MNIST project.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "97a4d702-2a4e-4b4b-ae12-93d653a84626",
   "metadata": {},
   "source": [
    "Exercises:\n",
    "Train a deep MLP on the MNIST dataset (you can load it using\n",
    "keras.datasets.mnist.load_data(). See if you can get over\n",
    "98% precision. Try searching for the optimal learning rate by\n",
    "using the approach presented in this chapter (i.e., by growing the\n",
    "learning rate exponentially, plotting the error, and finding the\n",
    "point where the error shoots up). Try adding all the bells and\n",
    "whistles—save checkpoints, use early stopping, and plot learning\n",
    "curves using TensorBoard."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "99e1ec91-7787-4514-8f4b-4beb366b8aa5",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "%load_ext tensorboard"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5c9f285b-6403-4d31-a2a8-1f3f826298e6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# imports\n",
    "import os\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "from datetime import datetime\n",
    "from packaging import version"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "ace4a97a-2008-451a-afc3-17940d6c8ef8",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "(X_train_full, y_train_full), (X_test, y_test) = keras.datasets.mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "33e07488-6b2a-4de2-b2d2-82ec0ec63e2e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(60000, 28, 28)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_full.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "ad145dc9-bc42-4dcd-a75c-942fdf26189f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dtype('uint8')"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_full.dtype"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "b83e1dee-9696-4227-a527-d43422f3d7fc",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_valid, X_train = X_train_full [:5000]/255.0, X_train_full[5000:]/255.0\n",
    "y_valid, y_train = y_train_full [:5000], y_train_full[5000:]\n",
    "X_test = X_test/255.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b267f493-b5bc-4849-803a-d458a061b87f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function matplotlib.pyplot.show(close=None, block=None)>"
      ]
     },
     "execution_count": 7,
     "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": [
    "plt.imshow(X_train[0], cmap=\"binary\")\n",
    "plt.axis('off')\n",
    "plt.show"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "c739d976-ab30-49ec-95bd-7387a8b723f8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(55000, 28, 28)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "b461f7ea-7431-41ec-9480-3e82b48bde2c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(55000,)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "7655ccf5-e9f2-48e5-b822-b467cfe19053",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5000, 28, 28)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_valid.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "72c950ac-dfa5-4236-bdc9-28e32da21860",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 28, 28)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "7bdd1ee5-7f64-4e9a-842b-fd0816e963e6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x345.6 with 40 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "n_rows = 4\n",
    "n_cols = 10\n",
    "plt.figure(figsize=(n_cols * 1.2, n_rows * 1.2))\n",
    "for row in range(n_rows):\n",
    "    for col in range(n_cols):\n",
    "        index = n_cols * row + col\n",
    "        plt.subplot(n_rows, n_cols, index + 1)\n",
    "        plt.imshow(X_train[index], cmap=\"binary\", interpolation=\"nearest\")\n",
    "        plt.axis('off')\n",
    "        plt.title(y_train[index], fontsize=12)\n",
    "plt.subplots_adjust(wspace=0.2, hspace=0.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "d5223f96-8fe5-447b-8bf5-768042cb9a7c",
   "metadata": {},
   "outputs": [],
   "source": [
    "K = keras.backend\n",
    "\n",
    "class ExponentialLearningRate (keras.callbacks.Callback):\n",
    "    def __init__(self, factor):\n",
    "        self.factor = factor\n",
    "        self.rate = []\n",
    "        self.losses = []\n",
    "    def on_batch_end (self, batch, logs):\n",
    "        self.rates.append(K.get_value(seld.model.optimizer.learning_rate))\n",
    "        seld.losses.append(logs[\"loss\"])\n",
    "        K.set_value(self.model.optimizer.learning_rate, self.model.optimizer.learning_rate * self.factor)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "409b9c02-df23-4b56-a482-b501b1fdd78c",
   "metadata": {},
   "outputs": [],
   "source": [
    "keras.backend.clear_session()\n",
    "np.random.seed(42)\n",
    "tf.random.set_seed(42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "aa21a5bd-37f3-46ca-8651-4c8266267c3d",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.Sequential ([\n",
    "    keras.layers.Flatten(input_shape = [28, 28]),\n",
    "    keras.layers.Dense(300, activation=\"relu\"),\n",
    "    keras.layers.Dense(100, activation=\"relu\"),\n",
    "    keras.layers.Dense(10, activation=\"softmax\")\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "55db05b7-06f2-41bc-8276-88e818162451",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(\n",
    "    loss = \"sparse_categorical_crossentropy\",\n",
    "    optimizer = keras.optimizers.SGD(learning_rate = 3e-1),\n",
    "    metrics = [\"accuracy\"]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "4e626b3f-f249-4f21-9813-9ba47c5ce264",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'.\\\\my_mnist_logs\\\\run_001'"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "run_index = 1 # increment this at every run\n",
    "run_logdir = os.path.join(os.curdir, \"my_mnist_logs\", \"run_{:03d}\".format(run_index))\n",
    "run_logdir"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "e58f1101-7519-40c7-b2b5-a6a8fca9e5ac",
   "metadata": {},
   "outputs": [],
   "source": [
    "checkpoint_cb = keras.callbacks.ModelCheckpoint(\"my_keras_model.h5\", save_best_only=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "0d66b83e-00e5-4c9a-9de4-cf8c81a15453",
   "metadata": {},
   "outputs": [],
   "source": [
    "early_stopping_cb = keras.callbacks.EarlyStopping(patience = 10, restore_best_weights=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "aa272c32-e1f7-4dfd-af2f-527f9d4272c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "tensorboard_cb = keras.callbacks.TensorBoard(run_logdir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "33edf334-73b9-4e34-9cba-232c7787df2d",
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "1719/1719 [==============================] - 5s 3ms/step - loss: 0.2200 - accuracy: 0.9390 - val_loss: 0.1004 - val_accuracy: 0.9704\n",
      "Epoch 2/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0925 - accuracy: 0.9713 - val_loss: 0.0870 - val_accuracy: 0.9762\n",
      "Epoch 3/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0682 - accuracy: 0.9785 - val_loss: 0.0818 - val_accuracy: 0.9774\n",
      "Epoch 4/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0502 - accuracy: 0.9840 - val_loss: 0.0921 - val_accuracy: 0.9750\n",
      "Epoch 5/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0391 - accuracy: 0.9870 - val_loss: 0.0983 - val_accuracy: 0.9748\n",
      "Epoch 6/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0316 - accuracy: 0.9898 - val_loss: 0.0934 - val_accuracy: 0.9746\n",
      "Epoch 7/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0251 - accuracy: 0.9921 - val_loss: 0.0856 - val_accuracy: 0.9800\n",
      "Epoch 8/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0205 - accuracy: 0.9938 - val_loss: 0.0910 - val_accuracy: 0.9780\n",
      "Epoch 9/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0176 - accuracy: 0.9943 - val_loss: 0.0815 - val_accuracy: 0.9804\n",
      "Epoch 10/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0173 - accuracy: 0.9942 - val_loss: 0.0852 - val_accuracy: 0.9814\n",
      "Epoch 11/100\n",
      "1719/1719 [==============================] - 4s 3ms/step - loss: 0.0158 - accuracy: 0.9949 - val_loss: 0.0868 - val_accuracy: 0.9822\n",
      "Epoch 12/100\n",
      "1719/1719 [==============================] - 4s 3ms/step - loss: 0.0175 - accuracy: 0.9946 - val_loss: 0.1239 - val_accuracy: 0.9758\n",
      "Epoch 13/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0174 - accuracy: 0.9944 - val_loss: 0.0826 - val_accuracy: 0.9794\n",
      "Epoch 14/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0105 - accuracy: 0.9965 - val_loss: 0.0904 - val_accuracy: 0.9832\n",
      "Epoch 15/100\n",
      "1719/1719 [==============================] - 4s 3ms/step - loss: 0.0051 - accuracy: 0.9983 - val_loss: 0.1055 - val_accuracy: 0.9820\n",
      "Epoch 16/100\n",
      "1719/1719 [==============================] - 4s 3ms/step - loss: 0.0123 - accuracy: 0.9962 - val_loss: 0.1156 - val_accuracy: 0.9762\n",
      "Epoch 17/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0073 - accuracy: 0.9977 - val_loss: 0.1005 - val_accuracy: 0.9844\n",
      "Epoch 18/100\n",
      "1719/1719 [==============================] - 4s 3ms/step - loss: 0.0038 - accuracy: 0.9988 - val_loss: 0.1233 - val_accuracy: 0.9796\n",
      "Epoch 19/100\n",
      "1719/1719 [==============================] - 4s 2ms/step - loss: 0.0042 - accuracy: 0.9988 - val_loss: 0.0951 - val_accuracy: 0.9830\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x21fb4ea4c40>"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(\n",
    "    X_train, y_train, epochs=100,\n",
    "    validation_data = (X_valid, y_valid),\n",
    "    callbacks = [checkpoint_cb, early_stopping_cb, tensorboard_cb]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "e441b091-f874-4abf-9c00-c3f135781e2e",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.load_model(\"my_keras_model.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "7149a272-a6c0-42c7-9e2e-5d1ad994914a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 1s 3ms/step - loss: 0.0885 - accuracy: 0.9801\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.08849705010652542, 0.9800999760627747]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "bed86f87-6121-4eff-8e96-bbf426dc6df8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "      <iframe id=\"tensorboard-frame-4716abc5e8f44088\" width=\"100%\" height=\"800\" frameborder=\"0\">\n",
       "      </iframe>\n",
       "      <script>\n",
       "        (function() {\n",
       "          const frame = document.getElementById(\"tensorboard-frame-4716abc5e8f44088\");\n",
       "          const url = new URL(\"/\", window.location);\n",
       "          const port = 6006;\n",
       "          if (port) {\n",
       "            url.port = port;\n",
       "          }\n",
       "          frame.src = url;\n",
       "        })();\n",
       "      </script>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%tensorboard --logdir=./my_mnist_logs --port=6006"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ebea20a1-2e9d-4c1e-a395-5ed34452769e",
   "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": "3.8.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}