUe1mKMMaYfvAyLDNoPZlZE+4HOEJGZQJaqvtzJ9mNFZL2IrBSRizp7AxG5S0QKRaSwtLQ0YIUbY4xpz8s+i84GY/O/0UsIzrDKt3Wy3kEgW1XLReQs4C8iMtUdmbNtZ6qLcIaUJj8//6RzgBsaGigqKqK21hotgRIVFUVmZibh4eHBLsUYM4C8DIsiIMtvOhMo9puOB6YBK9wh9k8DlorI1apaCNQBqOpaEdkN5AK96sEuKioiPj6enJwc2obxN32lqpSXl1NUVMTYsX2+QZwxZgjy8jDUGmCiiIwVkQhgAbC0ZaGqVqpqqqrmqGoOsBq42j0byud2kLcMUTAR2NPbAmpra0lJSbGgCBARISUlxVpqxoxAnrUsVLVRRO4DXsc5dfZJVd0iIg8Ahaq6tIvN5wAPiEgjzjDN96jqkb7UYUERWPZ9GjMyeXqdhaq+ArzSYd5/nGLdi/1evwC84GVtLRqbmik/Xk9CVBjREcPmGkVjjAkoG+5D4PCxOiprGj3ZfUVFBY899livt7vyyiupqKjwoCJjjOm9ER8WYSEhxESEUlXb4Mn+TxUWTU1Nnazd5pVXXiEpKcmTmowxprdGfFgAxEeFUdPQRENTc8D3ff/997N7927y8vI4++yzueSSS7jpppuYPt25pcC1117LWWedxdSpU1m0aFHrdjk5OZSVlbF3714mT57MF77wBaZOncrcuXOpqakJeJ3GGNOVEXOQ/ocvbWFr8bFOlzWrUlPfRGR4KGEhPe/AnTImge9fNbXLdR588EE2b97Mhg0bWLFiBZ/85CfZvHlz66mnTz75JKNGjaKmpoazzz6bz3zmM6SkpLTbx86dO/nDH/7Ar371K66//npeeOEFbr755h7XaYwx/TViwqIrISKICE3N2quw6IvZs2e3u0bhF7/4BS+++CIA+/fvZ+fOnSeFxdixY8nLc4bJOuuss9i7d6+nNRpjTEcjJiy6awHsP3KCqtpGJqfHe3p6aGxsbOvrFStW8Pe//513332XmJgYLr744k6vYYiMjGx9HRoaaoehjDEDzvosXHFRYTQ2N1PT0HXHc2/Fx8dTVVXV6bLKykqSk5OJiYnhww8/ZPXq1QF9b2OMCZQR07LoTnyk81VU1TYSE8DrLVJSUrjggguYNm0a0dHRpKWltS6bN28eTzzxBDNmzGDSpEmce+65AXtfY4wJpGFzD+78/HztePOjbdu2MXny5B7vY+fhKkIQxo+OC3R5w0pvv1djzOAlImtVNb+79ewwlJ/4yHBO1DfS6MEptMYYM5RZWPiJjwpDgeN13lzNbYwxQ5WFhZ+YiFBCQ4SqWgsLY4zxZ2HhR0SIiwyjqq6R4dKXY4wxgWBh0UF8VBgNTc3UNVq/hTHGtLCw6CAu0rldqFcDCxpjzFBkYdFBRFgIUeGhQeu3iItzTtstLi5m/vz5na5z8cUX0/E04Y4eeeQRTpw40TptQ54bY/rD07AQkXkisl1EdonI/V2sN19EVETy/eZ9291uu4j8i5d1dhQfFcbx+iaamoPXbzFmzBiWLFnS5+07hoUNeW6M6Q/PwsK9h/ZC4ApgCnCjiEzpZL144CvAe37zpuDcs3sqMA94rOWe3AMhPjIMVQ3IKbTf+ta32t3P4gc/+AE//OEPufTSS5k1axbTp0/nr3/960nb7d27l2nTpgFQU1PDggULmDFjBjfccEO7saHuvfde8vPzmTp1Kt///vcBZ3DC4uJiLrnkEi655BKgbchzgIcffphp06Yxbdo0Hnnkkdb3s6HQjTGn4uVwH7OBXaq6B0BEFgPXAFs7rPcj4CHgm37zrgEWq2od8JGI7HL3926fq3n1fji0qUerxqKMr29yRqAN6yKjTpsOVzzY5b4WLFjA1772Nb74xS8C8Pzzz/Paa6/x9a9/nYSEBMrKyjj33HO5+uqrTzmA4eOPP05MTAwbN25k48aNzJo1q3XZT37yE0aNGkVTUxOXXnopGzdu5Ctf+QoPP/wwy5cvJzU1td2+1q5dy1NPPcV7772HqnLOOedQUFBAcnKyDYVujDklLw9DZQD7/aaL3HmtRGQmkKWqL/d2W3f7u0SkUEQKS0tLA1M1IAih7pDl/TVz5kwOHz5McXExH3zwAcnJyaSnp/Od73yHGTNmcNlll3HgwAFKSkpOuY9Vq1a1/tGeMWMGM2bMaF32/PPPM2vWLGbOnMmWLVvYurVjFrf39ttvc9111xEbG0tcXByf/vSneeuttwAbCt0Yc2petiw6+5nc+tdXREKA/wFu6+22rTNUFwGLwBkbqstqumkBdFRdXUdxRQ2T0uKJDO/fEbD58+ezZMkSDh06xIIFC3j22WcpLS1l7dq1hIeHk5OT0+nQ5P46a3V89NFH/Pd//zdr1qwhOTmZ2267rdv9dHX9iA2Fbow5FS9bFkVAlt90JlDsNx0PTANWiMhe4FxgqdvJ3d22nouPckehDUC/xYIFC1i8eDFLlixh/vz5VFZWMnr0aMLDw1m+fDkff/xxl9vPmTOHZ599FoDNmzezceNGAI4dO0ZsbCyJiYmUlJTw6quvttV/iqHR58yZw1/+8hdOnDjB8ePHefHFF7nooov6/RmNMcObly2LNcBEERkLHMDpsL6pZaGqVgKtB9RFZAXwTVUtFJEa4DkReRgYA0wE3vew1pNEhoUSGRZCdW0jqXGR3W/QhalTp1JVVUVGRgbp6el87nOf46qrriI/P5+8vDzOOOOMLre/9957uf3225kxYwZ5eXnMnj0bgDPPPJOZM2cydepUxo0bxwUXXNC6zV133cUVV1xBeno6y5cvb50/a9YsbrvtttZ93HnnncycOdMOORljuuTpEOUiciXwCBAKPKmqPxGRB4BCVV3aYd0VuGHhTn8X+FegEfiaqr5KFwIxRHlHBypqOHq8ninpCYR4fLvVocSGKDdm+OjpEOWe3vxIVV8BXukw7z9Ose7FHaZ/AvzEs+J6ID4yjPLqOo7XNxIfFR7MUowxJqjsCu4uxEaGISJU2yi0xpgRbtiHRX8Os4WGCLERoQHp5B4ubDReY0amYR0WUVFRlJeX9+sPXHxUOLUNTdTbKLSoKuXl5URFRQW7FGPMAPO0zyLYMjMzKSoqoj8X7DU0NVNyrI76snBiI4f119UjUVFRZGZmBrsMY8wAG9Z//cLDwxk7dmy/9qGq3P3gP8jLSuLxm6cHqDJjjBlahvVhqEAQEQpyfby9s4yGJjsUZYwZmSwseqAg10dVXSMb9tv9IIwxI5OFRQ+cPyGV0BBh5fbADVZojDFDiYVFDyRGhzMrO4mVOywsjDEjk4VFDxXk+th0oJKy6rpgl2KMMQPOwqKHCnJHA/DWTmtdGGNGHguLHpo6JoGU2AjrtzDGjEgWFj0UEiLMyfWxamcZzQG4g54xxgwlFha9UJDr48jxejYXVwa7FGOMGVAWFr1w0cRURLBDUcaYEcfTsBCReSKyXUR2icj9nSy/R0Q2icgGEXlbRKa483NEpMadv0FEnvCyzp5KiYtkekainUJrjBlxPAsLEQkFFgJXAFOAG1vCwM9zqjpdVfOAh4CH/ZbtVtU893GPV3X2VkGuj3X7jlJ5oiHYpRhjzIDxsmUxG9ilqntUtR5YDFzjv4KqHvObjAUGfc9xQa6PZoV3dpcFuxRjjBkwXoZFBrDfb7rIndeOiHxJRHbjtCy+4rdorIisF5GVInJRZ28gIneJSKGIFPZnGPLeyMtKIj4qzPotjDEjipdhIZ3MO6nloKoLVXU88C3ge+7sg0C2qs4EvgE8JyIJnWy7SFXzVTXf5/MFsPRTCwsN4aKJqazcUWp3jTPGjBhehkURkOU3nQkUd7H+YuBaAFWtU9Vy9/VaYDeQ61GdvVaQ6+PQsVp2lFQHuxRjjBkQXobFGmCiiIwVkQhgAbDUfwURmeg3+Ulgpzvf53aQIyLjgInAHg9r7ZU5uU4rZuWOw0GuxBhjBoZnYaGqjcB9wOvANuB5Vd0iIg+IyNXuaveJyBYR2YBzuOlWd/4cYKOIfAAsAe5R1SNe1dpb6YnRTEqLt1NojTEjhqe3VVXVV4BXOsz7D7/XXz3Fdi8AL3hZW38VTPLx23f2cryu0e7NbYwZ9uwK7j4qyPVR39TM6j3lwS7FGGM8Z2HRR/k5yUSHh9qhKGPMiGBh0UeRYaGcPz7FwsIYMyJYWPRDwSQfH5efYG/Z8WCXYowxnrKw6IeC1lNorXVhjBneLCz64fSUWHJSYiwsjDHDnoVFPxXk+nh3dzm1DU3BLsUYYzxjYdFPBZN81DQ0Ubj3aLBLMcYYz1hY9NO541KICA2xoT+MMcOahUU/xUSEMXvsKOu3MMYMaxYWAVCQ62NHSTXFFTXBLsUYYzxhYREABZOcU2hXWevCGDNMWVgEwMTRcaQnRtmhKGPMsGVhEQAiQkGuj7d3ltHQ1BzscowxJuAsLAKkINdHVV0jG/ZXBLsUY4wJOE/DQkTmich2EdklIvd3svweEdkkIhtE5G0RmeK37NvudttF5F+8rDMQzp+QSmiIsHK7HYoyxgw/noWFe1vUhcAVwBTgRv8wcD2nqtNVNQ94CHjY3XYKzm1YpwLzgMdabrM6WCVGhzMrO8n6LYwxw5KXLYvZwC5V3aOq9cBi4Br/FVT1mN9kLKDu62uAxapap6ofAbvc/Q1qBbk+Nh2opKy6LtilGGNMQHkZFhnAfr/pIndeOyLyJRHZjdOy+Eovt71LRApFpLC0NPi/6AtyRwPw1s7g12KMMYHkZVhIJ/P0pBmqC1V1PPAt4Hu93HaRquarar7P5+tXsYEwdUwCKbER1m9hjBl2vAyLIiDLbzoTKO5i/cXAtX3cdlAICRHm5PpYtbOM5uaTss0YY4YsL8NiDTBRRMaKSAROh/VS/xVEZKLf5CeBne7rpcACEYkUkbHAROB9D2sNmIJcH0eO17O5uDLYpRhjTMCEebVjVW0UkfuA14FQ4ElV3SIiDwCFqroUuE9ELgMagKPAre62W0TkeWAr0Ah8SVWHxA0jLpqYigis3F7KjMykYJdjjDEBIarD43BJfn6+FhYWBrsMAK5+9G0iQkNYcu/5wS7FGGO6JCJrVTW/u/XsCm4PFOT6WLfvKJUnGoJdijHGBISFhQcKcn00K7yzuyzYpRhjTEBYWHggLyuJ+KgwO4XWGDNsWFh4ICw0hIsmprJyRynDpU/IGDOyWVh4pCDXx6FjtewoqQ52KcYY028WFh6Zk+tcUb5yx+EgV2KMMf1nYeGR9MRoJqXF2yi0xphhwcLCQwWTfKz56CjH6xqDXYoxxvSLhYWHCnJ91Dc1s3pPebBLMcaYfrGw8FB+TjLR4aF2KMoYM+RZWHgoMiyU88enWFgYY4Y8CwuPFUzy8XH5CfaWHQ92KcYY02cWFh4raD2F1loXxpihy8LCY6enxJKTEmNhYYwZ0iwsBkBBro93d5dT2zAkbslhjDEnsbAYAAWTfNQ0NFG492iwSzHGmD7xNCxEZJ6IbBeRXSJyfyfLvyEiW0Vko4i8KSKn+y1rEpEN7mNpx22HknPHpRARGmJDfxhjhizPwkJEQoGFwBXAFOBGEZnSYbX1QL6qzgCWAA/5LatR1Tz3cbVXdQ6EmIgwZo8dZf0Wxpghy8uWxWxgl6ruUdV6YDFwjf8KqrpcVU+4k6uBTA/rCaqCXB87SqoprqgJdinGGNNrXoZFBrDfb7rInXcqdwCv+k1HiUihiKwWkWs720BE7nLXKSwtHdy/2gsmOafQrrLWhTFmCOpRWIjIV0UkQRy/EZF1IjK3u806mdfpnYBE5GYgH/iZ3+xs9ybiNwGPiMj4k3amukhV81U13+fz9eSjBM3E0XGkJ0bZoShjzJDU05bFv6rqMWAu4ANuBx7sZpsiIMtvOhMo7riSiFwGfBe4WlXrWuararH7vAdYAczsYa2DkohQkOvj7Z1lNDQ1B7scY4zplZ6GRUsr4UrgKVX9gM5bDv7WABNFZKyIRAALgHZnNYnITOCXOEFx2G9+sohEuq9TgQuArT2sddAqyPVRVdfIhv0VwS7FGGN6padhsVZE3sAJi9dFJB7o8uexqjYC9wGvA9uA51V1i4g8ICItZzf9DIgD/tThFNnJQKGIfAAsBx5U1SEfFudPSCU0RFi53Q5FGWOGFlHttBuh/UoiIUAesEdVK0RkFJCpqhu9LrCn8vPztbCwMNhldOuzT/yT2oZmXvryhcEuxRhjEJG1bv9wl3rasjgP2O4Gxc3A94DK/hQ4UhXk+th0oJKy6rruVzbGmEGip2HxOHBCRM4E/h34GHjas6qGsYLc0QC8tdMORRljho6ehkWjOserrgF+rqo/B+K9K2v4mjomgZTYCOu3MMYMKWE9XK9KRL4N3AJc5A7lEe5dWcNXSIgwJ9fHyh2lNDcrISHdnVRmjDHB19OWxQ1AHc71FodwrsT+WdebmFMpyPVx5Hg9m4ut28cYMzT0KCzcgHgWSBSRTwG1qmp9Fn100cRURLBDUcaYIaOnw31cD7wPfBa4HnhPROZ7WdhwlhIXyfSMRBv6wxgzZPS0z+K7wNktV1mLiA/4O86w4qYPCnJ9LFy+i8oTDSTGWPePMWZw62mfRYj/cBxAeS+2HdyaGuClr0LZzgF924JcH80K7+wuG9D3NcaYvujpH/zXROR1EblNRG4D/ga84l1ZA6hiH3z4N/jNXNi/ZsDeNi8rifioMOu3MMYMCT3t4P43YBEwAzgTWKSq3/KysAGTMh7ueAOiEuF3V8H21wbkbcNCQ7hoYiord5TSkyFXjDEmmHp8KElVX1DVb6jq11X1RS+LGnCjxsEdy2D0GbD4Jlj3+wF524JcH4eO1bKjpHpA3s8YY/qqy7AQkSoROdbJo0pEjg1UkQMizge3vgzjLoal98HKn4HHv/jn5Do3bFq543A3axpjTHB1GRaqGq+qCZ084lU1YaCKHDCRcXDTH2HGAlj+Y/jbN6C5ybO3S0+MZlJavJ1Ca4wZ9IbHGU2BFBoO1z0BF3wNCp+E5z8PDTWevV3BJB9rPjrK8bpGz97DGGP6y9OwEJF5IrJdRHaJyP2dLP+GiGwVkY0i8qaInO637FYR2ek+bvWyzk4Kh8t/CPP+yzlT6ulr4cQRT96qINdHfVMzq/eUe7J/Y4wJBM/Cwh1scCFwBTAFuFFEpnRYbT2Qr6ozcC7we8jddhTwfeAcYDbwfRFJ9qrWUzr3Hpj/JBSvg6eugMqigL9Ffk4y0eGhdijKGDOoedmymA3sUtU9qloPLMYZ4ryVqi5X1RPu5Gog0339L8AyVT2iqkeBZcA8D2s9tWmfhpv/DMeK4deXQ0lg7+4aGRbK+eNTLCyMMYOal2GRAez3my5y553KHcCrvdlWRO4SkUIRKSwt9fCP7diL4PZXQZvhqXmw952A7r5gko+Py0+wt+x4QPdrjDGB4mVYdHajhk7PRXVv1ZpP27DnPdpWVRepar6q5vt8vj4X2iOnTYM7l0HsaPj9dbB1acB2XdB6Cq21Lowxg5OXYVEEZPlNZwLFHVcSkctwBiq8WlXrerPtgEvKdq72Tj/TOUvq/V8FZLenp8SSkxJjYWGMGbS8DIs1wEQRGSsiEcACoN3PcRGZCfwSJyj8r0x7HZgrIslux/Zcd17wxYyCz/8VcufBK9+ENx8IyMV7Bbk+3t1dTm2Dd9d1GGNMX3kWFqraCNyH80d+G/C8qm4RkQdE5Gp3tZ8BccCfRGSDiCx1tz0C/AgncNYAD7jzBoeIGLjhGZh1K7z1/+Cv9zmj1/bDJyanUdPQxDf/9AEn6u2aC2PM4CLDZRC7/Px8LSwsHNg3VYUVD8LKB2HiXPjsbyEito+7Uh5fuZufvb6dSWnxLLoln+yUmMDWa4wxHYjIWlXN7249u4K7P0Tgkm/Dpx6BXX93Rq093rf7U4gIX7x4Ak/ddjYHK2u56tG3rQ/DGDNoWFgEQv7tzmGpki3OfTGO7u3zri6eNJql911AemIUtz31Po+t2GVDmBtjgs7CIlDO+KTT8X2i3AmMgx/0eVenp8Ty5y+ez6dmjOGh17bzpefWUW1jRxljgsjCIpCyz3VOrQ0Jh6c+CbuX93lXMRFh/GJBHt+9cjKvbT7EdQvf4SO7aM8YEyQWFoHmm+RcvJeUDc9+FjYt6fOuRIQvzBnH0/96DmXVdVz96Nv848OSABZrjDE9Y2HhhYQxcPsrkHUOvHAH/PPRfu3uwompLL3vQrKSY7jjd4X84s2dNDdbP4YxZuBYWHglOglufgGmXANvfBde/y40N/d5d1mjYnjh3vO5Ni+Dh5ft4O5n1lJV279rO4wxpqcsLLwUHgXzn4LZd8G7j8KfvwCN9X3eXXREKA9ffybfv2oK//jwMNcufIddh+3+3cYY71lYeC0kFK54CC79PmxeAs99Fmr7fvtyEeH2C8byzB3nUHGigWsXvsMbWw4FsGBjjDmZhcVAEIGLvgHXPg4fvQW//SRU9a+j+rzxKbz05QsZ54vlrt+v5eE3tls/hjHGMxYWAynvJrjpj1C+C35zOZTt6tfuxiRF8/zd5zH/rEx+8Y9d3Pl0IZU11o9hjAk8C4uBNvFyuO1lqD8OT86ForX92l1UeCg/mz+DH10zlVU7Srl24TvsKKkKULHGGOOwsAiGjLOci/ci4uB3n4Idb/RrdyLCLefl8Ie7zqWqtpFrF77Dq5sOBqhYY4yxsAielPFwxzJImQB/WADrn+33Ls/OGcXLX76Q3LR47n12HQ+99iFN1o9hjAkAC4tgik9zLt4bOwf++kVY9bN+XYsBcFpiFH+8+1xunJ3FYyt2c/tv11Bxou+n6xpjDHgcFiIyT0S2i8guEbm/k+VzRGSdiDSKyPwOy5rcGyK13hRpWIqMh5ueh+nXwz9+DI+eBasf79fptZFhofznp2fw0+um8+7uMq5+9B22Hez7/owxxrOwEJFQYCFwBTAFuFFEpnRYbR9wG/BcJ7uoUdU893F1J8uHj7AIuO6XMP9JiPXBa/fDw5PhlX+Dsp193u1N52Sz+K7zqG1o4tOP/ZOXPgj+bcyNMUOTly2L2cAuVd2jqvXAYuAa/xVUda+qbgT6d+xlOAgJgWmfcTq+v7AcJl8Fa38Lj+bDM5+Bncv6dIjqrNOTefnLFzJ1TAJf/sN6fvrKNhqb7Os2xvSOl2GRAez3my5y5/VUlIgUishqEbm2sxVE5C53ncLS0mF0V7mMWXDdE/D1LXDJd+HQZnh2vhMc7/2y14eoRidE8dwXzuWWc09n0ao93PrU+xw5bv0Yxpie8zIspJN5vTk1J9u9L+xNwCMiMv6knakuUtV8Vc33+Xx9rXPwihsNBf8OX9sEn/kNxIyCV/8dHp4Cr34Lynf3eFcRYSH86NppPDR/Bmv2HuWq/32bzQcqPSzeGDOceBkWRUCW33Qm0OOD5qpa7D7vAVYAMwNZ3JASFgHT58Odf4c7/wFnXAkC4NbpAAAV0ElEQVRrfgP/exY8e71z/+8eHqK6Pj+LP919Hs2qfObxf/Li+iKPizfGDAdehsUaYKKIjBWRCGAB0KOzmkQkWUQi3depwAXAVs8qHUoyz4JPL3IOUV18PxSvd/o0Fs6G938Fdd1fvX1mVhIvfflC8rKS+PofP+CBl7bSYP0YxpguiKp3F22JyJXAI0Ao8KSq/kREHgAKVXWpiJwNvAgkA7XAIVWdKiLnA7/E6fgOAR5R1d909V75+flaWFjo2WcZtBrrYetfnNNti9dBZALMvBlmfwFGjety04amZn76yjaeemcv544bxaM3zSI1LnKACjfGDAYistY95N/1el6GxUAasWHhr6gQ3nsCtrwIzU2QOw/OuRvGXeyMfHsKL64v4v4XNpESG8ETt5zFjMykASvZGBNcFhYj2bGDsPYpKHwSjpdC6iQnNM5cABGxnW6y+UAld/9+LaXVdTxw9VSuz88iJOTUAWOMGR4sLAw01jmtjNWPw8ENEJkIs26Bs++EUWNPWv3I8Xrue24d/9xdzsTRcdxTMJ6r88YQHmqjwhgzXFlYmDaqsP99eP+XsPWvziGqSVfAOfc441L5HaJqalZe3ljM4yt28+GhKjKSovnCRWO54exsoiNCg/ghjDFesLAwnTtW7Jx2u/YpOFEOvsnOIaoZ17c7RKWqrNheymMrdrFm71FGxUZw+/k5fP68HBJjwoP4AYwxgWRhYbrWUAubX3A6xA9thKgkmPV55xBV8untVl2z9wiPr9jNPz48TGxEKJ8793TuuHAsaQlRQSreGBMoFhamZ1Rh32onNLa9BChMuAyyZkP6TBiTB7GpAGw7eIwnVu7mpQ+KCQsJ4TNnZXD3nPHkpHbeaW6MGfwsLEzvVRY5h6i2/gWO7Gmbn5AB6XmQfiaMyaMoahJPrKvi+cIiGpuauWJ6OvcWjGdaRmLwajfG9ImFhemfmgo4tMk5i+rgB1C8Acp30Tq8V9xp1I2eQWFdFouLUni/LptJE3P54iUTOGfsKKSL6zqMMYOHhYUJvLoqJ0CK3QA5uAHKdoA6Q4WUk8jGphzK4s9gYt5FzDi7gJCkrC4vCDTGBJeFhRkY9cedIdQPbqDxwHqq9hQSX72bMPcWJXURyYRnziRkzJlth7KScyxAjBkkehoWYQNRjBnGImIh+xzIPocwnEG+GmuPs+qfK9m4ZiWpVR8y66OPmPDRKkK00dkmKskJDbcPhPQ8SB7r3ADKGDMoWcvCeEZVWb79MI8t382mj0s4O+YQd4yr5PzYA0Qe3giHt0KTexOmyIS2AGlpgaRMsAAxxmN2GMoMKv7XasRFhvG5c7K547wMRtd+1L4P5NBmaKpzNoqIc4JjTJ5z98AxM50WiB3CMiZgLCzMoHTytRqZ3D1nXNu1Gk0NULrdCY7iDc6w64c2tbVAopKc0GgJjzEznVN7LUCM6RMLCzOo7Ss/wS9X7eZPa51rNa6cns49p7pWo7HeOWRVvN59rIOSraBNzvLY0R0CZBbEDcPb7BrjgUERFiIyD/g5zs2Pfq2qD3ZYPgfn5kgzgAWqusRv2a3A99zJH6vq77p6LwuLoelwVS1Pvr2XZ1Z/THVdIwW5Pu69eHz312o01DiHrFrCo3i90yJpuQ4kIbP94asxMyE6eUA+kzFDSdDDQkRCgR3A5Tj3414D3KiqW/3WyQESgG8CS1vCQkRGAYVAPs6//rXAWap69FTvZ2ExtFXWNPDM6o956p2PKKuuZ1Z2EndcOI5LzvARE9HDk/bqquDgxvYB4n8levLY9q2P9BkQGe/NBzJmiBgMp87OBnap6h63oMXANfjdS1tV97rLOt4A+l+AZap6xF2+DJgH/MHDek0QJUaH86VLJnDHhWP5U+F+frlqD196bh0RYSFcNCGVy6ekcenkNHzxXdz2NTIeci5wHi1qjrb1fRSvd4Zq3/yCu1AgNbd9gJw2DcKjPf2sxgxFXoZFBrDfb7oIOKcf22YEqC4ziEWFh3LLeTncODubNXuPsmxrCW9sPcSbHx5GZBOzspOZOyWNy6ekMc4X1/0Oo5Nh/CXOo0X14fYBsutN+MD9HRISBqMntx26GjPLuZd5ZLx1opsRzcuw6OxfVk+PefVoWxG5C7gLIDs7u+eVmUEvLDSE88ancN74FP7vpybz4aEq3thSwrJth/jPVz/kP1/9kPG+WOZOPY3Lp6SRl5nU89vAxo2G3LnOA5yRd48Vtz98tXUprHu6bRsJhegk52ys1ufk7udFJ0N4jAWNGfK8DIsiIMtvOhMo7sW2F3fYdkXHlVR1EbAInD6LvhRpBj8RYXJ6ApPTE/jqZRM5UFHD390Wx69W7eHxFbvxxUdy2eQ05k5N4/zxKUSG9eKufiKQmOE8Jn/KmacKR/c6wXHsgHM4q6YCaiuc55qjcPSjtnna8Uiqn5DwrsOkq7CxQ2JmkPCygzsMp4P7UuAATgf3Taq6pZN1fwu83KGDey0wy11lHU4H95FTvZ91cI9MlScaWL79MMu2lrBi+2GO1zcRGxHKxZNGc/mUNC6ZNNr7O/upOp3rNUfbwqS2ovOAabe8Amor6bLBHRbVPjhCI9xHuBNCoeHt5/X5dVgP14+EsAhvv08zoIJ+NpRbxJU4p8aGAk+q6k9E5AGgUFWXisjZwIs4QwrVAodUdaq77b8C33F39RNVfaqr97KwMHWNTfxzdzlvbCnh79tKKK2qIyxEOGfcKOZOOY3LpqSRkTTIfqk3N0NdZc8CprHOuTixqcF9PtVrv3kt16IEUkyqczfFpNPbnpOynQEiE7MsTIaYQREWA8nCwvhrblY2FFU4HeRbDrG79DgA0zISuHzyacydmsYZp8UP//tuNDd1HiI9fd3cYduGGucmWRUfw9GPoXI/NDf6vaFAwphOgsR9nTAGQnpxiNB4zsLCGD+7S6tZtrWEZVtLWLfvKKqQmRzN5VPSmDvlNM7OSSYs1AYt7LXmJufkgIp9bQHS+rzP6e/xP8wWEg6JmW54ZLuhktMWLrE+OxlggFlYGHMKpVV1vLmthDe2lvD2rjLqG5tJignnE2eMZu6UNObk9uJCQNO1xnqn9eEfJBX72l4fL22/flh0+5ZIu+dsuwrfAxYWxvTA8bpGVu0oZdnWEt788DCVNQ1EhoVw4YRU5k51LgRMjeviQkDTP/XH/cKjpXWy133e5/Tn+ItMhORsZ/DI0HCQkA6PUL/X0n5ZSGiHdaWTbfzX72Re6/rSts+WTv/QSAiLdE4GaPfsv9xvvUFyOM7CwpheamhqZs3eI871HFtLOFBRgwhMHZPAzKxk8rKSmJmdxNjU2OHf1zFY1Bxt3xJpeT520Om81+a2R3PLtLaf33G9luXNHec30/NLwQJAQnoQNN0ETsvyxCyY+bm+lWFhYUzfqSpbDx5j2dYS1uw9wgf7K6muczpyk2LCOTPTCY6Z2cnkZSZ5f3quGRiq7qOTIGkNGL8wam507r/SWN/h2T1zrd1zT9er73zeqfbR3AiZs+HOZX36yINhbChjhiwRYeqYRKaOcYZMb2pWdh2uZv2+o2zYX8H6fRX8/M2dtPzWGueLZWZWMjOzk8jLSuKM0+Ktw3woEnE72IfQf7vmpg5npHnDWhbG9FFVbQObiipZv7+C9fuOsn5fBeXHnZs0RYeHMj0zkZlZbS2QtISoIFdszMmsZWGMx+Kjwjl/QirnT0gFnENXRUdrWOcGx4b9FTz5zkc0rHJ+kI1JjCIvO6m1BTItI5Go8MHRyWlMdywsjAkQESFrVAxZo2K4Js8ZJLm2oYmtB4+xYV9FawvklU2HAAgLcca8cloeSeRlJZOTEmOd52ZQssNQxgyw0qo6t9/D6f/4YH8Fx+udYTmSY8LJy3KCY2Z2EmdmJZEYbZ3nxjt2GMqYQcoXH8nl7j05wOk833m4yml97Ktg/f6jrNhR2tp5Pt4Xy8zsZKakJ5CbFk9uWhy++EhrgZgBZS0LYwahqtoGNhZVtjv7qqXzHJw7C+amxTExLZ7c0XHkpsUzMS2e1LgICxHTK9ayMGYIi48K54IJqVzg13leVl3PzpIqdpRUseNwNTtLqvjbxoM8V9PQul1yTLgTIGlugIx2XqfYVeimnywsjBkCRARffCS++MjWs6/ACZHSqjp2lFSzo6SKnYer2FFSzV83FFNV23bufUpsBBPcFkhriyQtnlGxNpy46RkLC2OGMBFhdEIUoxOiuHBi+xApOVbntEJKqthZUs2Ow1W8uP5A65XoAKlxEa2tj5YAyU2LIynGQsS0Z2FhzDAkIpyWGMVpiVHMyfW1zldVDlbWtgWIe0hrydqi1jOywOmEz02Lc4OkLUzszKyRy9OwEJF5wM9x7pT3a1V9sMPySOBp4CygHLhBVfeKSA6wDdjurrpaVe/xslZjRgIRYUxSNGOSorl40ujW+arKgYoadpZUtx7K2llSxfOF+znhFyJpCZFkJceQkRxNRlJ027P72oZ2H748+y8rIqHAQuByoAhYIyJLVXWr32p3AEdVdYKILAD+C7jBXbZbVfO8qs8Y00ZEyEyOITM5hkvOaAuR5mY3RFoDpJqioydY+/FR/rbxII3N7c+mTI4J9wuQGMYkRZGZ7LzOSI4mOSbcztYaorz8GTAb2KWqewBEZDFwDeAfFtcAP3BfLwEeFfs/yZhBIySk7ar0T5yR1m5ZU7NyuKqWA0drOFBRQ5H7fOBoDbtLj7NqRxk1De3vAR4dHkpGstOyyUiKdoOkrYWSlhBFaIj9CRiMvAyLDGC/33QRcM6p1lHVRhGpBFLcZWNFZD1wDPieqr7V8Q1E5C7gLoDs7OzAVm+M6VJoiJCeGE16YjSdnaSvqlScaDgpSIornNebD1RyxO/akZZ9npYQRUZyNJl+ITLG77WNpxUcXoZFZz8POl4BeKp1DgLZqlouImcBfxGRqap6rN2KqouAReBclBeAmo0xASIiJMdGkBwbwbSMxE7XOVHfSLEbJsUVtRyoONHaUlm9p5xDx2rpcKSL1LgIp98lMZr0pKjW5/TEaMYkRTE63lonXvAyLIqALL/pTKD4FOsUiUgYkAgcUeey8joAVV0rIruBXMAu0TZmGImJCGPC6HgmjI7vdHlDUzOHKmtbWyMtQdLSj7JqZ2m7DnhwWidp8ZGkJ0WTnhjFGL/nlmBJibUr3XvLy7BYA0wUkbHAAWABcFOHdZYCtwLvAvOBf6iqiogPJzSaRGQcMBHY42GtxphBKDw0pLXPpDOqyrGaRoorazhY6bRODlbWcLCiluLKGjYdqOSNrSXUNza32y4iLIT0xCgnRDq0TNITnVBJiA6zQPHjWVi4fRD3Aa/jnDr7pKpuEZEHgEJVXQr8Bvi9iOwCjuAECsAc4AERaQSagHtU9YhXtRpjhiYRITEmnMSYcCanJ3S6jqpSfry+NUAOVtRQ7LZWDlbWsnpPOSVVdTR1ON4VGxHa1jo5xSGvkXSqsA0kaIwZ8VrO7OrYMjnoThdX1lJaVXfSdglRYa3DsKTGtX/2xUfic1+Pio0gfJDeZtcGEjTGmB7yP7MLkjtdp76xmZJjbS2S4soaDlXWUlZdR1lVPVuKj1FaVdduOBV/o2Ij8MVFkhof0RoinQVMckzEoOygt7AwxpgeiAjruv+kRU19E2XVdRyuqqOsuo5Sv+eW12v3HaW0qo7ahuaTtg8RSInzC5AOAeMfNEkDeJGjhYUxxgRQdERoj0JFVTle39QuTDqGSmlVHbsPV1NaVUd908nBEh4qpMZFkp8ziv+9caZXHwmwsDDGmKAQEeIiw4iLDGNsamyX67ac9VXaMVSq6yirqsMX7/39SiwsjDFmkPM/62vC6Lig1DA4u+eNMcYMKhYWxhhjumVhYYwxplsWFsYYY7plYWGMMaZbFhbGGGO6ZWFhjDGmWxYWxhhjujVsRp0VkVLg437sIhUoC1A5Q519F+3Z99GefR9thsN3cbqq+rpbadiERX+JSGFPhukdCey7aM++j/bs+2gzkr4LOwxljDGmWxYWxhhjumVh0WZRsAsYROy7aM++j/bs+2gzYr4L67MwxhjTLWtZGGOM6ZaFhTHGmG6N+LAQkXkisl1EdonI/cGuJ5hEJEtElovINhHZIiJfDXZNwSYioSKyXkReDnYtwSYiSSKyREQ+dP8fOS/YNQWTiHzd/XeyWUT+ICJRwa7JSyM6LEQkFFgIXAFMAW4UkSnBrSqoGoH/o6qTgXOBL43w7wPgq8C2YBcxSPwceE1VzwDOZAR/LyKSAXwFyFfVaUAosCC4VXlrRIcFMBvYpap7VLUeWAxcE+SagkZVD6rqOvd1Fc4fg4zgVhU8IpIJfBL4dbBrCTYRSQDmAL8BUNV6Va0IblVBFwZEi0gYEAMUB7keT430sMgA9vtNFzGC/zj6E5EcYCbwXnArCapHgH8HmoNdyCAwDigFnnIPy/1aRGKDXVSwqOoB4L+BfcBBoFJV3whuVd4a6WEhncwb8ecSi0gc8ALwNVU9Fux6gkFEPgUcVtW1wa5lkAgDZgGPq+pM4DgwYvv4RCQZ5yjEWGAMECsiNwe3Km+N9LAoArL8pjMZ5k3J7ohIOE5QPKuqfw52PUF0AXC1iOzFOTz5CRF5JrglBVURUKSqLS3NJTjhMVJdBnykqqWq2gD8GTg/yDV5aqSHxRpgooiMFZEInA6qpUGuKWhERHCOSW9T1YeDXU8wqeq3VTVTVXNw/r/4h6oO61+OXVHVQ8B+EZnkzroU2BrEkoJtH3CuiMS4/24uZZh3+IcFu4BgUtVGEbkPeB3nbIYnVXVLkMsKpguAW4BNIrLBnfcdVX0liDWZwePLwLPuD6s9wO1BridoVPU9EVkCrMM5i3A9w3zoDxvuwxhjTLdG+mEoY4wxPWBhYYwxplsWFsYYY7plYWGMMaZbFhbGGGO6ZWFhTC+ISJOIbPB7BOwqZhHJEZHNgdqfMYE0oq+zMKYPalQ1L9hFGDPQrGVhTACIyF4R+S8Red99THDnny4ib4rIRvc5252fJiIvisgH7qNlqIhQEfmVe5+EN0QkOmgfyhg/FhbG9E50h8NQN/gtO6aqs4FHcUasxX39tKrOAJ4FfuHO/wWwUlXPxBljqWXkgInAQlWdClQAn/H48xjTI3YFtzG9ICLVqhrXyfy9wCdUdY87GOMhVU0RkTIgXVUb3PkHVTVVREqBTFWt89tHDrBMVSe6098CwlX1x95/MmO6Zi0LYwJHT/H6VOt0ps7vdRPWr2gGCQsLYwLnBr/nd93X/6TtdpufA952X78J3Aut9/lOGKgijekL+9ViTO9E+43IC849qVtOn40UkfdwfoTd6M77CvCkiPwbzp3mWkZq/SqwSETuwGlB3ItzxzVjBiXrszAmANw+i3xVLQt2LcZ4wQ5DGWOM6Za1LIwxxnTLWhbGGGO6ZWFhjDGmWxYWxhhjumVhYYwxplsWFsYYY7r1/wFZ8KzjFqvBugAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_train_history(train_history,'loss','val_loss') #畫出loss誤差執行結果"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s 16us/step\n",
      "\n",
      "accuracy= 0.9761\n"
     ]
    }
   ],
   "source": [
    "#評估模型準確率\n",
    "scores = model.evaluate(x_Test_normalize,y_Test_OneHot) #使用model.evaluate進行評估模型準確率，評估後的準確率會儲存在scores\n",
    "print()\n",
    "print('accuracy=',scores[1]) #顯示準確率"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "prediction=model.predict_classes(x_Test) #使用model.predict_classes輸入操數x_Test(測試資料的數字影像)進行預測，並將結果儲存於prediction變數"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "#建立plot_images_labels_prediction函數\n",
    "def plot_images_labels_prediction(images, labels, prediction, idx, num=10): #定義函數，並傳入5個參數：images(數字影像)、labels(真實值)、prediction(預測結果)、idx(開始顯示的資料index)及\n",
    "    fig = plt.gcf() #設定圖形大小                                                                                                                                          num(資料筆數)\n",
    "    fig.set_size_inches(12, 14)\n",
    "    if num>25: num=25 #如果資料筆數超過25則設為25，避免出錯\n",
    "    for i in range(0, num): #定義for迴圈\n",
    "        ax=plt.subplot(5, 5, 1+i) #建立subfraph子圖形為5行5列\n",
    "        ax.imshow(images[idx], cmap='binary') #畫出subgraph子圖形\n",
    "        title= \"label=\" + str(labels[idx]) #設定子圖形title，顯示標籤欄位\n",
    "        if len(prediction)>0: #如果有傳入預測結果\n",
    "            title=title+\",prediction=\"+str(prediction[idx]) #標題title加入預測結果\n",
    "            \n",
    "        ax.set_title(title, fontsize=10) #設定子圖形的標題title與大小\n",
    "        ax.set_xticks([]); ax.set_yticks([]) #設定為不顯示刻度\n",
    "        idx+=1 #讀取下一筆資料\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x1008 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_images_labels_prediction(x_test_image,y_test_label,prediction, idx=716) #顯示從716~725等10筆資料"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>predict</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>label</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>971</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>1124</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>995</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>961</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>15</td>\n",
       "      <td>1</td>\n",
       "      <td>854</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>939</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>13</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>989</td>\n",
       "      <td>6</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>956</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>11</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>971</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "predict    0     1     2    3    4    5    6    7    8    9\n",
       "label                                                      \n",
       "0        971     0     1    1    1    0    2    1    3    0\n",
       "1          0  1124     4    0    0    1    2    0    4    0\n",
       "2          5     0  1010    1    1    0    2    4    9    0\n",
       "3          0     0     3  995    0    0    0    3    5    4\n",
       "4          1     0     5    1  961    0    3    0    3    8\n",
       "5          3     0     0   15    1  854    7    1    8    3\n",
       "6          5     3     2    1    3    3  939    0    2    0\n",
       "7          0     5    13    6    1    0    0  989    6    8\n",
       "8          5     0     3    6    0    0    1    2  956    1\n",
       "9          3     5     0   11    8    2    1    4    4  971"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd #匯入pandas模組\n",
    "pd.crosstab(y_test_label,prediction,rownames=['label'],colnames=['predict']) \n",
    "#y_test_label為測試資料影像的真實值，prediction為測試資料影像的預測結果，rownames為設定行的名稱，colnames則為設定列的名稱"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>label</th>\n",
       "      <th>predict</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   label  predict\n",
       "0      7        7\n",
       "1      2        2\n",
       "2      1        1\n",
       "3      0        0\n",
       "4      4        4"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.DataFrame({'label':y_test_label,'predict':prediction}) #建立dataframe(label為真實值，prediction為預測值)\n",
    "df[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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>label</th>\n",
       "      <th>predict</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>340</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1003</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1393</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2035</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2526</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2597</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2810</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3117</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4271</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4355</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4360</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5937</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5972</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6028</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6043</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      label  predict\n",
       "340       5        3\n",
       "1003      5        3\n",
       "1393      5        3\n",
       "2035      5        3\n",
       "2526      5        3\n",
       "2597      5        3\n",
       "2810      5        3\n",
       "3117      5        3\n",
       "4271      5        3\n",
       "4355      5        3\n",
       "4360      5        3\n",
       "5937      5        3\n",
       "5972      5        3\n",
       "6028      5        3\n",
       "6043      5        3"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[(df.label==5)&(df.predict==3)] #顯示真實值是5但預測結果是3的資料(顯示前10筆)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x1008 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_images_labels_prediction(x_test_image,y_test_label,prediction,idx=716,num=1) #查看第716筆資料"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.5.5"
  },
  "widgets": {
   "state": {},
   "version": "1.1.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}