TF2_Fashion_MNIST-checkpoint.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'2.8.0'"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "tf.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from tensorflow.keras.layers import Input, Conv2D, Dense, Flatten, Dropout\n",
    "from tensorflow.keras.models import Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train.shape: (60000, 28, 28)\n"
     ]
    }
   ],
   "source": [
    "# Load in the data\n",
    "fashion_mnist = tf.keras.datasets.fashion_mnist\n",
    "\n",
    "(x_train, y_train), (x_test, y_test) = fashion_mnist.load_data()\n",
    "x_train, x_test = x_train / 255.0, x_test / 255.0\n",
    "print(\"x_train.shape:\", x_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(60000, 28, 28, 1)\n"
     ]
    }
   ],
   "source": [
    "# the data is only 2D\n",
    "# convolution expects height x width x color\n",
    "x_train = np.expand_dims(x_train, -1)\n",
    "x_test = np.expand_dims(x_test, -1)\n",
    "print(x_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number of classes: 10\n"
     ]
    }
   ],
   "source": [
    "# number of classes\n",
    "K = len(set(y_train))\n",
    "print(\"number of classes:\", K)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_1\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_2 (InputLayer)        [(None, 28, 28, 1)]       0         \n",
      "                                                                 \n",
      " conv2d_3 (Conv2D)           (None, 13, 13, 32)        320       \n",
      "                                                                 \n",
      " conv2d_4 (Conv2D)           (None, 6, 6, 64)          18496     \n",
      "                                                                 \n",
      " conv2d_5 (Conv2D)           (None, 2, 2, 128)         73856     \n",
      "                                                                 \n",
      " flatten_1 (Flatten)         (None, 512)               0         \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 512)               0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 512)               262656    \n",
      "                                                                 \n",
      " dropout_3 (Dropout)         (None, 512)               0         \n",
      "                                                                 \n",
      " dense_3 (Dense)             (None, 10)                5130      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 360,458\n",
      "Trainable params: 360,458\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Build the model using the functional API\n",
    "i = Input(shape=x_train[0].shape)\n",
    "x = Conv2D(32, (3, 3), strides=2, activation='relu')(i)\n",
    "x = Conv2D(64, (3, 3), strides=2, activation='relu')(x)\n",
    "x = Conv2D(128, (3, 3), strides=2, activation='relu')(x)\n",
    "x = Flatten()(x)\n",
    "x = Dropout(0.2)(x)\n",
    "x = Dense(512, activation='relu')(x)\n",
    "x = Dropout(0.2)(x)\n",
    "x = Dense(K, activation='softmax')(x)\n",
    "\n",
    "model = Model(i, x)\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/15\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.5279 - accuracy: 0.8056 - val_loss: 0.3869 - val_accuracy: 0.8523\n",
      "Epoch 2/15\n",
      "1875/1875 [==============================] - 12s 7ms/step - loss: 0.3656 - accuracy: 0.8628 - val_loss: 0.3640 - val_accuracy: 0.8614\n",
      "Epoch 3/15\n",
      "1875/1875 [==============================] - 11s 6ms/step - loss: 0.3163 - accuracy: 0.8807 - val_loss: 0.3250 - val_accuracy: 0.8798\n",
      "Epoch 4/15\n",
      "1875/1875 [==============================] - 11s 6ms/step - loss: 0.2847 - accuracy: 0.8935 - val_loss: 0.3272 - val_accuracy: 0.8812\n",
      "Epoch 5/15\n",
      "1875/1875 [==============================] - 11s 6ms/step - loss: 0.2550 - accuracy: 0.9024 - val_loss: 0.3125 - val_accuracy: 0.8892\n",
      "Epoch 6/15\n",
      "1875/1875 [==============================] - 12s 6ms/step - loss: 0.2351 - accuracy: 0.9122 - val_loss: 0.2925 - val_accuracy: 0.8953\n",
      "Epoch 7/15\n",
      "1875/1875 [==============================] - 11s 6ms/step - loss: 0.2162 - accuracy: 0.9180 - val_loss: 0.3069 - val_accuracy: 0.8933\n",
      "Epoch 8/15\n",
      "1875/1875 [==============================] - 12s 7ms/step - loss: 0.1983 - accuracy: 0.9244 - val_loss: 0.2975 - val_accuracy: 0.9000\n",
      "Epoch 9/15\n",
      "1875/1875 [==============================] - 12s 6ms/step - loss: 0.1854 - accuracy: 0.9298 - val_loss: 0.3091 - val_accuracy: 0.9013\n",
      "Epoch 10/15\n",
      "1875/1875 [==============================] - 12s 6ms/step - loss: 0.1747 - accuracy: 0.9340 - val_loss: 0.3135 - val_accuracy: 0.8954\n",
      "Epoch 11/15\n",
      "1875/1875 [==============================] - 11s 6ms/step - loss: 0.1634 - accuracy: 0.9379 - val_loss: 0.3284 - val_accuracy: 0.8982\n",
      "Epoch 12/15\n",
      "1875/1875 [==============================] - 11s 6ms/step - loss: 0.1512 - accuracy: 0.9429 - val_loss: 0.3294 - val_accuracy: 0.9018\n",
      "Epoch 13/15\n",
      "1875/1875 [==============================] - 11s 6ms/step - loss: 0.1471 - accuracy: 0.9440 - val_loss: 0.3478 - val_accuracy: 0.8964\n",
      "Epoch 14/15\n",
      "1875/1875 [==============================] - 11s 6ms/step - loss: 0.1377 - accuracy: 0.9473 - val_loss: 0.3381 - val_accuracy: 0.9021\n",
      "Epoch 15/15\n",
      "1875/1875 [==============================] - 11s 6ms/step - loss: 0.1304 - accuracy: 0.9511 - val_loss: 0.3656 - val_accuracy: 0.8999\n"
     ]
    }
   ],
   "source": [
    "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
    "r = model.fit(x_train, y_train, validation_data=(x_test, y_test), epochs=15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x14ea14544c0>"
      ]
     },
     "execution_count": 20,
     "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": [
    "# Plot loss per iteration\n",
    "import matplotlib.pyplot as plt\n",
    "plt.plot(r.history['loss'], label='loss')\n",
    "plt.plot(r.history['val_loss'], label='val_loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x14ea1521490>"
      ]
     },
     "execution_count": 23,
     "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": [
    "# Plot accuracy per iteration\n",
    "import matplotlib.pyplot as plt\n",
    "plt.plot(r.history['accuracy'], label='accuracy')\n",
    "plt.plot(r.history['val_accuracy'], label='val_accuracy')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Consusion matrix, without normalization\n",
      "[[874   1  23  18   2   1  77   0   4   0]\n",
      " [  2 976   1  14   3   0   2   0   2   0]\n",
      " [ 14   1 884   9  33   0  55   0   4   0]\n",
      " [ 21   3  17 913  12   0  31   0   3   0]\n",
      " [  0   2 114  36 782   0  62   0   4   0]\n",
      " [  0   0   0   1   0 971   0  17   0  11]\n",
      " [124   1  72  17  78   1 695   0  12   0]\n",
      " [  0   0   0   0   0   2   0 949   0  49]\n",
      " [  2   0   4   4   2   3   0   5 979   1]\n",
      " [  1   0   0   0   0   5   0  17   1 976]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "import itertools\n",
    "def plt_confusion_matrix(cm, classes, \n",
    "                          normalize=False, \n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Blues):\n",
    "  '''\n",
    "  This function prints and plots the confusion matrix.\n",
    "  Normalization can be appled by setting `normalize=True`.\n",
    "  '''\n",
    "  if normalize:\n",
    "    cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "    print('normalized confusion matrix')\n",
    "  else:\n",
    "    print('Consusion matrix, without normalization')\n",
    "  print(cm)\n",
    "\n",
    "  plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "  plt.title(title)\n",
    "  plt.colorbar()\n",
    "  tick_marks = np.arange(len(classes))\n",
    "  plt.xticks(tick_marks, classes, rotation=45)\n",
    "  plt.yticks(tick_marks, classes)\n",
    "\n",
    "  fmt = '.2f' if normalize else 'd'\n",
    "  thresh = cm.max() / 2.\n",
    "  for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "    plt.text(j, i, format(cm[i, j], fmt),\n",
    "             horizontalalignment=\"center\",\n",
    "             color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "  \n",
    "  plt.tight_layout()\n",
    "  plt.ylabel('True label')\n",
    "  plt.xlabel('Predicted label')\n",
    "  plt.show()\n",
    "\n",
    "p_test = model.predict(x_test).argmax(axis=1)\n",
    "cm = confusion_matrix(y_test, p_test)\n",
    "plt_confusion_matrix(cm, list(range(10)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Label mapping\n",
    "labels = '''T-shirt/top\n",
    "Trouser\n",
    "Pullover\n",
    "Dress\n",
    "Coat\n",
    "Sandal\n",
    "Shirt\n",
    "Sneaker\n",
    "Bag\n",
    "Ankle boot'''.split()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'True label: Coat, Predicted: Pullover')"
      ]
     },
     "execution_count": 26,
     "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": [
    "# Show some misclassified examples\n",
    "misclassified_idx = np.where(p_test != y_test)[0]\n",
    "i = np.random.choice(misclassified_idx)\n",
    "plt.imshow(x_test[i].reshape(28, 28), cmap='gray')\n",
    "plt.title(\"True label: %s, Predicted: %s\" % (labels[y_test[i]], labels[p_test[i]]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}