02-Keras-CNN-CIFAR-10.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CIFAR-10 Multiple Classes\n",
    "\n",
    "Let's go over another example of using Keras and building out CNNs. This time will use another famous data set, the CIFAR-10 dataset which consists of 10 different image types."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "-----"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# The Data\n",
    "\n",
    "CIFAR-10 is a dataset of 50,000 32x32 color training images, labeled over 10 categories, and 10,000 test images."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.datasets import cifar10\n",
    "\n",
    "(x_train, y_train), (x_test, y_test) = cifar10.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50000, 32, 32, 3)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(32, 32, 3)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x295e6e1cba8>"
      ]
     },
     "execution_count": 5,
     "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": [
    "# FROG\n",
    "plt.imshow(x_train[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x295e6ebb0b8>"
      ]
     },
     "execution_count": 6,
     "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": [
    "# HORSE\n",
    "plt.imshow(x_train[12])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# PreProcessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[[ 59,  62,  63],\n",
       "        [ 43,  46,  45],\n",
       "        [ 50,  48,  43],\n",
       "        ...,\n",
       "        [158, 132, 108],\n",
       "        [152, 125, 102],\n",
       "        [148, 124, 103]],\n",
       "\n",
       "       [[ 16,  20,  20],\n",
       "        [  0,   0,   0],\n",
       "        [ 18,   8,   0],\n",
       "        ...,\n",
       "        [123,  88,  55],\n",
       "        [119,  83,  50],\n",
       "        [122,  87,  57]],\n",
       "\n",
       "       [[ 25,  24,  21],\n",
       "        [ 16,   7,   0],\n",
       "        [ 49,  27,   8],\n",
       "        ...,\n",
       "        [118,  84,  50],\n",
       "        [120,  84,  50],\n",
       "        [109,  73,  42]],\n",
       "\n",
       "       ...,\n",
       "\n",
       "       [[208, 170,  96],\n",
       "        [201, 153,  34],\n",
       "        [198, 161,  26],\n",
       "        ...,\n",
       "        [160, 133,  70],\n",
       "        [ 56,  31,   7],\n",
       "        [ 53,  34,  20]],\n",
       "\n",
       "       [[180, 139,  96],\n",
       "        [173, 123,  42],\n",
       "        [186, 144,  30],\n",
       "        ...,\n",
       "        [184, 148,  94],\n",
       "        [ 97,  62,  34],\n",
       "        [ 83,  53,  34]],\n",
       "\n",
       "       [[177, 144, 116],\n",
       "        [168, 129,  94],\n",
       "        [179, 142,  87],\n",
       "        ...,\n",
       "        [216, 184, 140],\n",
       "        [151, 118,  84],\n",
       "        [123,  92,  72]]], dtype=uint8)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(32, 32, 3)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "255"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train.max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train = x_train/225"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_test = x_test/255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50000, 32, 32, 3)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 32, 32, 3)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.utils import to_categorical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50000, 1)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([6], dtype=uint8)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_cat_train = to_categorical(y_train,10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50000, 10)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_cat_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0., 0., 0., 0., 0., 0., 1., 0., 0., 0.], dtype=float32)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_cat_train[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_cat_test = to_categorical(y_test,10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "----------\n",
    "# Building the Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Conv2D, MaxPool2D, Flatten"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "\n",
    "## FIRST SET OF LAYERS\n",
    "\n",
    "# CONVOLUTIONAL LAYER\n",
    "model.add(Conv2D(filters=32, kernel_size=(4,4),input_shape=(32, 32, 3), activation='relu',))\n",
    "# POOLING LAYER\n",
    "model.add(MaxPool2D(pool_size=(2, 2)))\n",
    "\n",
    "## SECOND SET OF LAYERS\n",
    "\n",
    "# CONVOLUTIONAL LAYER\n",
    "model.add(Conv2D(filters=32, kernel_size=(4,4),input_shape=(32, 32, 3), activation='relu',))\n",
    "# POOLING LAYER\n",
    "model.add(MaxPool2D(pool_size=(2, 2)))\n",
    "\n",
    "# FLATTEN IMAGES FROM 28 by 28 to 764 BEFORE FINAL LAYER\n",
    "model.add(Flatten())\n",
    "\n",
    "# 256 NEURONS IN DENSE HIDDEN LAYER (YOU CAN CHANGE THIS NUMBER OF NEURONS)\n",
    "model.add(Dense(256, activation='relu'))\n",
    "\n",
    "# LAST LAYER IS THE CLASSIFIER, THUS 10 POSSIBLE CLASSES\n",
    "model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer='rmsprop',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 29, 29, 32)        1568      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 14, 14, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 11, 11, 32)        16416     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 5, 5, 32)          0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 800)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 256)               205056    \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                2570      \n",
      "=================================================================\n",
      "Total params: 225,610\n",
      "Trainable params: 225,610\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "50000/50000 [==============================] - 5s 109us/step - loss: 0.8937 - acc: 0.6919\n",
      "Epoch 2/10\n",
      "50000/50000 [==============================] - 5s 108us/step - loss: 0.8094 - acc: 0.7249\n",
      "Epoch 3/10\n",
      "50000/50000 [==============================] - 5s 109us/step - loss: 0.7420 - acc: 0.7461\n",
      "Epoch 4/10\n",
      "50000/50000 [==============================] - 5s 109us/step - loss: 0.6902 - acc: 0.7656\n",
      "Epoch 5/10\n",
      "50000/50000 [==============================] - 5s 109us/step - loss: 0.6466 - acc: 0.7780\n",
      "Epoch 6/10\n",
      "50000/50000 [==============================] - 5s 110us/step - loss: 0.6105 - acc: 0.7958\n",
      "Epoch 7/10\n",
      "50000/50000 [==============================] - 5s 110us/step - loss: 0.5891 - acc: 0.8020\n",
      "Epoch 8/10\n",
      "50000/50000 [==============================] - 5s 109us/step - loss: 0.5631 - acc: 0.8124\n",
      "Epoch 9/10\n",
      "50000/50000 [==============================] - 6s 110us/step - loss: 0.5416 - acc: 0.8203\n",
      "Epoch 10/10\n",
      "50000/50000 [==============================] - 5s 109us/step - loss: 0.5252 - acc: 0.8275\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x295dce78dd8>"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(x_train,y_cat_train,verbose=1,epochs=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Careful, don't overwrite our file!\n",
    "# model.save('cifar_10epochs.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['loss', 'acc']"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.metrics_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 56us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[1.3332478387832642, 0.6444]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(x_test,y_cat_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "\n",
    "predictions = model.predict_classes(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.80      0.59      0.68      1000\n",
      "          1       0.83      0.76      0.80      1000\n",
      "          2       0.44      0.65      0.52      1000\n",
      "          3       0.50      0.40      0.44      1000\n",
      "          4       0.50      0.75      0.60      1000\n",
      "          5       0.52      0.57      0.54      1000\n",
      "          6       0.70      0.74      0.72      1000\n",
      "          7       0.88      0.56      0.69      1000\n",
      "          8       0.82      0.74      0.78      1000\n",
      "          9       0.81      0.68      0.74      1000\n",
      "\n",
      "avg / total       0.68      0.64      0.65     10000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test,predictions))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Optional: Large Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "\n",
    "## FIRST SET OF LAYERS\n",
    "\n",
    "# CONVOLUTIONAL LAYER\n",
    "model.add(Conv2D(filters=32, kernel_size=(4,4),input_shape=(32, 32, 3), activation='relu',))\n",
    "# CONVOLUTIONAL LAYER\n",
    "model.add(Conv2D(filters=32, kernel_size=(4,4),input_shape=(32, 32, 3), activation='relu',))\n",
    "\n",
    "# POOLING LAYER\n",
    "model.add(MaxPool2D(pool_size=(2, 2)))\n",
    "\n",
    "## SECOND SET OF LAYERS\n",
    "\n",
    "# CONVOLUTIONAL LAYER\n",
    "model.add(Conv2D(filters=64, kernel_size=(4,4),input_shape=(32, 32, 3), activation='relu',))\n",
    "# CONVOLUTIONAL LAYER\n",
    "model.add(Conv2D(filters=64, kernel_size=(4,4),input_shape=(32, 32, 3), activation='relu',))\n",
    "\n",
    "# POOLING LAYER\n",
    "model.add(MaxPool2D(pool_size=(2, 2)))\n",
    "\n",
    "# FLATTEN IMAGES FROM 28 by 28 to 764 BEFORE FINAL LAYER\n",
    "model.add(Flatten())\n",
    "\n",
    "# 512 NEURONS IN DENSE HIDDEN LAYER (YOU CAN CHANGE THIS NUMBER OF NEURONS)\n",
    "model.add(Dense(512, activation='relu'))\n",
    "\n",
    "# LAST LAYER IS THE CLASSIFIER, THUS 10 POSSIBLE CLASSES\n",
    "model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer='rmsprop',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "50000/50000 [==============================] - 7s 148us/step - loss: 1.6025 - acc: 0.4197\n",
      "Epoch 2/20\n",
      "50000/50000 [==============================] - 8s 159us/step - loss: 1.1793 - acc: 0.5849\n",
      "Epoch 3/20\n",
      "50000/50000 [==============================] - 8s 155us/step - loss: 1.0083 - acc: 0.6508\n",
      "Epoch 4/20\n",
      "50000/50000 [==============================] - 8s 156us/step - loss: 0.9479 - acc: 0.6782\n",
      "Epoch 5/20\n",
      "50000/50000 [==============================] - 8s 154us/step - loss: 0.9145 - acc: 0.6952\n",
      "Epoch 6/20\n",
      "50000/50000 [==============================] - 8s 151us/step - loss: 0.9005 - acc: 0.7015\n",
      "Epoch 7/20\n",
      "50000/50000 [==============================] - 8s 151us/step - loss: 0.8867 - acc: 0.7061\n",
      "Epoch 8/20\n",
      "50000/50000 [==============================] - 8s 152us/step - loss: 0.8796 - acc: 0.7121\n",
      "Epoch 9/20\n",
      "50000/50000 [==============================] - 8s 151us/step - loss: 0.8693 - acc: 0.7158\n",
      "Epoch 10/20\n",
      "50000/50000 [==============================] - 8s 151us/step - loss: 0.8591 - acc: 0.7194\n",
      "Epoch 11/20\n",
      "50000/50000 [==============================] - 7s 145us/step - loss: 0.8586 - acc: 0.7235\n",
      "Epoch 12/20\n",
      "50000/50000 [==============================] - 8s 151us/step - loss: 0.8497 - acc: 0.7248\n",
      "Epoch 13/20\n",
      "50000/50000 [==============================] - 7s 148us/step - loss: 0.8462 - acc: 0.7296\n",
      "Epoch 14/20\n",
      "50000/50000 [==============================] - 7s 139us/step - loss: 0.8422 - acc: 0.7263\n",
      "Epoch 15/20\n",
      "50000/50000 [==============================] - 7s 147us/step - loss: 0.8321 - acc: 0.7330\n",
      "Epoch 16/20\n",
      "50000/50000 [==============================] - 7s 143us/step - loss: 0.8154 - acc: 0.7363\n",
      "Epoch 17/20\n",
      "50000/50000 [==============================] - 7s 147us/step - loss: 0.8241 - acc: 0.7353\n",
      "Epoch 18/20\n",
      "50000/50000 [==============================] - 7s 138us/step - loss: 0.8061 - acc: 0.7411\n",
      "Epoch 19/20\n",
      "50000/50000 [==============================] - 7s 140us/step - loss: 0.8120 - acc: 0.7409\n",
      "Epoch 20/20\n",
      "50000/50000 [==============================] - 7s 142us/step - loss: 0.8097 - acc: 0.7437\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x295e6eb3e80>"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(x_train,y_cat_train,verbose=1,epochs=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 75us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.9843294318199157, 0.6902]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(x_test,y_cat_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "\n",
    "predictions = model.predict_classes(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.82      0.61      0.70      1000\n",
      "          1       0.77      0.88      0.82      1000\n",
      "          2       0.64      0.57      0.60      1000\n",
      "          3       0.66      0.30      0.41      1000\n",
      "          4       0.66      0.67      0.66      1000\n",
      "          5       0.45      0.81      0.58      1000\n",
      "          6       0.80      0.75      0.77      1000\n",
      "          7       0.76      0.69      0.72      1000\n",
      "          8       0.78      0.81      0.80      1000\n",
      "          9       0.75      0.82      0.79      1000\n",
      "\n",
      "avg / total       0.71      0.69      0.69     10000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test,predictions))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('larger_CIFAR10_model.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}