12.2 - Keras.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Apresentando Keras\n",
    "\n",
    "Certifique-se de estar usando tensorflow 1.9 ou mais recente!\n",
    "\n",
    "Keras é uma API para TensorFlow que facilita muito as coisas. Não é apenas mais fácil de usar, é mais fácil de ajustar.\n",
    "\n",
    "Vamos configurar a mesma rede neural profunda que configuramos com o TensorFlow para aprender com o conjunto de dados MNIST.\n",
    "\n",
    "Primeiro nós vamos importar todas as coisas que precisamos, que irá inicializar o Keras:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/leo/anaconda3/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n"
     ]
    }
   ],
   "source": [
    "from tensorflow import keras\n",
    "from tensorflow.keras.datasets import mnist\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Dropout\n",
    "from tensorflow.keras.optimizers import RMSprop\n",
    "import matplotlib.pyplot as plt\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Vamos carregar o conjunto de dados MNIST. No Keras, é um pouco diferente - há 60 mil amostras de treinamento e 10 mil amostras de teste. Nenhuma amostra de \"validação\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "(mnist_train_images, mnist_train_labels), (mnist_test_images, mnist_test_labels) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Precisamos converter explicitamente os dados no formato esperado pelo Keras / TensorFlow. Nós dividimos os dados da imagem em 255 para normalizá-los em um intervalo de 0-1, após convertê-los em valores de ponto flutuante."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_images = mnist_train_images.reshape(60000, 784)\n",
    "test_images = mnist_test_images.reshape(10000, 784)\n",
    "train_images = train_images.astype('float32')\n",
    "test_images = test_images.astype('float32')\n",
    "train_images /= 255\n",
    "test_images /= 255"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Agora vamos converter os rótulos 0-9 em um formato \"one-hot\", como fizemos para o TensorFlow."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_labels = keras.utils.to_categorical(mnist_train_labels, 10)\n",
    "test_labels = keras.utils.to_categorical(mnist_test_labels, 10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Vamos dar uma olhada em uma das imagens de treinamento só para ter certeza de que está tudo bem:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def display_sample(num):\n",
    "    #Print the one-hot array of this sample's label \n",
    "    print(train_labels[num])  \n",
    "    #Print the label converted back to a number\n",
    "    label = train_labels[num].argmax(axis=0)\n",
    "    #Reshape the 768 values to a 28x28 image\n",
    "    image = train_images[num].reshape([28,28])\n",
    "    plt.title('Sample: %d  Label: %d' % (num, label))\n",
    "    plt.imshow(image, cmap=plt.get_cmap('gray_r'))\n",
    "    plt.show()\n",
    "    \n",
    "display_sample(1234)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Aqui é onde as coisas ficam empolgantes. Todo esse código que escrevemos no Tensorflow criando espaços reservados, variáveis e definindo um monte de álgebra linear para cada camada em nossa rede neural? Nada disso é necessário com Keras!\n",
    "\n",
    "Podemos configurar as mesmas camadas como esta. A camada de entrada de 784 recursos alimenta uma camada ReLU de 512 nós, que então entra em 10 nós com softmax aplicado. Não poderia ser mais simples:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Dense(512, activation='relu', input_shape=(784,)))\n",
    "model.add(Dense(10, activation='softmax'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Podemos até obter uma boa descrição do modelo resultante:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 512)               401920    \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 10)                5130      \n",
      "=================================================================\n",
      "Total params: 407,050\n",
      "Trainable params: 407,050\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Configurar nosso otimizador e função de perda é tão simples. Vamos usar o otimizador RMSProp aqui. Outras opções incluem Adagrad, SGD, Adam, Adamax e Nadam. Veja https://keras.io/optimizers/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer=RMSprop(),\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Treinar nosso modelo é também apenas uma linha de código com Keras. Aqui vamos fazer 10 épocas com um tamanho de lote de 100. Keras é mais lento, e se não estivermos correndo em cima de um Tensorflow acelerado por GPU, isso pode levar um bom tempo (é por isso que eu limito isso apenas 10 épocas.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 60000 samples, validate on 10000 samples\n",
      "Epoch 1/10\n",
      " - 4s - loss: 0.2406 - acc: 0.9297 - val_loss: 0.1252 - val_acc: 0.9620\n",
      "Epoch 2/10\n",
      " - 4s - loss: 0.0978 - acc: 0.9712 - val_loss: 0.0846 - val_acc: 0.9740\n",
      "Epoch 3/10\n",
      " - 4s - loss: 0.0646 - acc: 0.9802 - val_loss: 0.0880 - val_acc: 0.9728\n",
      "Epoch 4/10\n",
      " - 4s - loss: 0.0465 - acc: 0.9860 - val_loss: 0.0758 - val_acc: 0.9776\n",
      "Epoch 5/10\n",
      " - 4s - loss: 0.0353 - acc: 0.9891 - val_loss: 0.0805 - val_acc: 0.9758\n",
      "Epoch 6/10\n",
      " - 4s - loss: 0.0274 - acc: 0.9918 - val_loss: 0.0627 - val_acc: 0.9823\n",
      "Epoch 7/10\n",
      " - 4s - loss: 0.0209 - acc: 0.9939 - val_loss: 0.0675 - val_acc: 0.9820\n",
      "Epoch 8/10\n",
      " - 4s - loss: 0.0160 - acc: 0.9953 - val_loss: 0.0761 - val_acc: 0.9818\n",
      "Epoch 9/10\n",
      " - 4s - loss: 0.0131 - acc: 0.9961 - val_loss: 0.0710 - val_acc: 0.9819\n",
      "Epoch 10/10\n",
      " - 5s - loss: 0.0098 - acc: 0.9970 - val_loss: 0.0781 - val_acc: 0.9824\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_images, train_labels,\n",
    "                    batch_size=100,\n",
    "                    epochs=10,\n",
    "                    verbose=2,\n",
    "                    validation_data=(test_images, test_labels))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Mas, mesmo com apenas 10 épocas, superamos consideravelmente nossa versão do Tensorflow!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.0780573598645111\n",
      "Test accuracy: 0.9824\n"
     ]
    }
   ],
   "source": [
    "score = model.evaluate(test_images, test_labels, verbose=0)\n",
    "print('Test loss:', score[0])\n",
    "print('Test accuracy:', score[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Como antes, vamos visualizar os erros. Como esse modelo é muito melhor, teremos que pesquisar mais profundamente para encontrar erros a serem observados."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAEmxJREFUeJzt3X2wVPV9x/H3R1RUwBSGK6CIqNGZZGwFeqttbRON0RFbq2ZiAn8gNImkaUy1xqHWZgZsasMwMZo+xBQVxSYh2kYTjPisGWuasd5QG4nWxxIec7mIGHyICfrtH3uus17vnt27e3bPwu/zmtm5e8/3PHzvgc+e3XN296eIwMzSs0/ZDZhZORx+s0Q5/GaJcvjNEuXwmyXK4TdLlMNfMknTJYWkfbPf75I0v4n1TJP0iqRRxXfZeZIWSHqk08umxOFvgKT1kl7PwtUv6UZJY9uxrYiYHRErG+zpw1XLbYiIsRHxZjv6Gmb7F0n6P0mvSnpK0rE15lsi6Rud6KlVkuZnD8SfKruXTnD4G3dWRIwFZgG/A3xh6Ayq2Ov3aRaOTwJ/BIwF/hjYXmpTLZI0Hvhr4Kdl99Ipe/1/1KJFxGbgLuA4AEk/kHSlpB8CrwFHSXqPpBskbZW0WdLfDT4dlzRK0pclbZf0ApUAvS1b36eqfr8gO7LukvSkpFmS/hWYBtyRPRtZNMzLh0MlrZa0Q9Jzki6oWucSSbdKujlb708l9Tby92cPbouBv4yIJ6Pi+YjYMdJ9KekySc9X/W3nvnsW/aOklyX9r6RTqwo193GTvgT8A3v4g9hIOPwjJOlw4Ezgv6smzwMWAuOAnwErgd3Ae4GZwOnAYKAvoHKknAn0Ah/N2dZ5wBLgfOBg4E+AFyNiHrCB7NlIRCwbZvFVwCbg0Gwbf18dnmxd3wZ+A1gN/FPVdr8m6Ws12pqa3Y6TtDF76n9Fk894ngf+EHgPcAXwDUlTquonAi8AE6k84NwmaUJWy9vH7yDp+5Iuq9WEpBOo/Ft8vYm/Yc8VEb7VuQHrgVeAnVTC/TXgwKz2A+Bvq+adBLwxWM+mzQUeyu4/CPxZVe10IIB9q9b3qez+PcBFOT19uOr36YPrAQ4H3gTGVdW/BNyU3V8C3F9Vez/weoP74vez7dxJ5YFjOvAMcEGN+ZcA32hw3Y8DZ2f3FwBbAFXV/4vKA229fbwAeKTBbY4C+oDfG7r/9/bbvnUfHWzQORFxf43axqr7RwD7AVslDU7bp2qeQ4fM/7OcbR5O5eg4UocCOyJi15DtVD+1/3nV/deAAyTtGxG766z79eznsojYCeyU9C9Ung1dN5ImJZ0PXELlAQQq5w8mVs2yObJEVv0Nh1J/H4/EnwM/iYgfNbHsHs3hL0b1f9CNVI5KE2sEaSuVUA+alrPejcDRDWxzqC3ABEnjqh4ApgGbc5Zp1NPAr+psvy5JR1B5sDgV+FFEvCnpcUBVsx0mSVUPANOovESpt49H4lTgg5LOzH6fAMyUNCMiLmxx3V3Nr/kLFhFbgXuBqyQdLGkfSUdL+mA2y63AX0iamp1hrvlaFLgeuFTSb2dXEt6bhQagHziqRg8bgf8EviTpAEm/ReXs/DcL+PteA24BFkkaJ2kqlfMY389ZbJ+sj8HbaGAMlQeQAQBJf0p2ErXKIVT21X7Z+Y/3AWsa2McjsSBb74zs1kfl/MPfNLGuPYrD3x7nA/sDTwIvAf8ODJ7Iuo7Ka/n/AdYCt9VaSUT8G3Al8C1gF/BdKkcmqLyG/4KknZIuHWbxuVSeTm8BbgcWR8R9jTQv6euS8k5+XUjlHMgW4EdZfyty5p9L5eXC4O35iHgSuCpbvh/4TeCHQ5Z7FDiGyhn4K4GPRsSLWS1vHw/9e+6SdPlwtYjYGRE/H7xReVbzi4h4Oefv2SvonS+pzCwVPvKbJcrhN0uUw2+WKIffLFEdvc4/ceLEmD59eic3aZaU9evXs337dtWfs8XwSzoD+CqVt0heHxFL8+afPn06fX19rWzSzHL09jb0+Syghaf92Seo/hmYTeW94XMlvb/Z9ZlZZ7Xymv8E4LmIeCEifkXlE2JnF9OWmbVbK+E/jHd+kGJTNu0dJC2U1Cepb2BgoIXNmVmRWgn/cCcV3vV2wYhYHhG9EdHb09PTwubMrEithH8T7/x02lQq7/U2sz1AK+F/DDhG0pGS9gfmUPm4pZntAZq+1BcRuyVdSOUTaqOAFRGRzJcfmu3pWrrOHxFrgDUF9WJmHeS395olyuE3S5TDb5Yoh98sUQ6/WaIcfrNE+Xv7rSWvvfZabn3OnDk1a0cdNew3j7/tmmuuaaona4yP/GaJcvjNEuXwmyXK4TdLlMNvliiH3yxRvtRnLdm0aVNu/Y477qhZO/DAA3OXXbx4cW59/PjxuXXL5yO/WaIcfrNEOfxmiXL4zRLl8JslyuE3S5TDb5YoX+e30kyaNCm3vv/++3eokzT5yG+WKIffLFEOv1miHH6zRDn8Zoly+M0S5fCbJcrX+a00s2fPzq2PGTOmQ52kqaXwS1oP7ALeBHZHRG8RTZlZ+xVx5D8lIrYXsB4z6yC/5jdLVKvhD+BeST+WtHC4GSQtlNQnqW9gYKDFzZlZUVoN/0kRMQuYDXxW0geGzhARyyOiNyJ6e3p6WtycmRWlpfBHxJbs5zbgduCEIpoys/ZrOvySxkgaN3gfOB1YV1RjZtZerZztnwTcLmlwPd+KiLsL6cr2GNdee21uffTo0TVrF198cdHt2Ag0Hf6IeAE4vsBezKyDfKnPLFEOv1miHH6zRDn8Zoly+M0S5Y/0Wq4NGzbk1m+66abc+kEHHVSzduyxxzbTkhXER36zRDn8Zoly+M0S5fCbJcrhN0uUw2+WKIffLFG+zm+57r///tz6zp07c+tLly4tsh0rkI/8Zoly+M0S5fCbJcrhN0uUw2+WKIffLFEOv1mifJ0/cdu2bcutL1u2LLc+efLk3PqCBQtG2pJ1iI/8Zoly+M0S5fCbJcrhN0uUw2+WKIffLFEOv1mifJ0/cXfddVdu/emnn86tn3feebn1SZMm1ay9/vrrucvu3r07tz5u3LjcuuWre+SXtELSNknrqqZNkHSfpGezn+Pb26aZFa2Rp/03AWcMmXYZ8EBEHAM8kP1uZnuQuuGPiIeBHUMmnw2szO6vBM4puC8za7NmT/hNioitANnPQ2rNKGmhpD5JfQMDA01uzsyK1vaz/RGxPCJ6I6K3p6en3ZszswY1G/5+SVMAsp/5Hw0zs67TbPhXA/Oz+/OB7xXTjpl1St3r/JJWAScDEyVtAhYDS4FbJX0S2ADkX+y10rz66qu59Ztvvrml9S9atCi3nnetfs6cObnL9vf359bXrFmTW58wYUJuPXV1wx8Rc2uUTi24FzPrIL+91yxRDr9Zohx+s0Q5/GaJcvjNEuWP9O7lrr766tz6gw8+mFs/5ZRTcuu9vb259XvvvbdmbfXq1bnL1rNx48bcui/15fOR3yxRDr9Zohx+s0Q5/GaJcvjNEuXwmyXK4TdLlK/z7wXWrVtXs7Z8+fKW1v2JT3wit759+/bc+uc+97mmtz1lypTcer3hwS2fj/xmiXL4zRLl8JslyuE3S5TDb5Yoh98sUQ6/WaJ8nb8L/PrXv86t33333bn1z3zmMzVrmzdvbqqnQR/5yEdy6/fcc09u/Zlnnml62/vum//fs94Q3m+88UbN2ujRo5vqaW/iI79Zohx+s0Q5/GaJcvjNEuXwmyXK4TdLlMNvlihf5++Al19+Obd+7rnn5tYfeuihItsZkTFjxpS27Xrfyz916tTc+rRp02rWrr/++txlTzvttNz63qDukV/SCknbJK2rmrZE0mZJj2e3M9vbppkVrZGn/TcBZwwz/eqImJHd1hTblpm1W93wR8TDwI4O9GJmHdTKCb8LJf0ke1kwvtZMkhZK6pPUNzAw0MLmzKxIzYb/WuBoYAawFbiq1owRsTwieiOit6enp8nNmVnRmgp/RPRHxJsR8RZwHXBCsW2ZWbs1FX5J1d+pfC5Q+7ujzawr1b3OL2kVcDIwUdImYDFwsqQZQADrgU+3sceuV+86/qWXXppbb/U6/tixY5ve9sEHH5xbX7VqVW79sccey62XKe/7ANauXZu7bArX+euGPyLmDjP5hjb0YmYd5Lf3miXK4TdLlMNvliiH3yxRDr9ZovyR3gblfb12vctp9T4+2qorrriiZu2SSy7JXfaXv/xlbv2LX/xiUz0NklSzdvzxx+cu+6EPfSi3ftZZZ+XWZ82aVbNW7xJnCnzkN0uUw2+WKIffLFEOv1miHH6zRDn8Zoly+M0S5ev8DXr22Wdr1tp9HX/evHm59Ysuuqjpdd9yyy259ZdeeqnpdQOcccZw3/1asWaNv/e1TD7ymyXK4TdLlMNvliiH3yxRDr9Zohx+s0Q5/GaJ8nX+Bi1btqxt6z7yyCNz6/U+Uz9q1Kimt93qEGrnn39+bv3GG29saf3WPj7ymyXK4TdLlMNvliiH3yxRDr9Zohx+s0Q5/GaJamSI7sOBm4HJwFvA8oj4qqQJwC3AdCrDdH8sIlr78HeJXnzxxdx6K8Nojx49OrdebxjsI444oult17N58+bc+gEHHJBb//jHP55b32cfH1+6VSP/MruBz0fE+4DfBT4r6f3AZcADEXEM8ED2u5ntIeqGPyK2RsTa7P4u4CngMOBsYGU220rgnHY1aWbFG9FzMknTgZnAo8CkiNgKlQcI4JCimzOz9mk4/JLGAt8BLo6IX4xguYWS+iT1tfo+cjMrTkPhl7QfleB/MyJuyyb3S5qS1acA24ZbNiKWR0RvRPT29PQU0bOZFaBu+FUZZvUG4KmI+EpVaTUwP7s/H/he8e2ZWbs08pHek4B5wBOSHs+mXQ4sBW6V9ElgA3Bee1rsjLwhuKH+UNZ57rzzztz6iSee2PS6W7Vo0aLcer2P7M6cObPIdqyD6oY/Ih4Bag2yfmqx7ZhZp/gdGGaJcvjNEuXwmyXK4TdLlMNvliiH3yxR/uruzOTJk3Pr/f39Heqks6ZMmdJS3fZcPvKbJcrhN0uUw2+WKIffLFEOv1miHH6zRDn8Zoly+M0S5fCbJcrhN0uUw2+WKIffLFEOv1miHH6zRDn8Zoly+M0S5fCbJcrhN0uUw2+WKIffLFEOv1miHH6zRDn8ZomqG35Jh0t6SNJTkn4q6aJs+hJJmyU9nt3ObH+7ZlaURgbt2A18PiLWShoH/FjSfVnt6oj4cvvaM7N2qRv+iNgKbM3u75L0FHBYuxszs/Ya0Wt+SdOBmcCj2aQLJf1E0gpJ42sss1BSn6S+gYGBlpo1s+I0HH5JY4HvABdHxC+Aa4GjgRlUnhlcNdxyEbE8Inojorenp6eAls2sCA2FX9J+VIL/zYi4DSAi+iPizYh4C7gOOKF9bZpZ0Ro52y/gBuCpiPhK1fTq4VvPBdYV356ZtUsjZ/tPAuYBT0h6PJt2OTBX0gwggPXAp9vSoZm1RSNn+x8BNExpTfHtmFmn+B1+Zoly+M0S5fCbJcrhN0uUw2+WKIffLFEOv1miHH6zRDn8Zoly+M0S5fCbJcrhN0uUw2+WKIffLFGKiM5tTBoAflY1aSKwvWMNjEy39tatfYF7a1aRvR0REQ19X15Hw/+ujUt9EdFbWgM5urW3bu0L3FuzyurNT/vNEuXwmyWq7PAvL3n7ebq1t27tC9xbs0rprdTX/GZWnrKP/GZWEoffLFGlhF/SGZKelvScpMvK6KEWSeslPZENO95Xci8rJG2TtK5q2gRJ90l6Nvs57BiJJfXWFcO25wwrX+q+67bh7jv+ml/SKOAZ4DRgE/AYMDcinuxoIzVIWg/0RkTpbwiR9AHgFeDmiDgum7YM2BERS7MHzvER8Vdd0tsS4JWyh23PRpOaUj2sPHAOsIAS911OXx+jhP1WxpH/BOC5iHghIn4FfBs4u4Q+ul5EPAzsGDL5bGBldn8llf88HVejt64QEVsjYm12fxcwOKx8qfsup69SlBH+w4CNVb9vosQdMIwA7pX0Y0kLy25mGJMiYitU/jMBh5Tcz1B1h23vpCHDynfNvmtmuPuilRH+4Yb+6qbrjSdFxCxgNvDZ7OmtNaahYds7ZZhh5btCs8PdF62M8G8CDq/6fSqwpYQ+hhURW7Kf24Db6b6hx/sHR0jOfm4ruZ+3ddOw7cMNK08X7LtuGu6+jPA/Bhwj6UhJ+wNzgNUl9PEuksZkJ2KQNAY4ne4benw1MD+7Px/4Xom9vEO3DNtea1h5St533TbcfSnv8MsuZVwDjAJWRMSVHW9iGJKOonK0h8oIxt8qszdJq4CTqXzksx9YDHwXuBWYBmwAzouIjp94q9HbyVSeur49bPvga+wO9/YHwH8ATwBvZZMvp/L6urR9l9PXXErYb357r1mi/A4/s0Q5/GaJcvjNEuXwmyXK4TdLlMNvliiH3yxR/w9rfncoTBZz9wAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for x in range(1000):\n",
    "    test_image = test_images[x,:].reshape(1,784)\n",
    "    predicted_cat = model.predict(test_image).argmax()\n",
    "    label = test_labels[x].argmax()\n",
    "    if (predicted_cat != label):\n",
    "        plt.title('Prediction: %d Label: %d' % (predicted_cat, label))\n",
    "        plt.imshow(test_image.reshape([28,28]), cmap=plt.get_cmap('gray_r'))\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Você pode quais imagens são difíceis de ser reconhecidas pelo modelo. Observe que são imagens que humanos também teriam problemas!\n",
    "\n",
    "## Exercício\n",
    "\n",
    "Como antes, veja se você pode melhorar os resultados! A execução de mais épocas ajuda consideravelmente? Que tal tentar diferentes otimizadores?\n",
    "\n",
    "Você também pode aproveitar a facilidade de uso do Keras para experimentar topologias diferentes rapidamente. Keras inclui um exemplo MNIST, onde eles adicionam uma camada adicional, e usam Dropout em cada etapa para evitar overfitting, assim:\n",
    "\n",
    "`\n",
    "model = Sequential()\n",
    "model.add(Dense(512, activation='relu', input_shape=(784,)))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Dense(512, activation='relu'))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Dense(10, activation='softmax'))\n",
    "`\n",
    "\n",
    "Tente adaptar isso ao nosso código acima e veja se ele faz diferença ou não."
   ]
  },
  {
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}