cifar-10Drop.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training set: 50000\n",
      "Testing set: 10000\n"
     ]
    }
   ],
   "source": [
    "# Libraries\n",
    "import tensorflow as tf\n",
    "from keras.datasets import cifar10\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "from sklearn.compose import make_column_transformer\n",
    "from sklearn.preprocessing import MinMaxScaler, OneHotEncoder\n",
    "from keras.utils import np_utils\n",
    "from sklearn.model_selection import train_test_split\n",
    "import numpy as np\n",
    "\n",
    "# Tensorflow aliases\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D\n",
    "from tensorflow.keras.regularizers import l2, l1\n",
    "from tensorflow.keras.utils import plot_model\n",
    "\n",
    "(X_train, y_train),(X_test, y_test) = cifar10.load_data()\n",
    "\n",
    "# Convert to float32 for scaling\n",
    "X_train = X_train.astype('float32')\n",
    "X_test = X_test.astype('float32')\n",
    "\n",
    "# Normalize data by scaling by max\n",
    "X_train = X_train / X_train.max()\n",
    "X_test = X_test / X_test.max()\n",
    "\n",
    "# One-hot encode target labels and reshape\n",
    "y_train = np_utils.to_categorical(y_train.transpose()).reshape(50000,10)\n",
    "y_test = np_utils.to_categorical(y_test.transpose()).reshape(10000,10)\n",
    "input_shape = (32,32,3)\n",
    "\n",
    "print(f\"Training set: {len(X_train)}\\nTesting set: {len(X_test)}\")\n",
    "\n",
    "def build_cnn2(activation='relu', kernel_size=(3,3), pool_size=(2,2), regularizer=None, lam=0.01, summarize=False, dropout=False, drop_prob1=0.2, drop_prob2=0.2):\n",
    "  model = Sequential()\n",
    "  \n",
    "  if regularizer == None:\n",
    "    # Convolutional layers\n",
    "    model.add(Conv2D(filters=32, kernel_size=kernel_size, activation=activation, input_shape=(32, 32, 3)))\n",
    "    model.add(MaxPool2D(pool_size))\n",
    "\n",
    "    model.add(Conv2D(filters=64, kernel_size=kernel_size, activation=activation))\n",
    "    model.add(MaxPool2D(pool_size))\n",
    "    if dropout:\n",
    "      model.add(Dropout(drop_prob1))\n",
    "\n",
    "    model.add(Conv2D(filters=128, kernel_size=kernel_size, activation=activation))\n",
    "    model.add(MaxPool2D(pool_size))\n",
    "    if dropout:\n",
    "      model.add(Dropout(drop_prob2))\n",
    "\n",
    "    # Fully connected layers\n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(512, activation=activation))\n",
    "    if dropout:\n",
    "      model.add(Dropout(drop_prob2))\n",
    "\n",
    "    # Output layer\n",
    "    model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "    model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "    \n",
    "    if summarize:\n",
    "      model.build()\n",
    "      model.summary()\n",
    "    return model\n",
    "  else:\n",
    "    # Convolutional layers\n",
    "    model.add(Conv2D(filters=32, kernel_size=kernel_size, kernel_regularizer=regularizer(lam), activation=activation, input_shape=(32, 32, 3)))\n",
    "    model.add(MaxPool2D(pool_size))\n",
    "\n",
    "    model.add(Conv2D(filters=64, kernel_size=kernel_size, kernel_regularizer=regularizer(lam), activation=activation))\n",
    "    model.add(MaxPool2D(pool_size))\n",
    "    if dropout:\n",
    "      model.add(Dropout(drop_prob1))\n",
    "\n",
    "    model.add(Conv2D(filters=128, kernel_size=kernel_size, kernel_regularizer=regularizer(lam), activation=activation))\n",
    "    model.add(MaxPool2D(pool_size))\n",
    "    if dropout:\n",
    "      model.add(Dropout(drop_prob2))\n",
    "\n",
    "    # Fully connected layers\n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(512, activation=activation, kernel_regularizer=regularizer(lam)))\n",
    "    if dropout:\n",
    "      model.add(Dropout(drop_prob2))\n",
    "\n",
    "    # Output layer\n",
    "    model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "    model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "\n",
    "    if summarize:\n",
    "      model.build()\n",
    "      model.summary()\n",
    "    return model\n",
    "\n",
    "# Helper functions\n",
    "def show_acc(h, e):\n",
    "  plt.plot(range(e), h.history['accuracy'], label='Training')\n",
    "  plt.plot(range(e), h.history['val_accuracy'], label='Validation')\n",
    "  plt.ylim([0, 1])\n",
    "  plt.legend()\n",
    "  plt.show()\n",
    "\n",
    "def build_and_plot(activation='relu', kernel_size=(3,3), pool_size=(2,2), regularizer=None, lam=0.01, dropout=False, drop_prob1=0.2, drop_prob2=0.2, summarize=False, e=100, verbose=0):\n",
    "  model = build_cnn2(activation=activation, kernel_size=kernel_size, pool_size=pool_size, regularizer=regularizer, lam=lam, dropout=dropout, drop_prob1=drop_prob1, drop_prob2=drop_prob2, summarize=summarize)\n",
    "  history = model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=e, verbose=verbose)\n",
    "  show_acc(history, e)\n",
    "  print(\"Max accuracy: \", np.max(history.history['accuracy']))\n",
    "  print(\"Min accuracy: \", np.min(history.history['accuracy']))\n",
    "  print(\"Test performance \", model.evaluate(X_test, y_test))\n",
    "  return model, history"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Same dropout probabilities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-04-17 22:29:33.093237: W tensorflow/tsl/platform/profile_utils/cpu_utils.cc:128] Failed to get CPU frequency: 0 Hz\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Max accuracy:  0.9642800092697144\n",
      "Min accuracy:  0.47975999116897583\n",
      "313/313 [==============================] - 1s 3ms/step - loss: 2.1989 - accuracy: 0.7081\n",
      "Test performance  [2.1989264488220215, 0.7081000208854675]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Max accuracy:  0.944920003414154\n",
      "Min accuracy:  0.4681600034236908\n",
      "313/313 [==============================] - 1s 3ms/step - loss: 1.4189 - accuracy: 0.7456\n",
      "Test performance  [1.4188770055770874, 0.7455999851226807]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Max accuracy:  0.8870000243186951\n",
      "Min accuracy:  0.4459399878978729\n",
      "313/313 [==============================] - 1s 3ms/step - loss: 0.8080 - accuracy: 0.7735\n",
      "Test performance  [0.807950496673584, 0.7735000252723694]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Max accuracy:  0.806879997253418\n",
      "Min accuracy:  0.41756001114845276\n",
      "313/313 [==============================] - 1s 3ms/step - loss: 0.6571 - accuracy: 0.7829\n",
      "Test performance  [0.657109260559082, 0.7828999757766724]\n"
     ]
    },
    {
     "ename": "TypeError",
     "evalue": "build_and_plot() got an unexpected keyword argument 'drop_pro1b'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m/Users/alexgeorgiev/Desktop/STOR 590/Final Project/FNN-Regularization/cifar-10Drop.ipynb Cell 3\u001b[0m in \u001b[0;36m5\n\u001b[1;32m      <a href='vscode-notebook-cell:/Users/alexgeorgiev/Desktop/STOR%20590/Final%20Project/FNN-Regularization/cifar-10Drop.ipynb#W2sZmlsZQ%3D%3D?line=2'>3</a>\u001b[0m drop_2, drop_2_history \u001b[39m=\u001b[39m build_and_plot(dropout\u001b[39m=\u001b[39m\u001b[39mTrue\u001b[39;00m, drop_prob1\u001b[39m=\u001b[39m\u001b[39m0.2\u001b[39m, drop_prob2\u001b[39m=\u001b[39m\u001b[39m0.2\u001b[39m, e\u001b[39m=\u001b[39m\u001b[39m100\u001b[39m)\n\u001b[1;32m      <a href='vscode-notebook-cell:/Users/alexgeorgiev/Desktop/STOR%20590/Final%20Project/FNN-Regularization/cifar-10Drop.ipynb#W2sZmlsZQ%3D%3D?line=3'>4</a>\u001b[0m drop_3, drop_3_history \u001b[39m=\u001b[39m build_and_plot(dropout\u001b[39m=\u001b[39m\u001b[39mTrue\u001b[39;00m, drop_prob1\u001b[39m=\u001b[39m\u001b[39m0.3\u001b[39m, drop_prob2\u001b[39m=\u001b[39m\u001b[39m0.3\u001b[39m, e\u001b[39m=\u001b[39m\u001b[39m100\u001b[39m)\n\u001b[0;32m----> <a href='vscode-notebook-cell:/Users/alexgeorgiev/Desktop/STOR%20590/Final%20Project/FNN-Regularization/cifar-10Drop.ipynb#W2sZmlsZQ%3D%3D?line=4'>5</a>\u001b[0m drop_4, drop_4_history \u001b[39m=\u001b[39m build_and_plot(dropout\u001b[39m=\u001b[39;49m\u001b[39mTrue\u001b[39;49;00m, drop_pro1b\u001b[39m=\u001b[39;49m\u001b[39m0.4\u001b[39;49m, drop_prob2\u001b[39m=\u001b[39;49m\u001b[39m0.4\u001b[39;49m, e\u001b[39m=\u001b[39;49m\u001b[39m100\u001b[39;49m)\n",
      "\u001b[0;31mTypeError\u001b[0m: build_and_plot() got an unexpected keyword argument 'drop_pro1b'"
     ]
    }
   ],
   "source": [
    "# drop_05, drop_05_history = build_and_plot(dropout=True, drop_prob1=0.05, drop_prob2=0.05, e=100)\n",
    "# drop_1, drop_1_history = build_and_plot(dropout=True, drop_prob1=0.1, drop_prob2=0.1, e=100)\n",
    "# drop_2, drop_2_history = build_and_plot(dropout=True, drop_prob1=0.2, drop_prob2=0.2, e=100)\n",
    "# drop_3, drop_3_history = build_and_plot(dropout=True, drop_prob1=0.3, drop_prob2=0.3, e=100)\n",
    "drop_4, drop_4_history = build_and_plot(dropout=True, drop_prob1=0.4, drop_prob2=0.4, e=100)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Smaller then larger dropout probabilites"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from keras.backend import image_data_format\n",
    "# image_data_format()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "drop_s05, drop_s05_history = build_and_plot(dropout=True, drop_prob1=0.05, drop_prob2=0.1, e=100)\n",
    "drop_s1, drop_s1_history = build_and_plot(dropout=True, drop_prob1=0.1, drop_prob2=0.2, e=100)\n",
    "drop_s2, drop_s2_history = build_and_plot(dropout=True, drop_prob1=0.2, drop_prob2=0.3, e=100)\n",
    "drop_s3, drop_s3_history = build_and_plot(dropout=True, drop_prob1=0.4, drop_prob2=0.5, e=100)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Larger then smaller dropout probabilities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "drop_b05, drop_b05_history = build_and_plot(dropout=True, drop_prob1=0.1, drop_prob2=0.05, e=100)\n",
    "drop_b1, drop_b1_history = build_and_plot(dropout=True, drop_prob1=0.2, drop_prob2=0.1, e=100)\n",
    "drop_b2, drop_b2_history = build_and_plot(dropout=True, drop_prob1=0.3, drop_prob2=0.2, e=100)\n",
    "drop_b3, drop_b3_history = build_and_plot(dropout=True, drop_prob1=0.4, drop_prob2=0.3, e=100)"
   ]
  }
 ],
 "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": "3.10.9"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}