CIFAR_10 image classification (ANN vs CNN).ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow.keras import datasets, layers, models\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np    # essential libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "(X_train,y_train), (X_test, y_test) = datasets.cifar10.load_data()     # load cifar 10 dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50000, 32, 32, 3)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape       # checking the shape of X_train, 50 k training samples, each sample is 32X32(pixel image) X3 (RGB channels) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 32, 32, 3)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape                      # in X_test we have 10 K images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x10674e7aa00>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (15,2))      # Reducing the size of the image and printing  it using imshow\n",
    "plt.imshow(X_train[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50000, 1)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train.shape          # Checking the shape of y_train, - 50 K samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[6],\n",
       "       [9],\n",
       "       [9],\n",
       "       [4],\n",
       "       [1]], dtype=uint8)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train[:5]           # first 5 samples of y_train are numbers - 6 is frog, 9 is truck, 4-deer, all these are in a 2D array, \n",
    "                      # Lets reshape this 2D array to 1D to convieninetly use y_train "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([6, 9, 9, 4, 1], dtype=uint8)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "                                # reshape y_train using reshape(), numpy\n",
    "y_train = y_train.reshape(-1,)  # \"-1\"- keep 50 K as it is, then flatten the array [6] as 6 now leave a \",blank\" for this  \n",
    "y_train[:5]                     # now we got a 1D array instead of 2D "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# airoplane label 0, automobile 1, bird 2 and so-on, defined this way in y_train\n",
    "\n",
    "classes = [\"airplane\",\"automobile\",\"bird\",\"cat\",\"deer\",\"dog\",\"frog\",\"horse\",\"ship\",\"truck\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train[0]       # y_train has the information about what the corresponding X_train images are, we can use this for training \n",
    "                 # here 6 means 0th sample is a frog (inbuilt, defined in that way) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Defining a function named plot_sample, takes X,y,index and prints that image, and label it (about what it is) using y_train\n",
    "\n",
    "def plot_sample(X, y, index):\n",
    "    plt.figure(figsize = (15,2) ) \n",
    "    plt.imshow(X[index])\n",
    "    plt.xlabel(classes[y[index]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_train, y_train, 0)         # cheking our above defined function "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# normalize the data by dividing with 255 ,each pixel value will be normalized into 0-1 range\n",
    "X_train = X_train/255\n",
    "X_test = X_test/255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# ANN model \n",
    "ann = models.Sequential([                         \n",
    "      layers.Flatten(input_shape=(32,32,3)),     # Define the input shape and flatten in to 1D\n",
    "      layers.Dense(3000, activation='relu'),     # 3D image data has a lot of features\n",
    "      layers.Dense(1000, activation='relu'),     # more no. of neurons to capture more features min dense layer \n",
    "      layers.Dense(10, activation='sigmoid')     # last o/p layer 10 neurons for 10 categories\n",
    "      ])\n",
    "\n",
    "ann.compile(optimizer='adam',                         \n",
    "            loss='sparse_categorical_crossentropy',  \n",
    "            metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "1563/1563 [==============================] - 118s 75ms/step - loss: 2.2310 - accuracy: 0.2753\n",
      "Epoch 2/5\n",
      "1563/1563 [==============================] - 122s 78ms/step - loss: 1.6843 - accuracy: 0.3964\n",
      "Epoch 3/5\n",
      "1563/1563 [==============================] - 130s 83ms/step - loss: 1.5864 - accuracy: 0.4297\n",
      "Epoch 4/5\n",
      "1563/1563 [==============================] - 114s 73ms/step - loss: 1.5318 - accuracy: 0.4491\n",
      "Epoch 5/5\n",
      "1563/1563 [==============================] - 116s 75ms/step - loss: 1.4936 - accuracy: 0.4663\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x106730ad280>"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ann.fit(X_train, y_train, epochs=5)                # training \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 3s 9ms/step - loss: 1.5274 - accuracy: 0.4527\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[1.5273629426956177, 0.4526999890804291]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ann.evaluate(X_test, y_test)          # evaluating accuracy on testsamples, poor accuracy "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "flatten_1 (Flatten)          (None, 3072)              0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 3000)              9219000   \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 1000)              3001000   \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 10)                10010     \n",
      "=================================================================\n",
      "Total params: 12,230,010\n",
      "Trainable params: 12,230,010\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "ann.summary()   # no of parameters is soo high in ANN, hence more complex and a slower training "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report: \n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "           0       0.55      0.46      0.50      1000\n",
      "           1       0.63      0.48      0.55      1000\n",
      "           2       0.42      0.15      0.22      1000\n",
      "           3       0.32      0.38      0.35      1000\n",
      "           4       0.35      0.51      0.41      1000\n",
      "           5       0.39      0.29      0.33      1000\n",
      "           6       0.41      0.63      0.50      1000\n",
      "           7       0.62      0.39      0.48      1000\n",
      "           8       0.47      0.73      0.57      1000\n",
      "           9       0.54      0.50      0.52      1000\n",
      "\n",
      "    accuracy                           0.45     10000\n",
      "   macro avg       0.47      0.45      0.44     10000\n",
      "weighted avg       0.47      0.45      0.44     10000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix, classification_report\n",
    "import numpy as np\n",
    "y_pred = ann.predict(X_test)\n",
    "y_pred_classes = [np.argmax(element) for element in y_pred]\n",
    "\n",
    "print(\"Classification Report: \\n\", classification_report(y_test, y_pred_classes))       "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "# CNN model \n",
    "cnn = models.Sequential([\n",
    "  \n",
    "    # cnn\n",
    "    layers.Conv2D(filters=32, kernel_size=(3, 3), activation='relu',input_shape=(32,32,3)),\n",
    "    layers.MaxPooling2D((2, 2)),\n",
    "    \n",
    "    layers.Conv2D(filters=64, kernel_size=(3, 3), activation='relu'),\n",
    "    layers.MaxPooling2D((2, 2)),    \n",
    "    \n",
    "    \n",
    "    #dense\n",
    "    layers.Flatten(),    \n",
    "    layers.Dense(64, activation='relu'),     \n",
    "    layers.Dense(10, activation='softmax')    \n",
    "    \n",
    "])\n",
    "\n",
    "#  2 convolution layers, 2 max poling layers \n",
    "#  conv-pooling-activation-conv-pooling-activation-dense network-output layer    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "cnn.compile(optimizer = 'adam',                          # compiling the model\n",
    "              loss = 'sparse_categorical_crossentropy',\n",
    "              metrics=['accuracy'])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "1563/1563 [==============================] - 32s 20ms/step - loss: 1.6905 - accuracy: 0.3815\n",
      "Epoch 2/20\n",
      "1563/1563 [==============================] - 30s 19ms/step - loss: 1.1361 - accuracy: 0.6007\n",
      "Epoch 3/20\n",
      "1563/1563 [==============================] - 31s 20ms/step - loss: 0.9905 - accuracy: 0.6558\n",
      "Epoch 4/20\n",
      "1563/1563 [==============================] - 31s 20ms/step - loss: 0.8934 - accuracy: 0.6917\n",
      "Epoch 5/20\n",
      "1563/1563 [==============================] - 28s 18ms/step - loss: 0.8151 - accuracy: 0.7175\n",
      "Epoch 6/20\n",
      "1563/1563 [==============================] - 32s 21ms/step - loss: 0.7484 - accuracy: 0.7377\n",
      "Epoch 7/20\n",
      "1563/1563 [==============================] - 32s 20ms/step - loss: 0.7030 - accuracy: 0.7562\n",
      "Epoch 8/20\n",
      "1563/1563 [==============================] - 30s 19ms/step - loss: 0.6602 - accuracy: 0.7678\n",
      "Epoch 9/20\n",
      "1563/1563 [==============================] - 29s 19ms/step - loss: 0.6131 - accuracy: 0.7861\n",
      "Epoch 10/20\n",
      "1563/1563 [==============================] - 30s 19ms/step - loss: 0.5679 - accuracy: 0.8023\n",
      "Epoch 11/20\n",
      "1563/1563 [==============================] - 29s 18ms/step - loss: 0.5234 - accuracy: 0.8168\n",
      "Epoch 12/20\n",
      "1563/1563 [==============================] - 28s 18ms/step - loss: 0.4976 - accuracy: 0.8253\n",
      "Epoch 13/20\n",
      "1563/1563 [==============================] - 28s 18ms/step - loss: 0.4620 - accuracy: 0.8396\n",
      "Epoch 14/20\n",
      "1563/1563 [==============================] - 28s 18ms/step - loss: 0.4278 - accuracy: 0.8493\n",
      "Epoch 15/20\n",
      "1563/1563 [==============================] - 29s 19ms/step - loss: 0.4004 - accuracy: 0.8589\n",
      "Epoch 16/20\n",
      "1563/1563 [==============================] - 38s 24ms/step - loss: 0.3698 - accuracy: 0.8698\n",
      "Epoch 17/20\n",
      "1563/1563 [==============================] - 34s 22ms/step - loss: 0.3569 - accuracy: 0.8745\n",
      "Epoch 18/20\n",
      "1563/1563 [==============================] - 30s 19ms/step - loss: 0.3191 - accuracy: 0.8881\n",
      "Epoch 19/20\n",
      "1563/1563 [==============================] - 34s 22ms/step - loss: 0.3055 - accuracy: 0.8921\n",
      "Epoch 20/20\n",
      "1563/1563 [==============================] - 28s 18ms/step - loss: 0.2860 - accuracy: 0.8994\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x1067c4e20d0>"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cnn.fit(X_train, y_train, epochs=20)                    # train the model using the data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 2s 5ms/step - loss: 1.2938 - accuracy: 0.6791\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[1.2937636375427246, 0.679099977016449]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cnn.evaluate(X_test,y_test)               # evaluate the above model using test sample data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[3],\n",
       "       [8],\n",
       "       [8],\n",
       "       [0],\n",
       "       [6]], dtype=uint8)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test[:5]                               # Similar to y_train ,y_test is also in 2D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([3, 8, 8, 0, 6], dtype=uint8)"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test = y_test.reshape(-1,)            # Reshaping it to 1D so that we can use y_test variable in plotting function\n",
    " y_test[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAI4AAACcCAYAAACp45OYAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAVSklEQVR4nO1daWxc13X+znuzDzkckqKohbQlW6plS7Yj11uRpg2cuBDSICkapIkDBAnQokjRFi3QAg3yrz+KOmhR9EeBBgZq2EWKBkYbtKkbpI6dtImLoLHjOLIdWwtkLZSohaQoLrPPnP6Y0bvnXHEZP4oj0jwfIOjO3Dvv3Xk8c896v0vMDIPhvSK41RMwbE6Y4BhiwQTHEAsmOIZYMMExxIIJjiEW1iQ4RHSEiI4R0Uki+vLNmpRh44PixnGIKARwHMDjACYAvALgCWb++c2bnmGjIrGGzz4M4CQznwIAIvoGgE8CWFZwhoeHeXx8HABwKwOPRLT2i6wwfdV1w61EL680DzfOn668Pnk36Pa5rvQM5DWOHj06xcwj/pi1CM5uAOfE6wkAj6z0gfHxcbz44osAgEajofpuyh+zS9x0weEVugK/z/UGN3Y6UMs1Rbs9zM2fPGvjZgvOjh07ziw1Zi02zlJ3vmHWRPS7RPQqEb06PT29htsZNhLWsuJMABgXr8cAXPAHMfNTAJ4CgMOHD3MYhmu45c3BzV7dqNVUr9WvJ9D3asnfG3vPQqguCoSqgl5x9B3Wd8VZDmtZcV4BsJ+I9hJRCsBnAXxrDdczbCLEXnGYuUFEfwDgvwCEAJ5m5rdu2swMGxprUVVg5m8D+PZNmothE2FNgvNewcyR/ryV7vgaYlfyIqLt2SBqmG/TOeugWteeZSKZdC+a7pohrTRf3/5ZO9bbxjFsYZjgGGKhp6qKiKLlfr0DfuuuCsX0m969uOU6Gy2tSuoN57qfOHVK9Y3u2B61W7Va1B4ZGlTjMmmn0lrr8D27+dvYimOIBRMcQyyY4BhiYcO44xslydm9beSuESZTqqcpUgflharqm722GLUvTc2ovmx/PmoP9/dH7YD071smNv0E6PLT1d95rU/bVhxDLJjgGGKh5+540MkWS5c1Lm6og1pBy0j1FKygqppiEW95rnQYut9ZrVaP2lem59S4ucVK1C5XdeZ8seRUV5DO6b6yc8H7cu7LNLzvJRVjXA2/VtPAVhxDLJjgGGKhp6qq1WphsVTuvNDrb0IUeLHXFybCJdvkJf+k6gpay/8mAulTeEv2QtWpGd/Dyibc46qIBOWkp6ouX3WvW57/Uhd6pzS/oD8nvKyJ85NR+579d6hxd+4Zi9ohe0VkKvkqnoGvmeQj8FRh0IXPZSuOIRZMcAyxYIJjiIWe2jiNVguz5bY72pfLq74g4TK+zZYucFLmilC/oaeKA1XsvcJvQtgBvlt6cfJ81B4aGlJ92YxzhKuVUtTOpXXkeMfINncrz15YLDkbKp/Sn6tVylE7DFwoYKGqo88NMWci/SfUdpkcp4bpWfn7troIntuKY4gFExxDLPQ2chwmkCgMAwCaniqpB6I2l7SLKV83xR6mwE+Uite8QhhZue3eMt2oObVAnqsLoUKLIiFZr3v3Cp3azfX1qy6pqihMqz4Sujedddcgb5INkfT0y52x7HfTcxTVzUvsUraaY8M6wQTHEAsmOIZY6KmNMzU9g6f/8esAAPLSCknhjvf1Z1Tfvr23Re2H7rsnaid8sgdxTT9dwMHSaYaG5/oPChc8ldbzkK51KuXsk+HB0BvnXic8lzsl0hZI6utXBIPH7NxV1752TY2bvzYbteulsuqT+YPh4WLU3r9Ppy2SKTcP36TxbaqlsOqKQ0RPE9FlInpTvDdERN8lohOd/wdXuobh/YduVNUzAI54730ZwEvMvB/AS53Xhi2EVVUVM/+AiPZ4b38SwIc77WcB/DeAP1v1Wq0Wyh13tFauqL6kWMLn9cqMnOhr3n0gale4psYFQlWlU1l9b7Ecy31Q7IVUB4Yc+dQNBV8ihFATRV6hp44g3GXfW24Jt/j0Gb2v6vzly1F7RnAJlctaHTWrTqXVyvoZVKsuoj02Phq1bxsfU+PyKfmn99T6OmbHR5l5EgA6/29fZbzhfYZ196okI1d5cXH1Dxg2BeJ6VZeIaCczTxLRTgCXlxsoGbkOHLyPf+s3PwUAqHreQD7rVAt5S2dWLKtyN8jcnC6gajVcHXAyoT2WRNa9ZlEMVq7rpZ5b7l6BF92Wnl9CXCOZ9LaeBMurwrpQk5VWXfXlC31Re7BYjNrNmh6XCd2zmp3Wen3i/OmovW/vvqgdBvpPLdV16M1xPZOc3wLwhU77CwD+PeZ1DJsU3bjj/wzgRwDuIqIJIvptAE8CeJyITqDNc/zk+k7TsNHQjVf1xDJdH7nJczFsIvQ0cgxmtOptIyX0FjsZe+1L6SKvbMZFacsVZ9eU6jp7ffrU6aid8tzx2/beHrXfPefIUZ//zktqXD1wdkzGK9DKiXnkhc00UCioccUBlxE/fPg+1TeyzcVK7xzbrfoCEkX5wqWvVXQhV0LYK+Xtuths186ia+/eGbWbTf2sSiVnN0n7ElDRhGVhuSpDLJjgGGKhp6rq6rU5/Nt/vAAAaNW1ixlAbH9N6a2x/UIV7NnvIqAjw31q3PBOlwwd2qZjkpm8Uy2zbzuW+TffPqfGlYUvmvB4HxMiTNAvrrfvttvVuF96+AE3p7wu5MqHIrnoBWhrNRcRbjSdeiqJpCYA1Jvu2WVzOuxQLDo1f+nipag95TNj5J16kkxgAJDL6QKzpWArjiEWTHAMsWCCY4iFnto4pVIZr/60XdaT8VisalXnZidTWp4fefShqH3mvLNJpifVMBw6eDBqp7Ja95eqzoZKCrf68APaXa6UnW2RSurHs/+OvVH74N13Re1d24pqXCHn7IdWRac0zl28ErUvX72q+ianXN/igsvrzc7OqnG1uptjMqXnKIvPmmKfet0j484Vne11CAdV38CAtsuWgq04hlgwwTHEQm+3ANdquDLRdoWHBnW16e4x5xLec99+1ZdMO7/1rdd/HLVHM1od9Yn9V5entB7LFwai9nDBfe4TR35FjZNEjQMDA6pv2/Bw1J6ZcYVW7545ocZdm3Vqd+7avOqbn3OFVrNemcnMnMt0N0S4IinPeACQEgTZQah/+wMF96yKIsM+uF2rn3TOhTxSWR3+WPCK7JaCrTiGWDDBMcRCT1VVrVrB+ePtQ4LnCjrq+/Ff+1LUPnJEJ95f/N4LUXu78Aa2e4wX2YRbpjMe/+/ogIs+94t2xou8NkR02N8e0xBHAV085lgtzl6+pMbVxJbgREbPsb/fJSW3Z7SKqHsFW9eRTGlVJUksQ09V9Qt+5EKhX4zzmMcWncq8dGlK9VUEE8dysBXHEAsmOIZYMMExxEJvz3JoNVEptV3Qe+8/pPoe+8hjUXu4OKz6PviIc5kDUQjen9RZ3EKfsyfClFesLgq7WFyjBR3ZvXbVudmFhL5+S5Sb3XGXm//2sV9Q42YE62i/cIkBoN4UbGDegfVJQfUiybkrFe0eLyw6tlL2jq5eKLm+c5MuJFEpa7ulLuhW/CKvXN6y44Z1ggmOIRZ6qqpSmRz27LsfAPCZz/+O6is1nct57KR2b1sk6oCFG1/3KqFmZsWS29JLc7Pp9nFJvsUWdD3v/JyL9IaXtHt8QWzRrVZdX6uiE4h5ESY4dWJC9b179qyYh3azh7Y5FV0ThJHXPLaK6SnnPrOnZgJBOkmi7dcVF0WYIJPRqqm84DFgLAFbcQyxYIJjiAUTHEMs9NTGGRwawqc+97l2e4em3fjZm84WqHmhd0kp0oQ8LMTfmyUIob39502RLpCMpDfyaIvip4ZOW0xNO9ur0XB2QOBxmRQLRTf3mrahZqZFRjzU1fBTU85Frtbd9RtetropjpYOvUKunCDxTsvUREPfq1aRz1jbSdm8DmUshW62AI8T0feJ6G0ieouI/qjzvrFybWF0o6oaAP6Eme8G8CiA3yeie2CsXFsa3ewdnwRwnURpnojeBrAbMVi5SqUSfvr6qwCAo2+8rvoIzl0MQ+2mJkSEOFT0JX7WWJI26t9ERhR9ycKoVFq7ooGIMIesr19IuUU1SIuwQKiX+kpT7I/yj0UUBVT1klZjpUUXca41BFG3twdNMYM1PZUsst6L8+4aOU+ljQy4+Se8CgEvGb8k3pNx3KF0Owzg/2CsXFsaXQsOEfUB+FcAf8zMc6uNF5+LGLlq1dUDS4bNga4Eh4iSaAvNPzHzNztvX+qwcWElVi5mfoqZH2TmB1Pp7FJDDJsQq9o41D7Q6R8AvM3MfyO6rrNyPYkuWbkWFubw8g9eBACU5mZVXyrpdH825+/rcdMMWey99uQ+SEobR6cjMqKaT4bYU14VXiLnwv6ZlC5WTwkKFEnOTRmPyk2QVNerHiuocK3rHo1cS1Ytimsk/ANN5IEpaW2QDOSTou2eVV9W72NLJ929kqRtKGpq22spdBPH+SCAzwN4g4he77z3FbQF5rkOQ9dZAJ/u4lqG9wm68apexhIn03RgrFxbFD2NHCcTIUZH2oXik+Urqq/ZnI3aBe9Iw4TIjs9NuW2z83N6X1K96Zb+VkMvt9zyqao7CLw9S1nnHHJSM201RFo9ELoq57F/5cU+paa39VYdm53WqpaEes0I9znrZa+HRMHamHce1thOd6yj9LKrFb2/K2CnMhNeIXuxsLotarkqQyyY4BhiocfkkS1wvR3ZHMhrK39e1NXWmwuq764Djk2Bdzo1dmVqWo27PO0KnBZmfbJEF1Ftishuq6ETiPmE86QO3Hen6rsgiryuCK+wXNMqsyxP8/XMw7Rg6ch7W3uLgiVrRNQq79i1Q43bt9ud0bA9rZOXCyL6PDPjzIHQi6Tn8i4K3tevVdPw8OppR1txDLFggmOIBRMcQyz0luakXsP0hXbBVrOubYuyiI6Wzp1VfUMiW75NFFknq7ogPSsqqsqhdwYTS7dY2D+kx5XKzk760EOaqerg3fdG7bNnHXPp9Kxm1qrKaLF3hGRCRH2zge7bJtzuYt59z6ZXaHVxyj2fYx6dC4lCrsJ2FwXPFrTbnhPHX8sieQDo8+hdloKtOIZYMMExxEJvI8fJBHZ03OmJs3q/UUMcFwjS0dZ3jx+L2tcEebYv9Yvi/KfFhk7ctZrymsuf1SQjrK/97wuq78N5V/x0SBRTlT2yxVbDqRZq6O9SqTkVfc1LJspwwpl3BLl1WVexVMT5WFnvLIfBHcWonS64ZxV6Sc6coHpJe3QxFK4uFrbiGGLBBMcQCyY4hljorY2TTmJ8/zgAYG5R6+3FCUknpu0OWfw9I/Y6pUhPvyZc7iZrFxa8dHac2C/Ccu2TR19Rfefmnd00EgjaFO8Qy6awfxa8TVcXRVb6pBdOmBAZ/VLOfbf+8Z1q3Kg4eytT1Bl8yLM3xb6qvj5NnZcT7nng0cVwFwdW2YpjiAUTHEMs9FRVhYkECoNt93FkVO+mmRSqyi83lMHXqoii1r1SXKmemlimcMsD+/W84ub1st6VsSjOWgjSxagdVnUU/IKY4+sejcrJhJvXYp/OjufHXFZ6ZNeuqD08MqrGpfPOza7Bj5C766fFgVuhd/iW3IMWJrQYBN7W5KVgK44hFkxwDLHQU1UVUIBsJ0mZ9upo5VFDzbpWM9LxaaikpKeOZJd/bqHn+URX8CLHLF4veHXK79ScFzQg6ozfqWgGsbcarrBrpqC33wyNu6OLdu7ZpfqKokgtLaLUQUvPsS7UUZjQEeFQeEiJlOujQF9DEkaS9wwC86oM6wUTHEMsmOAYYqG3BNkA6p0o8GJZ7/PpL7pNQJVF7cI2JSOX0L9N32yR5NNe4Hi5PYXs2UIsMsOLgc5sv1xz7J9nSiKandO/v8ToeNTesXtE9e0dcfuehgd0AVUg7JpFYbBVvGKzhHCtfcbQjMh0JwRJeMY7kyq9DO1Lt+iGkStDRD8mop91GLn+vPO+MXJtYXSjqqoAHmPm+wF8AMARInoUxsi1pdHN3nEGcH2jU7LzjxGDkYu5hXqneClM6eV3cMQtsfU+7WI2hHsuPfW65y6zUFU+oSNJYknhfrLnikKQVicSnhssiqGqA851vmNAR8EHh1zisa+gH3FfzqmZdEb3VUQBWE1En9lTJaE8ndifv3idFO64HzlOimuEXqT4hmj6EuiWHyfsMFVcBvBdZjZGri2OrgSHmZvM/AEAYwAeJqJDq3wkgmTkqlZW510xbA68J3ecmWfRVklHEIORy48WGzYvumHkGgFQZ+ZZIsoC+CiAryIGIxcREHYKrYtDurCoT7i0zZrWsdLGaSiiay9ULoqYyGfrEro/EHubJF0JACSS7t5Zzy7oF3uRRvvc3qM+j6IuL9IRKY8xqyZeLnj7ucuiYE2GHTLeYSEpETKQdgygM9skCsr8YjNJQp5K6cL+VHL17Hg3cZydAJ4lohDtFeo5Zn6eiH4EY+TasujGqzqKNkWt//40jJFry4L8JWxdb0Z0BcAZANsATK0yfCthIz+P25l5xH+zp4IT3ZToVWZ+sOc33qDYjM/DkpyGWDDBMcTCrRKcp27RfTcqNt3zuCU2jmHzw1SVIRZ6KjhEdISIjhHRSSLacmUY76fTBnumqjqR5+MAHgcwAeAVAE8w8897MoENgE5Obyczv0ZE/QB+AuA3AHwRwAwzP9n5QQ0y84olKrcavVxxHgZwkplPMXMNwDfQrunZMmDmSWZ+rdOeByBPG3y2M+xZtIVpQ6OXgrMbwDnxeqLz3pbEZj9tsJeCs1S1+JZ06eKeNriR0EvBmQAwLl6PAbjQw/tvCKzltMGNhF4KzisA9hPRXiJKAfgs2jU9WwZdnDYIdFnbdKvR6+z4xwD8LYAQwNPM/Bc9u/kGABH9MoAfAngDbuP7V9C2c54DcBs6tU3MPHNLJtklLHJsiAWLHBtiwQTHEAsmOIZYMMExxIIJjiEWTHBigIhOE9G2Jd7/xFbJ+ps7HgNEdBrAg8y8UXcmrDtsxVkFRJQnov/s8AO9SUSf6XT9IRG9RkRvENGBztgvEtHfddrPENHXiOiHRHSciD5+y77EOsAEZ3UcAXCBme9n5kMAvtN5f4qZHwDw9wD+dJnP7gHwqwB+HcDXiCizzLhNBxOc1fEGgI8S0VeJ6EPMfJ3P7XqC8idoC8hSeI6ZW8x8AsApAAfWd6q9Q28PrN+EYObjRPSLAD4G4C+J6Pqxedc5W5pY/jn6BuT7xqC0FWcVENEuACVm/jqAvwbwwHv4+KeJKCCiOwHcAeDYah/YLLAVZ3XcC+CviKgFoA7g9wD8S5efPQbgfwCMAvgSM1dWGb9pYO74OoGIngHwPDN3K2SbCqaqDLFgK44hFmzFMcSCCY4hFkxwDLFggmOIBRMcQyyY4Bhi4f8BGpzFiwLSV4UAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_test, y_test, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_test, y_test, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[4.41712388e-07, 1.92156691e-07, 2.36790856e-05, 9.97344315e-01,\n",
       "        3.70219873e-08, 2.60594441e-03, 1.43627485e-05, 1.35172595e-10,\n",
       "        1.09635121e-05, 2.11559907e-07],\n",
       "       [2.38699578e-02, 8.77629399e-01, 1.11689875e-08, 2.26918984e-10,\n",
       "        1.33433181e-13, 3.68789255e-13, 5.38155166e-14, 2.87107794e-12,\n",
       "        9.84979942e-02, 2.75592515e-06],\n",
       "       [2.83664912e-01, 2.14317277e-01, 3.84848448e-04, 9.23778571e-04,\n",
       "        1.28520740e-04, 5.02635612e-06, 7.22473175e-08, 1.22831678e-02,\n",
       "        4.79053557e-01, 9.23890248e-03],\n",
       "       [9.37574148e-01, 2.12705745e-05, 1.66812436e-06, 3.67018856e-07,\n",
       "        6.23899736e-02, 6.66893984e-10, 1.75688596e-11, 8.77269229e-11,\n",
       "        1.24983280e-05, 1.06843217e-07],\n",
       "       [1.30748665e-08, 1.17007530e-05, 6.82970703e-01, 2.44765133e-02,\n",
       "        2.60668606e-01, 1.11139918e-04, 3.17611806e-02, 1.67475145e-09,\n",
       "        3.66970587e-09, 6.40236024e-08]], dtype=float32)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = cnn.predict(X_test)           # predict the values and store in an array\n",
    "y_pred[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[3, 1, 8, 0, 2]"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_classes = [np.argmax(element) for element in y_pred]      # apply argmax on the predictions to get the same values as y_train\n",
    "y_classes[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([3, 8, 8, 0, 6], dtype=uint8)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_test, y_test, 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'frog'"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes[y_classes[5]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification_report :\n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "           0       0.68      0.73      0.71      1000\n",
      "           1       0.82      0.80      0.81      1000\n",
      "           2       0.53      0.62      0.57      1000\n",
      "           3       0.48      0.49      0.49      1000\n",
      "           4       0.65      0.57      0.61      1000\n",
      "           5       0.58      0.61      0.59      1000\n",
      "           6       0.79      0.74      0.76      1000\n",
      "           7       0.72      0.72      0.72      1000\n",
      "           8       0.79      0.75      0.77      1000\n",
      "           9       0.80      0.74      0.77      1000\n",
      "\n",
      "    accuracy                           0.68     10000\n",
      "   macro avg       0.68      0.68      0.68     10000\n",
      "weighted avg       0.68      0.68      0.68     10000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Classification_report :\\n\", classification_report(y_test, y_classes))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 30, 30, 32)        896       \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 15, 15, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 13, 13, 64)        18496     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 6, 6, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 2304)              0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 64)                147520    \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 10)                650       \n",
      "=================================================================\n",
      "Total params: 167,562\n",
      "Trainable params: 167,562\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "cnn.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_test, y_test, 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'airplane'"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes[y_classes[3]]                 # prediction is correct here"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_test, y_test, 21)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'bird'"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes[y_classes[21]]                 # prediction is a mismatch. Some evn us cant tell its an airplane/bird"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAI4AAACcCAYAAACp45OYAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAXbElEQVR4nO1daYxk11X+znu1V+8z3e6e6VkdxxljJw5xnKAQQBBLJggFkJDiH5FBkSIkkIIEUkL4AfxAMkJC/ERGBIwEBEsQESWRIhMgGyZxYmVx4tizZLqnZ3q6p5fq2rf3Dj+qXOecOz3u8ptxzfT0/aTR3Ff31n33vT51z34uMTM8PN4ogtu9AI/9CU84HongCccjETzheCSCJxyPRPCE45EIN0U4RPQ4Eb1CROeI6JO3alEedz4oqR2HiEIArwJ4DMAKgBcAPMHMP7p1y/O4U5G6ie8+CuAcM18AACL6DIAPAbgh4eTSIRdzvVvGDr1SmN613ZtbX8XSdImeZANlZzPVc0Td7qBdq9fNuNgszNx4l2u5243A1/Wpa77RfACpcUHgjNMP47yDIAgH7TiOVNuOI/0szvR6yiiON5h51l3fzRDOUQCX1PUKgPe83heKuRR++ZF5AECjbfvSk8cG7WBi3vRl0vICwrgxaHMUmXFRmB204zBv51Ave3trc9B+4cXvmXGVmswfpjKmj27wuiJnHXoXj+GskTpyEVviDlmuAxLiLhbsfUMS4oi6dv7x8fFBu1wuD9qtVseMS0HelSuwdBXBbVXqS9gFN0M4u/1crvvpEdHHAHwMAArZ8LoveOxP3AzhrAA4pq4XAVxxBzHz0wCeBoDDE1kOqXfLKO6acTnFnU6fusf0HZ4ZG7Rr2+uD9vrVVTOuHsmcUWB/id2msLifnJcfUXm7YsbFamfqdlumLwxkjkCxxVw+Z8blsnLd7tqtldQcnbZ9B4hkVzh+7PCgPTlu/0y1UlXmszOgUJCdtjUpvVsbJTOuXZd1LRxdMH1l1bdVsaz8NdyMVvUCgPuI6BQRZQB8GMDnbmI+j32ExDsOM3eJ6PcAfAlACODTzPzDW7YyjzsaN8OqwMxfBPDFW7QWj32EmyKcNw4CqCfMZPNWYzl+VOSak3Pjpo+7zUE7TInsEkxa2aLakr5GbLpwZUtpGA2RaxbuOWTGhRmlmTmqbr4oMoNWkbNZuw6tBjfqdo7tbZEZikX7vUZVZAvqirxTLdXMuPKmPMvxxTnTt7UpMmBBzT87VbT3Son8NjtTsOvH3vAuB49E8ITjkQgjZVUMQifs2XLmZq0KOJUV1rX2sjXKRW0xynVbsm3HkVVnw6yo7YXchDOHbO+ZjLCSbMGyi2xO1NlTp0+bvnqjpK5kQ19bWzfjNAuamXLYQFtYhGvVarOsq3RN7jU9bY2Zk8rIFwSW5afTMnZzQ9T2uUOTZtyRe0Tdr9cbpu/y1avYC37H8UgETzgeieAJxyMRRirjdDnGdqPHT9NVq2Iu7ZQGbapsmb7FeeHHrYqMK5dLZlyuOCX3ylpXQhiLqjuhTPhhxv52imMiM6RTVpWu7ohZoFAQeSqXtvLDxtrGoM1sXR/Hj4mjuVa2sgW3Zc7JcZGTMlaMQXlHVPorq/Zd5ZWM1o3EZHBhacOMmyiI7FXvWLdIy1nzbvA7jkcieMLxSISRsqp2J8KltW0AwNY1y0pyKrhq1lGRZw9pW6Yosd223VJb8c6g3dy2W3N+TLbwmZw8diWy/uWxMWE7V1evmb6JCelrt2V7d80Cc3Nizc04oSTFglw36q45QcXIKPNEm62XPpWWOdbWLatqFWVsrKzbnLZ/6jXFdjNpu38EdOMAs8GYPUd4eOwCTzgeiTBSVkUUIJXuSfPNyo7pa9Vl60x3bJhjaUfGpkPRdLqx447rKKss2760YoVNVlbqkmWZSy2Jhi1OWGdrEMjvrNEQjSibc2OkZasnJyiyWhWNaLtk34EOVa035X1wbLWv2QmxihPZcOCUYnEtxco3S2UzLoK8q6lxu/5arYm94Hccj0TwhOORCJ5wPBJhpDJOKpXC3EzPChw4ckG0Izw4rlveX6lKX1Gp6hRauk9n5HECR8bhtlznVI5Rxgn3LreEv3crVj5pNUUumJsV2aLVsury1paoyLVa1fTde++JQbtYsJ7zTRWgNTEp5oNy1Vp2f7KxIuuYmzF9zaaMbavUmdB5V0TybKnQvittMgCsvPka/I7jkQiecDwSYaSsKgCj0A+AOnXfW0zf3LjExC6ff8X0FXLCTlKG1K3lmNR1t2Wtsh2VyRgrtTeVsiwzlxFWGGatd3HusHK2Ni170igW5VmmpqZMX1Op2SnHmqvV+CgS9uGymXxR2JieDwBqTXnOMC3rHyvYmOMgFnbUaVh21G57J6fHmwRPOB6J4AnHIxFGKuPk02mcOdrzHOeLVhWdX1TVKtJW7mjsSI54rSyVJuCU/whDeZxOx8ogkQrYOqRypeNu1oxbUSpx7LgL6qqSha5Isbh41IzrKvfG2Jh9zlZL1PNzZ5dNXyol66/WJNBNu1kA68YgctRs1e50RFZJpV2VW32vbT34zfotcDkQ0aeJaJ2IXlKfzRDRc0R0tv//9J538rirMAyr+gcAjzuffRLAl5n5PgBf7l97HCDsyaqY+atEdNL5+EMAfqHffgbA/wD4xF5z5bIZnDm1CAA4f9nm7qxtbMu4MbuB5ULZ+lmVDSmXrWebAnkcDuyjzZwUVvjIz39A5oisyv3yuYsyR2S397yy9Gq2eOTIETOu2ZCtfnn5ounLKJb5rnf9tOlLp0TNrinr+dXVFTOuq+K1y06UQQR5J6F5B1blzqhU6rl5GzN9z7xEBZx7/hJ2Q1Lh+B5mXgWA/v9ze4z3uMvwpmtVRPQxIvo2EX27MoTQ5bE/kFSrWiOiBWZeJaIFAOs3Gqgrcp0+MsthP9cj5cQVnzsn1uLjczY9+PhhYV25BbGARmnL0mo12bbHJ60mcuqhRwftQ8fODNphzabpnFbO0bQTejt/SNbFyuIcO+M6ytHYbdgfC2VEgzlx4qTpy6pKGbroZNtxolZ2hB1dXDpn+i4vnR20ox1xtjYrlq0vXxXtdHPd9s0vatZ7a1nV5wA82W8/CeA/Es7jsU8xjDr+LwCeB3A/Ea0Q0UcBPAXgMSI6i16d46fe3GV63GkYRqt64gZdv3SL1+KxjzDiMieMiHtq4fS4tah2piUAezJjl1VRRZE7ynv91nfYssqdjlh2x8esRXh6QQKoWi2Zr1baNuPilgpCH7Me5U5d5CFSVbhSjhU8zEnfyZMnTV+gni3vBHLVVbFubX2enJgy4zJp+d745Jjpm1EBYMsvfXfQLjlpvvMLIq9t12ww/IVLa9gL3lflkQiecDwSYbR5VWAE3Nsyx5xyVFPHxFEYOlWmnvve9wftktJMP3jiPjNu4Ziwo8kJW8VKxwWvvCrHTaxdsY7GTE5Y3HTOmgyurF6QNSo2c+rMT5lxYVrmyKbsHJQiNc6+BO04rVRFRY6d/LGUslqD7HOefsvb5ELZzbJpyxbfdlRU7nLTsrHPfuE57AW/43gkgiccj0TwhOORCCOWcYB0/zylfD7r9Aq/v7JlS3e8ui5m72Ykpv7tqs1ZmgsWB+0obWWLzTUJBlu5KGb5uFYy43LTUjC7tmnXUd4UNTXTUfnbTgWrrC5X4sgx2mEdpmyfrqaVU/JV3LXuk5DkHYQZq46ntZ/kraLSE9t7zS6IX3pizuafXymVBu2XXt29cL7fcTwSwROORyKMvMxJPuxZYyllWZU+FWml5pwhpXT3pjoL4epVG+B0772qoLWTi9RWltM0S1DTjHMeBBTbqW2XbF9b1NvpcTnFr+iw3UDlSxFb00Kg0m1dVqXLqOhT9yKHFeYzKqDMSaVmldpbmBC22+7Yey2fEzPEqayd4/jxvcOr/I7jkQiecDwSYcSsKkS6rwV0nAKFOyptZL1jj/Mbn5OYWFJW09KOLRDZqCstK7ZVHKZn5Dpz5oFBO6zYObY2JBis0bTrSKkDVg8fUvNlrfW2qwy9Ab3OIT7Xnb6rqo3pgpTOuyJ10i/c9BidMqQ0y7STAryqAr6aHVutqxRZUWE3+B3HIxE84Xgkgiccj0QY7dGKREA/AKrasHz04sblQbvCVrYIVUHrVFFUx/WqLWBdVR5lsg5fzM6LVXlmSqy+28sXzLhMJGpqdsvG4Ld06RRlAm5HjkqshBzuOtU+VemRlKOq6+AtLQBlso6VXRW0jhzPedSROdrqOGo450EEKRkXNOz7LrjVXHeB33E8EsETjkcijJ5V9a2jlzctG7i4cWXQbmaspVSfxqvzjS5eWjLjXlGVvE4snjR9Y4o9pYuiSo/N2dTYICdqa7hqVdj1S+cHbRUGjWrVKbCoWE4QOecwjCk245wy3FGFwfN5UfEDx8IcQ+bvujlXW5Ivtb0mTllWxzUBwOFxcY6Opy0fqzZtDPJu8DuORyJ4wvFIBE84Hokw2qMVu11c6+cxnV+znu2ykgXaDj1zJLKADnAKUjaQ6+vf+sagvbBgS4+8b0rln6s5KGsDoVJTcu9Jh/eTCjynjJQCKZftOmJVmHqyYFXpUKnZNScQTVfkCpWnHGRloUh596u6QhmASz8e1L9CZ1sC0WYcDzimRM4jp5xLHN0CdZyIjhHRfxPRy0T0QyL6eP9zX5XrAGMYVtUF8AfMfAbAewH8LhE9AF+V60BjmNzxVQCvFVGqENHLAI4iQVWuVreDn2z0Yn83mnab7mbUMYCx3ZqhrrX3d+6IjZW9qlJXv/bNr5i+e09KQe75e4SNaW87ANSrwmYqDVtkGwVR6evKQttpWg87WNbYbjvsripsoF63JVYOHZbAq8lJVSXLeR0hiXpe3bGW6SuXxPI9l5d7Txbs2VsakcOarjsHbBe8IeG4X9LtnQC+CV+V60BjaMIhojEA/wbg95m5vNd49b1BRa5648Zl7D32F4YiHCJKo0c0/8TM/97/eK1fjQuvV5WLmZ9m5keY+ZHCdSkxHvsVe8o41DuZ4u8AvMzMf6W6XqvK9RSGrMrVjrpYKvU82k33zjpwzWWxpJOR1OLzlu7njwm3XLpsc8JfPS/54jOqpIrreW6pkmcXz543fWvKWx6rYPJMaOWYSkW8zZ2ue1CJPNzOjq0Y+vZ3vH3QnjbmA5v3nUupYHXYYPtqVZVpUd7xxXkrSej884zj0qC840rfBcPYcd4H4CMAfkBE3+1/9in0CObZfoWuZQC/OcRcHncJhtGqvo7romMH8FW5DihGaznmGFudnhrecYK4daB26ARg63wj7VB2zzHI6C02ZVnE8op40t/ziFQgJecIxrgt23tFpcICwA/PCbur1YWl5Zxzp2JlPmi2redcVxbtOn2lb4gVeGFeTA1n7n/Izk/ybPNHbIXWk/ffP2hfVhVJtyN7r7kx8b63nWpd7hkWu8H7qjwSwROORyKMtngkxWiFPVsOu9ZJfayg+0X1gXYgpp1jEbVukHOORSyXxUrLqnJD16lurW3FswuWDRzaEXbX3dSFr+2KJ8aFDUyQ1Xp04YnQiVVeXxLL93899yXpaNs/04IqrF2pWJNaRcUxb8Wifb140cZnP3hK2tS2rCqzt+HY7zgeyeAJxyMRPOF4JMJog9UBxEGPyYeOaUgHE1FkZYZMIDJDVpX4KORtznZWBWjl2coWY8r6qu/cco5fjlTO9tys9b7PLknwUzvQ3n0rFBRIWaMddb/VUIHgTrD6rDqna/mCBLo9u/Z5My49JWdvXV2znvl2W/yB+kwtYvucS8siTz30luOm7/TcFPaC33E8EsETjkcijJhVEdBXhYPIbtMFEjWyWLQsqJgRp2Q+KyxHW5QBIKeKWxdmLKvKpiW2OJuS71VbNv2VlYW1UbXbe16pzwsFcRq6lldS0SNR17Ecd7TZwbKxMC15XNkj8pxLqzbga/WaFNNkx3oehsoooW0Lsf1Tl8+K5fvapnXmlh+w8dq7we84HongCccjETzheCTCaEu5MSEd9WSPfGjN7dMFkUEKTnHruCWyQLum8rId/h41lVfaCVNNjYsM1VIl2gLHE9xQAeSrV66Yvo7K086mZL5O0wa1swpWz4bOISAs8lucsiYJLfatXZN1lDrWPBFlTIiA6YvVMdGsTAFB4JZikfe9tGFLzpS++TL2gt9xPBLBE45HIoyUVYWUwlR4GAAwPm7zfFKK7bScAKeuKkyN+DrfucyvVNFm07KqjvIA1yoS61twjjfcuiYW1fPqSGsAyKmxU5OSA1V00mvDQJkF8vY5W21ha022z1leEw/2SklYVT12ft9K42ZHpdecV5srYscar8+8ZrKRBJstH8jl8SbBE45HIoyYVYWYyPZSW2OnuGNbKQfttt0qWSktrLZYIpvWoTNZie2jTU0W1DhhEWXn3IjLV68O2mvO8UenZ6Zk/qzcO+WklxSLYgEu5J0TgtuyyM31kuk7uywOyx2dfhzYFB5u60oWVqsKaPf345xkCQ7kHbTYaoVp5LEX/I7jkQiecDwSwROORyKMPJBrUFzKCXAidb6UG2ieVpWxWC257VQhCdW5ThMTlqufPCHqc6hKpVxcsdZhVhbtMw8+bPqmlLySUqq/OXgDVkWuOYeMxCpQfn3NylCbm/JAsUorDpzzqsDyrtzcMh2m1umos7HYWaPKa4tSVuBMdfZOAR6mIleOiL5FRN/rV+T6s/7nviLXAcYwrKoF4BeZ+R0AHgbwOBG9F74i14HGMLnjDOC1ANt0/x8jQUWugAi5TH+rJnvrTEa2x3TK9inNF/o4hUrTWl5DtYUfnrZFIcfVmVKlVbHQbq2smnHz41OD9uSk3UTjpmzpOjbZrUjRUFZql43VFX9ddeKF9dlToWKZAdn5WZ0j4bIqrY6bcx6cPLYplftFoWXrOxuODLALhq2PE/YrVawDeI6ZfUWuA46hCIeZI2Z+GMAigEeJ6MFhb2AqcjkZBR77F29IHWfmEnos6XEkqciVy+02xGMfYpiKXLMAOsxcIqI8gA8A+AskqMiVSoWYm+m5HDpOvnKgY5Mc9bPbkJ0qpVTziTFrim+qSqaV6rbpy6j7RS1pj8F6tvNKFoqqNkicVUBVWucskZ0jq1wOXUcG+c6PfzBob5Rs5dUwqw8dUcWyHXkwNm4F97cvLzJQ+4LjncGD952Q+aol0/fipjVR7IZh7DgLAJ6hnuMjAPAsM3+eiJ6Hr8h1YDGMVvV99ErUup9vwlfkOrAg5r2Ddm7ZzYiuAVgCcBjAxh7DDxLu5Pdxgpln3Q9HSjiDmxJ9m5kfGfmN71Dsx/fhnZweieAJxyMRbhfhPH2b7nunYt+9j9si43jsf3hW5ZEIIyUcInqciF4honNEdODCMO6m0wZHxqr6ludXATwGYAXACwCeYOYfve4X7yL0fXoLzPwiEY0D+A6AXwPwWwC2mPmp/g9qmplfN0TldmOUO86jAM4x8wVmbgP4DHoxPQcGzLzKzC/22xUA+rTBZ/rDnkGPmO5ojJJwjgK4pK5X+p8dSOz30wZHSTi7nUBzIFW6pKcN3kkYJeGsADimrhcB7O2/v8twM6cN3kkYJeG8AOA+IjpFRBkAH0YvpufAYIjTBoEhY5tuN0btHf8ggL9GL5X508z85yO7+R0AIvpZAF8D8ANIVe1PoSfnPAvgOPqxTcy8teskdwi85dgjEbzl2CMRPOF4JIInHI9E8ITjkQiecDwSwRPOGwQR/SkR/eHtXsfthiccj0TwhDMEiOiP+3FE/wng/v5nDxPR/xHR94nos6/F0BDRu/ufPU9Ef0lEL93Wxb9J8ISzB4joXei5R94J4DcAvLvf9Y8APsHMb0fPEvwn/c//HsDvMPPPYJeTsO8WeMLZG+8H8Flmrvc92Z8DUAQwxcxf6Y95BsDPEdEUgHFm/t/+5/888tWOCJ5whsOwfpndQkfuSnjC2RtfBfDrRJTvh3v+KoAagG0ien9/zEcAfIWZtwFU+qXugB6Luysx8qqj+w39+OB/BfBd9PLev9bvehLA3xBRAcAFAL/d//yjAP6WiGro1RLawV0I7x2/xSCiMWau9tufRC84/eO3eVm3HH7HufX4FSL6I/Te7RJ6GQx3HfyO45EIXjj2SARPOB6J4AnHIxE84Xgkgiccj0TwhOORCP8Pdb3W4rSKes8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_test, y_test, 12)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'dog'"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes[y_classes[12]]                # correct prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_test, y_test, 13)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'horse'"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes[y_classes[13]]                # correct prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_test, y_test, 15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'frog'"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes[y_classes[15]]                # a mismatch"
   ]
  }
 ],
 "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}