CancerDetection_model_training_1level_transfer.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Mo4WETmo_ZVI"
   },
   "source": [
    "# Install the library and toolboxs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "umoPuj0VjLFS"
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import os\n",
    "import pickle\n",
    "import random\n",
    "import shutil\n",
    "import numpy as np\n",
    "from PIL import Image\n",
    "from skimage.color import rgb2gray\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "!pip install ipython-autotime\n",
    "\n",
    "%load_ext autotime\n",
    "%tensorflow_version 2.x\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "# Check GPU\n",
    "!nvidia-smi"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Sm2ZmN-wBGHn"
   },
   "source": [
    "# Load preprocessed data from drive and split to training and validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 138
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 182465,
     "status": "ok",
     "timestamp": 1587770332708,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "ML5e6F3rmIz1",
    "outputId": "d2b40081-855a-455f-ded5-014380a34b07"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3aietf%3awg%3aoauth%3a2.0%3aoob&response_type=code&scope=email%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdocs.test%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive.photos.readonly%20https%3a%2f%2fwww.googleapis.com%2fauth%2fpeopleapi.readonly\n",
      "\n",
      "Enter your authorization code:\n",
      "··········\n",
      "Mounted at /content/drive/\n",
      "time: 2min 50s\n"
     ]
    }
   ],
   "source": [
    "# Mount google drive\n",
    "from google.colab import drive\n",
    "drive.mount('/content/drive/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 27093,
     "status": "ok",
     "timestamp": 1587770359809,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "gYkvXBXeBO46",
    "outputId": "b12a8394-1797-4449-911a-bf3abda8a7d0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 25.9 s\n"
     ]
    }
   ],
   "source": [
    "# # Load the balanced dataset\n",
    "# level = 3\n",
    "# variable_folder =  \"/content/drive/My Drive/Dataset\"\n",
    "# file_name = \"dataset_preprocessed\" + \"_level\" + str(level) + \"_balanced.pkl\"\n",
    "# with open(os.path.join(variable_folder, file_name), 'rb') as f:\n",
    "#     dataset_patches, dataset_label = pickle.load(f)\n",
    "\n",
    "# OR Load the unbalanced dataset\n",
    "level = 3\n",
    "variable_folder =  \"/content/drive/My Drive/Dataset\"\n",
    "file_name = \"dataset_preprocessed\" + \"_level\" + str(level) + \"_unbalanced.pkl\"\n",
    "with open(os.path.join(variable_folder, file_name), 'rb') as f:\n",
    "    dataset_patches, dataset_label = pickle.load(f)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 27087,
     "status": "ok",
     "timestamp": 1587770359810,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "P-I2z1o-tKMc",
    "outputId": "1d119c13-b441-453f-ee9c-220a99eebfe6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 54.5 ms\n"
     ]
    }
   ],
   "source": [
    "# Split and get train data\n",
    "patches_with_tumor = [dataset_patches[ind] for ind in dataset_label if dataset_label[ind]==1]\n",
    "patches_without_tumor = [dataset_patches[ind] for ind in dataset_label if dataset_label[ind]==0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 101
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 31141,
     "status": "ok",
     "timestamp": 1587770363876,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "mNQPUR4tDY4q",
    "outputId": "691f52a0-5ce2-445d-a6d8-8ce2aab6a16a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Size of the train set: 347503\n",
      "Size of the validation set: 86876\n",
      "Size of the train data sample: (32, 32, 3)\n",
      "Size of the validation data sample: (32, 32, 3)\n",
      "time: 3.98 s\n"
     ]
    }
   ],
   "source": [
    "# Split train and validation set\n",
    "X_train, X_val, y_train, y_val = train_test_split(\n",
    "...     dataset_patches, dataset_label, test_size=0.2, random_state=42)\n",
    "X_train = (np.array(X_train).astype('float32') / 255.) * 2.0 - 1\n",
    "X_val = (np.array(X_val).astype('float32') / 255.) * 2.0 - 1.0\n",
    "\n",
    "# Check the split\n",
    "print(\"Size of the train set: {:d}\".format(len(y_train)))\n",
    "print(\"Size of the validation set: {:d}\".format(len(y_val)))\n",
    "print(\"Size of the train data sample: \" + str(X_train[0].shape))\n",
    "print(\"Size of the validation data sample: \" + str(X_val[0].shape))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "pIj6DAuMNOtK"
   },
   "source": [
    "### Function to plot the history of training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 31131,
     "status": "ok",
     "timestamp": 1587770363877,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "FV2NAAtGLHS4",
    "outputId": "99be392d-2a65-4142-fe4a-815d872a32d2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 14.2 ms\n"
     ]
    }
   ],
   "source": [
    "# Function to plot the history of training\n",
    "def plot_training(history):\n",
    "  \n",
    "  # The history object contains results on the training and test\n",
    "  # sets for each epoch\n",
    "  acc = history.history['accuracy']\n",
    "  val_acc = history.history['val_accuracy']\n",
    "  loss = history.history['loss']\n",
    "  val_loss = history.history['val_loss']\n",
    "\n",
    "  # Plot the result\n",
    "  epochs = range(len(acc))\n",
    "  fig = plt.figure(figsize=(15, 5))\n",
    "  ax1 = fig.add_subplot(121)\n",
    "  plt.title('Max val accuracy: {0:.4f}'.format(max(val_acc)))\n",
    "  plt.plot(epochs, acc, color='blue', label='Train')\n",
    "  plt.plot(epochs, val_acc, color='orange', label='Val')\n",
    "  plt.xlabel('Epoch')\n",
    "  plt.ylabel('Accuracy')\n",
    "  plt.legend()\n",
    "\n",
    "  ax1 = fig.add_subplot(122)\n",
    "  plt.title('Min val loss: {0:.4f}'.format(min(val_loss)))\n",
    "  plt.plot(epochs, loss, color='blue', label='Train')\n",
    "  plt.plot(epochs, val_loss, color='orange', label='Val')\n",
    "  plt.xlabel('Epoch')\n",
    "  plt.ylabel('Loss')\n",
    "  plt.legend()\n",
    "  if max(val_loss) > 3:\n",
    "      plt.ylim(0, 3)\n",
    "\n",
    "  # Save the figure\n",
    "  path = fig_save_path + \".png\"\n",
    "  plt.tight_layout()\n",
    "  plt.savefig(path, format=\"png\", dpi=300)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "S-2YALtORhPR"
   },
   "source": [
    "# InceptionV3"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Oqfb507nMU7j"
   },
   "source": [
    "## Method 1 \n",
    "* Resize the data patches to match required input size of InceptionV3 (at least 75x75)\n",
    "* Freeze up to layer 40\n",
    "* No data augmentation\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "eyn4hWezJwqw"
   },
   "source": [
    "### Load InceptionV3 and visualize the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VrqbpLSrRoId"
   },
   "outputs": [],
   "source": [
    "# Load the base model\n",
    "base_model = tf.keras.applications.InceptionV3(input_shape=[75, 75, 3], include_top=False, weights='imagenet')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "i9CuOpvEg975"
   },
   "outputs": [],
   "source": [
    "# Plot the base model\n",
    "from tensorflow.keras.utils import plot_model\n",
    "plot_model(base_model, to_file='model.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "qwsCsKVPh-o0"
   },
   "outputs": [],
   "source": [
    "# List all the layers\n",
    "base_model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "iu7fyoXZNhWv"
   },
   "source": [
    "### Resize the data and construct the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 7950,
     "status": "ok",
     "timestamp": 1587587643660,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "N-bmeQRZoP5k",
    "outputId": "6a2c7787-03c4-4e2f-fd46-46d938b920f1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 7.47 s\n"
     ]
    }
   ],
   "source": [
    "# Resize the data patches to match required input size of InceptionV3\n",
    "X_train_resize = tf.image.resize(\n",
    "                        X_train,\n",
    "                        [75, 75],\n",
    "                        method='bilinear',\n",
    "                        antialias=True)\n",
    "\n",
    "X_val_resize = tf.image.resize(\n",
    "                        X_val,\n",
    "                        [75, 75],\n",
    "                        method='bilinear',\n",
    "                        antialias=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 4041,
     "status": "ok",
     "timestamp": 1587588087512,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "c0lGMxdqLkju",
    "outputId": "747ea785-223c-422c-a77b-4128abc8beb4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 3.71 s\n"
     ]
    }
   ],
   "source": [
    "# Freeze all weights\n",
    "base_model.trainable = False \n",
    "\n",
    "# Add last layers\n",
    "global_average_layer = tf.keras.layers.GlobalAveragePooling2D()\n",
    "prediction_layer = tf.keras.layers.Dense(1, activation='sigmoid')\n",
    "\n",
    "# Build a new model reusing the pretrained base\n",
    "model = tf.keras.Sequential([\n",
    "  base_model,\n",
    "  global_average_layer,\n",
    "  prediction_layer\n",
    "])\n",
    "\n",
    "# Compile the model\n",
    "early_stopping_cb = keras.callbacks.EarlyStopping(patience=50, restore_best_weights=True, monitor='val_loss')\n",
    "optimizer = keras.optimizers.Adam(learning_rate=0.005)\n",
    "model.compile(optimizer=optimizer,\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "bH_mRrOrPyx_"
   },
   "outputs": [],
   "source": [
    "# Check the model\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "CiwsP_dSKTIu"
   },
   "source": [
    "### Train the last layer first"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "GTYW0isgJEao"
   },
   "outputs": [],
   "source": [
    "# Try training on a single batch first\n",
    "train_steps = 50\n",
    "loss = []\n",
    "accuracy = []\n",
    "for i in range(train_steps):\n",
    "  metrics = model.train_on_batch(X_train_resize[:128, :, :, :], y_train[:128])\n",
    "  loss.append(metrics[0])\n",
    "  accuracy.append(metrics[1])\n",
    "\n",
    "\n",
    "epochs = range(len(accuracy))\n",
    "fig = plt.figure(figsize=(15, 5))\n",
    "ax1 = fig.add_subplot(121)\n",
    "plt.title('Training accuracy')\n",
    "plt.plot(epochs, accuracy, color='blue', label='Train')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()\n",
    "\n",
    "ax1 = fig.add_subplot(122)\n",
    "plt.title('Training loss')\n",
    "plt.plot(epochs, loss, color='blue', label='Train')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Loss')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "6KiDH8wCdj1Y"
   },
   "outputs": [],
   "source": [
    "# Train the last layer first\n",
    "EPOCH = 10\n",
    "BATCH_SIZE = 128\n",
    "history = model.fit(X_train_resize, y_train, validation_data=(X_val_resize, y_val), \n",
    "                    epochs=EPOCH, callbacks=[early_stopping_cb], batch_size=BATCH_SIZE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 386
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1992,
     "status": "ok",
     "timestamp": 1587588395073,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "nwC2nW0rNsxv",
    "outputId": "4d7bbcd7-5945-47f7-944b-60ce29fd20e0"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 1.6 s\n"
     ]
    }
   ],
   "source": [
    "fig_save_path = \"/content/drive/My Drive/Train_figs/InceptionV3_lastLayer_0.005\"\n",
    "plot_training(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "DXCguUTKKph0"
   },
   "source": [
    "### Unfreeze last layers and train the model again"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 308,
     "status": "ok",
     "timestamp": 1587588412580,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "M0q9xlY3w-17",
    "outputId": "61184c3d-6b4e-4071-f591-dc08b17f0eec"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "40:mixed0\n",
      "time: 2.9 ms\n"
     ]
    }
   ],
   "source": [
    "# Find the layer at which we will freeze\n",
    "layer_freeze = \"mixed0\"\n",
    "for index, layer in enumerate(base_model.layers):\n",
    "    if layer.name == layer_freeze:\n",
    "        print(str(index) + \":\" + str(layer.name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 378,
     "status": "ok",
     "timestamp": 1587588416311,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "MDjIsGhUwyX_",
    "outputId": "e544d35d-d7c5-4557-f611-3ca2e98d07a8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 28.8 ms\n"
     ]
    }
   ],
   "source": [
    "# Unfreeze some layers and train everything again\n",
    "model.trainable = True\n",
    "max_layer_freeze = 40\n",
    "for layer in base_model.layers[:max_layer_freeze]:\n",
    "  layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 67
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 407,
     "status": "ok",
     "timestamp": 1587588420224,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "gc1AQKhRj4QA",
    "outputId": "ff43fa4b-45c2-4902-dc1e-2ff7916ae215"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checkpoints directory is /content/drive/My Drive/Checkpoints/InceptionV3-DataAug-Freeze40\n",
      "Creating a checkpoints directory\n",
      "time: 9.67 ms\n"
     ]
    }
   ],
   "source": [
    "# Make checkpoint\n",
    "drive_root = '/content/drive/My Drive/'\n",
    "checkpoint_dir = os.path.join(drive_root, \"Checkpoints\", \"InceptionV3-Freeze40\")\n",
    "\n",
    "# Create checkpoint folder\n",
    "print(\"Checkpoints directory is\", checkpoint_dir)\n",
    "if os.path.exists(checkpoint_dir):\n",
    "  print(\"Checkpoints folder already exists\")\n",
    "else:\n",
    "  print(\"Creating a checkpoints directory\")\n",
    "  os.makedirs(checkpoint_dir)\n",
    "\n",
    "# Make checkpoint callbacks\n",
    "checkpoint_path = os.path.join(checkpoint_dir, \"best_model_1.hdf5\")\n",
    "checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
    "                    filepath=checkpoint_path, monitor='val_loss', verbose=0, save_best_only=True,\n",
    "                    save_weights_only=False, mode='auto')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "lX4Typ61VH7g"
   },
   "outputs": [],
   "source": [
    "# Recompile the model (to set lr after initial training on 1 batch)\n",
    "optimizer = keras.optimizers.Adam(learning_rate=0.002)\n",
    "model.compile(optimizer=optimizer,\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "# Train the whole model again\n",
    "EPOCH = 200\n",
    "BATCH_SIZE = 128\n",
    "history = model.fit(X_train_resize, y_train, validation_data=(X_val_resize, y_val), \n",
    "                    epochs=EPOCH, callbacks=[checkpoint_callback, early_stopping_cb], batch_size=BATCH_SIZE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 386
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1949,
     "status": "ok",
     "timestamp": 1587590620902,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "u6X2rRF1VOfi",
    "outputId": "39fb75b1-04c6-44e2-c741-8385f2a15c40"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 1.51 s\n"
     ]
    }
   ],
   "source": [
    "fig_save_path = \"/content/drive/My Drive/Train_figs/InceptionV3_method1\"\n",
    "plot_training(history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 5611,
     "status": "ok",
     "timestamp": 1587591039497,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "Hl1icRdlxu4v",
    "outputId": "8ad75d16-da5c-4bcc-877b-bab907cde7e6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Val accuracy: 95.8%\n",
      "time: 5.27 s\n"
     ]
    }
   ],
   "source": [
    "# Check the model on val data\n",
    "n_sample = 10000\n",
    "label_predict = model.predict(X_val_resize[:n_sample])\n",
    "label_predict = np.round(np.squeeze(label_predict))\n",
    "train_acc = sum(label_predict == y_val[:n_sample]) / len(y_val[:n_sample])\n",
    "print(\"Val accuracy: {:.1f}%\".format(train_acc*100))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Fa6HESjDorDC"
   },
   "source": [
    "## Method 2\n",
    "* Resize the data patches to match required input size of InceptionV3 (at least 75x75)\n",
    "* Freeze up to layer 40\n",
    "* Add data augmentation\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "JJrGJbWKorDE"
   },
   "source": [
    "### Load InceptionV3 and visualize the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 7942,
     "status": "ok",
     "timestamp": 1587591159001,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "wmaHxIrAorDE",
    "outputId": "78eefcf1-656c-466f-d070-b8679d18e4b8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 7.58 s\n"
     ]
    }
   ],
   "source": [
    "# Load the base model\n",
    "base_model = tf.keras.applications.InceptionV3(input_shape=[75, 75, 3], include_top=False, weights='imagenet')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "O3XDzHPJorDK"
   },
   "source": [
    "### Resize the data and construct the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 11775,
     "status": "ok",
     "timestamp": 1587591167558,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "88cumDBkorDK",
    "outputId": "78b03602-765b-4f2f-d642-86cf7701f3e2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 8.48 s\n"
     ]
    }
   ],
   "source": [
    "# Resize the data patches to match required input size of InceptionV3\n",
    "X_train_resize = tf.image.resize(\n",
    "                        X_train,\n",
    "                        [75, 75],\n",
    "                        method='bilinear',\n",
    "                        antialias=True)\n",
    "\n",
    "X_val_resize = tf.image.resize(\n",
    "                        X_val,\n",
    "                        [75, 75],\n",
    "                        method='bilinear',\n",
    "                        antialias=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 13830,
     "status": "ok",
     "timestamp": 1587591172020,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "zrk4ZMFiorDO",
    "outputId": "328bd161-b266-4878-e4e3-2aa1ac1e1a90"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 4.54 s\n"
     ]
    }
   ],
   "source": [
    "# Freeze all weights\n",
    "base_model.trainable = False \n",
    "\n",
    "# Add last layers\n",
    "global_average_layer = tf.keras.layers.GlobalAveragePooling2D()\n",
    "prediction_layer = tf.keras.layers.Dense(1, activation='sigmoid')\n",
    "\n",
    "# Build a new model reusing the pretrained base\n",
    "model = tf.keras.Sequential([\n",
    "  base_model,\n",
    "  global_average_layer,\n",
    "  prediction_layer\n",
    "])\n",
    "\n",
    "# Compile the model\n",
    "early_stopping_cb = keras.callbacks.EarlyStopping(patience=50, restore_best_weights=True, monitor='val_loss')\n",
    "optimizer = keras.optimizers.Adam(learning_rate=0.005)\n",
    "model.compile(optimizer=optimizer,\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "1D-WCMDnorDR"
   },
   "outputs": [],
   "source": [
    "# Check the model\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "BYe83-J2orDU"
   },
   "source": [
    "### Train the last layer first"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ttOQEatnorDX"
   },
   "outputs": [],
   "source": [
    "# Train the last layer first\n",
    "EPOCH = 20\n",
    "BATCH_SIZE = 128\n",
    "history = model.fit(X_train_resize, y_train, validation_data=(X_val_resize, y_val), \n",
    "                    epochs=EPOCH, callbacks=[early_stopping_cb], batch_size=BATCH_SIZE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "dVVgJVPPorDc"
   },
   "source": [
    "### Unfreeze last layers and train the model again"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 384,
     "status": "ok",
     "timestamp": 1587591472651,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "dTP7noicorDc",
    "outputId": "656de48d-5176-4b3e-c60a-5c0c8a216a48"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "40:mixed0\n",
      "time: 3.13 ms\n"
     ]
    }
   ],
   "source": [
    "# Find the layer at which we will freeze\n",
    "layer_freeze = \"mixed0\"\n",
    "for index, layer in enumerate(base_model.layers):\n",
    "    if layer.name == layer_freeze:\n",
    "        print(str(index) + \":\" + str(layer.name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 543,
     "status": "ok",
     "timestamp": 1587591472825,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "aZPWqgUZorDe",
    "outputId": "4cde9269-2687-49ab-d393-24a68f9a033a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 22.5 ms\n"
     ]
    }
   ],
   "source": [
    "# Unfreeze some layers\n",
    "model.trainable = True\n",
    "max_layer_freeze = 40\n",
    "for layer in base_model.layers[:max_layer_freeze]:\n",
    "  layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 67
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 540,
     "status": "ok",
     "timestamp": 1587591472827,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "iEpcZgYCorDg",
    "outputId": "22213ba2-ac07-4353-b95f-2cbca3569cfd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checkpoints directory is /content/drive/My Drive/Checkpoints/InceptionV3-DataAug-Freeze40\n",
      "Checkpoints folder already exists\n",
      "time: 7.64 ms\n"
     ]
    }
   ],
   "source": [
    "# Make checkpoint\n",
    "drive_root = '/content/drive/My Drive/'\n",
    "checkpoint_dir = os.path.join(drive_root, \"Checkpoints\", \"InceptionV3-DataAug-Freeze40\")\n",
    "\n",
    "# Create checkpoint folder\n",
    "print(\"Checkpoints directory is\", checkpoint_dir)\n",
    "if os.path.exists(checkpoint_dir):\n",
    "  print(\"Checkpoints folder already exists\")\n",
    "else:\n",
    "  print(\"Creating a checkpoints directory\")\n",
    "  os.makedirs(checkpoint_dir)\n",
    "\n",
    "# Make checkpoint callbacks\n",
    "checkpoint_path = os.path.join(checkpoint_dir, \"best_model.hdf5\")\n",
    "checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
    "                    filepath=checkpoint_path, monitor='val_loss', verbose=0, save_best_only=True,\n",
    "                    save_weights_only=False, mode='auto')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1783,
     "status": "ok",
     "timestamp": 1587591474080,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "9LBtLxj2qVoS",
    "outputId": "31e86623-a8d9-4a46-b770-ed9978c7f2cd"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 1.26 s\n"
     ]
    }
   ],
   "source": [
    "# Data augmentation\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "# Create the augmentation generator\n",
    "train_datagen = ImageDataGenerator(\n",
    "    shear_range=0.2,    \n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2,  \n",
    "    horizontal_flip=True,\n",
    "    vertical_flip=True)\n",
    "\n",
    "train_datagen.fit(X_train_resize)\n",
    "\n",
    "# Note that the validation data should not be augmented!\n",
    "train_generator = train_datagen.flow(\n",
    "        X_train_resize, y_train,\n",
    "        batch_size=BATCH_SIZE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0qbA0vAeorDi"
   },
   "outputs": [],
   "source": [
    "# Recompile the model (to set lr after initial training on 1 batch)\n",
    "optimizer = keras.optimizers.Adam(learning_rate=0.002)\n",
    "model.compile(optimizer=optimizer,\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "# Train the whole model again\n",
    "EPOCH = 200\n",
    "BATCH_SIZE = 128\n",
    "history = model.fit(\n",
    "      train_generator,\n",
    "      steps_per_epoch=X_train_resize.shape[0] / BATCH_SIZE,\n",
    "      epochs=EPOCH,\n",
    "      validation_data=(X_val_resize, y_val), \n",
    "      callbacks=[checkpoint_callback, early_stopping_cb])   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 386
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1810,
     "status": "ok",
     "timestamp": 1587606305501,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "oflX2fzQorDk",
    "outputId": "0911d2e9-1a21-4d06-cced-db82c16ddca6"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 1.71 s\n"
     ]
    }
   ],
   "source": [
    "fig_save_path = \"/content/drive/My Drive/Train_figs/InceptionV3_method2\"\n",
    "plot_training(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "buA6BibsEgt1"
   },
   "source": [
    "## Method 3\n",
    "* Use tranpose convolution to upsample (3X) input to match required input size of InceptionV3 (at least 75x75)\n",
    "* Freeze up to layer 40\n",
    "* Add data augmentation\n",
    "* Use unbalanced data (10 times more patches with no tumor)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "GofT9weXEgt3"
   },
   "source": [
    "### Load InceptionV3 and visualize the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 67
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 44292,
     "status": "ok",
     "timestamp": 1587770377049,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "3TWKSDogEgt3",
    "outputId": "a0a41796-6346-45ad-e1cb-757093809159"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/inception_v3/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
      "87916544/87910968 [==============================] - 1s 0us/step\n",
      "time: 13.2 s\n"
     ]
    }
   ],
   "source": [
    "# Load the base model\n",
    "base_model = tf.keras.applications.InceptionV3(input_shape=[32*3, 32*3, 3], include_top=False, weights='imagenet')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "4Xn9KLBeEgt7"
   },
   "outputs": [],
   "source": [
    "# Plot the base model\n",
    "from tensorflow.keras.utils import plot_model\n",
    "plot_model(base_model, to_file='model.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "yGzL1L_WEgt9"
   },
   "outputs": [],
   "source": [
    "# List all the layers\n",
    "base_model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "onAvdVlYEgt_"
   },
   "source": [
    "### Construct the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 4029,
     "status": "ok",
     "timestamp": 1587770381083,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "wIiDuw69EguC",
    "outputId": "7c2bbded-c7a5-4f7e-fb49-0e3e98506328"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 2.83 s\n"
     ]
    }
   ],
   "source": [
    "# Freeze all weights\n",
    "base_model.trainable = False \n",
    "\n",
    "# Add input transpose conv layer\n",
    "input_layer = tf.keras.layers.InputLayer(\n",
    "    input_shape=[32, 32, 3])\n",
    "input_tranConv = tf.keras.layers.Conv2DTranspose(3, [3, 3], strides=3,\n",
    "                                    padding='same',\n",
    "                                    use_bias=False)\n",
    "\n",
    "# Add last layers\n",
    "global_average_layer = tf.keras.layers.GlobalAveragePooling2D()\n",
    "prediction_layer = tf.keras.layers.Dense(1, activation='sigmoid')\n",
    "\n",
    "# Build a new model reusing the pretrained base\n",
    "model = tf.keras.Sequential([\n",
    "  input_layer,\n",
    "  input_tranConv,                           \n",
    "  base_model,\n",
    "  global_average_layer,\n",
    "  prediction_layer\n",
    "])\n",
    "\n",
    "# Compile the model\n",
    "early_stopping_cb = keras.callbacks.EarlyStopping(patience=50, restore_best_weights=True, monitor='val_loss')\n",
    "optimizer = keras.optimizers.Adam(learning_rate=0.005)\n",
    "model.compile(optimizer=optimizer,\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "XL7wOUvFEguE"
   },
   "outputs": [],
   "source": [
    "# Check the model\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "AT_fs45wEguF"
   },
   "source": [
    "### Train the last layer first"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 370
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1844764,
     "status": "ok",
     "timestamp": 1587772221834,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "_ZDKftA_EguJ",
    "outputId": "0f62fa7b-9d98-46ba-b1f6-affa7d3909ca"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "679/679 [==============================] - 186s 274ms/step - loss: 0.1569 - accuracy: 0.9492 - val_loss: 0.1321 - val_accuracy: 0.9580\n",
      "Epoch 2/10\n",
      "679/679 [==============================] - 182s 269ms/step - loss: 0.1336 - accuracy: 0.9570 - val_loss: 0.1231 - val_accuracy: 0.9613\n",
      "Epoch 3/10\n",
      "679/679 [==============================] - 182s 268ms/step - loss: 0.1291 - accuracy: 0.9595 - val_loss: 0.1245 - val_accuracy: 0.9608\n",
      "Epoch 4/10\n",
      "679/679 [==============================] - 182s 269ms/step - loss: 0.1268 - accuracy: 0.9602 - val_loss: 0.1202 - val_accuracy: 0.9615\n",
      "Epoch 5/10\n",
      "679/679 [==============================] - 182s 268ms/step - loss: 0.1242 - accuracy: 0.9610 - val_loss: 0.1206 - val_accuracy: 0.9619\n",
      "Epoch 6/10\n",
      "679/679 [==============================] - 182s 268ms/step - loss: 0.1239 - accuracy: 0.9612 - val_loss: 0.1183 - val_accuracy: 0.9636\n",
      "Epoch 7/10\n",
      "679/679 [==============================] - 182s 268ms/step - loss: 0.1232 - accuracy: 0.9614 - val_loss: 0.1230 - val_accuracy: 0.9602\n",
      "Epoch 8/10\n",
      "679/679 [==============================] - 182s 268ms/step - loss: 0.1241 - accuracy: 0.9613 - val_loss: 0.1148 - val_accuracy: 0.9645\n",
      "Epoch 9/10\n",
      "679/679 [==============================] - 182s 268ms/step - loss: 0.1233 - accuracy: 0.9618 - val_loss: 0.1257 - val_accuracy: 0.9586\n",
      "Epoch 10/10\n",
      "679/679 [==============================] - 182s 268ms/step - loss: 0.1211 - accuracy: 0.9622 - val_loss: 0.1155 - val_accuracy: 0.9641\n",
      "time: 30min 40s\n"
     ]
    }
   ],
   "source": [
    "# Train the first and last layer first\n",
    "EPOCH = 10\n",
    "BATCH_SIZE = 512\n",
    "y_train = np.array(y_train)\n",
    "y_val = np.array(y_val)\n",
    "history = model.fit(X_train, y_train, validation_data=(X_val, y_val), \n",
    "                    epochs=EPOCH, callbacks=[early_stopping_cb], batch_size=BATCH_SIZE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "wkeUUnvOEguM"
   },
   "source": [
    "### Unfreeze last layers and train the model again"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 308,
     "status": "ok",
     "timestamp": 1587588412580,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "39f4RY8xEguM",
    "outputId": "61184c3d-6b4e-4071-f591-dc08b17f0eec"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "40:mixed0\n",
      "time: 2.9 ms\n"
     ]
    }
   ],
   "source": [
    "# Find the layer at which we will freeze\n",
    "layer_freeze = \"mixed0\"\n",
    "for index, layer in enumerate(base_model.layers):\n",
    "    if layer.name == layer_freeze:\n",
    "        print(str(index) + \":\" + str(layer.name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1844756,
     "status": "ok",
     "timestamp": 1587772221837,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "3AlvHcnHEguO",
    "outputId": "bb6b1a81-2fc4-46d5-9917-8f914284d739"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 9.69 ms\n"
     ]
    }
   ],
   "source": [
    "# Unfreeze some layers and train everything again\n",
    "model.trainable = True\n",
    "max_layer_freeze = 40\n",
    "for layer in base_model.layers[:max_layer_freeze]:\n",
    "  layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 67
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1837209,
     "status": "ok",
     "timestamp": 1587772221838,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "ntGYnCCgEguQ",
    "outputId": "ac203d02-cd39-4f1b-e33c-63e635812b32"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checkpoints directory is /content/drive/My Drive/Checkpoints/InceptionV3-Freeze40-UnbalData\n",
      "Creating a checkpoints directory\n",
      "time: 8.98 ms\n"
     ]
    }
   ],
   "source": [
    "# Make checkpoint\n",
    "drive_root = '/content/drive/My Drive/'\n",
    "checkpoint_dir = os.path.join(drive_root, \"Checkpoints\", \"InceptionV3-Freeze40-UnbalData\")\n",
    "\n",
    "# Create checkpoint folder\n",
    "print(\"Checkpoints directory is\", checkpoint_dir)\n",
    "if os.path.exists(checkpoint_dir):\n",
    "  print(\"Checkpoints folder already exists\")\n",
    "else:\n",
    "  print(\"Creating a checkpoints directory\")\n",
    "  os.makedirs(checkpoint_dir)\n",
    "\n",
    "# Make checkpoint callbacks\n",
    "checkpoint_path = os.path.join(checkpoint_dir, \"best_model.hdf5\")\n",
    "checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
    "                    filepath=checkpoint_path, monitor='val_loss', verbose=0, save_best_only=True,\n",
    "                    save_weights_only=False, mode='auto')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2150,
     "status": "ok",
     "timestamp": 1587772876368,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "zFAUhclWg9ft",
    "outputId": "5f731808-4f8d-461f-9ab2-cb89c9998494"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 1.55 s\n"
     ]
    }
   ],
   "source": [
    "# Data augmentation\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "# Create the augmentation generator\n",
    "train_datagen = ImageDataGenerator(\n",
    "    shear_range=0.2,    \n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2,  \n",
    "    horizontal_flip=True,\n",
    "    vertical_flip=True)\n",
    "\n",
    "train_datagen.fit(X_train)\n",
    "\n",
    "# Note that the validation data should not be augmented!\n",
    "train_generator = train_datagen.flow(\n",
    "        X_train, y_train,\n",
    "        batch_size=BATCH_SIZE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "eGEi8A-BIx1I"
   },
   "outputs": [],
   "source": [
    "# Recompile the model (to set lr after initial training on 1 batch)\n",
    "optimizer = keras.optimizers.Adam(learning_rate=0.001)\n",
    "model.compile(optimizer=optimizer,\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "# Train the whole model again\n",
    "EPOCH = 200\n",
    "history = model.fit(\n",
    "      train_generator,\n",
    "      steps_per_epoch=X_train.shape[0] / BATCH_SIZE,\n",
    "      epochs=EPOCH,\n",
    "      validation_data=(X_val, y_val), \n",
    "      callbacks=[checkpoint_callback, early_stopping_cb])   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 386
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2921,
     "status": "ok",
     "timestamp": 1587802871845,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "xBsWzP6bn_eh",
    "outputId": "0d4d429f-bbd0-4587-b349-ea0d9cdf6676"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 1.32 s\n"
     ]
    }
   ],
   "source": [
    "# Plot the history of training\n",
    "fig_save_path = \"/content/drive/My Drive/Train_figs/InceptionV3_method3\"\n",
    "plot_training(history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 5611,
     "status": "ok",
     "timestamp": 1587591039497,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "tzFMDLChEguW",
    "outputId": "8ad75d16-da5c-4bcc-877b-bab907cde7e6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Val accuracy: 95.8%\n",
      "time: 5.27 s\n"
     ]
    }
   ],
   "source": [
    "# Check the model on val data\n",
    "n_sample = 10000\n",
    "label_predict = model.predict(X_val_resize[:n_sample])\n",
    "label_predict = np.round(np.squeeze(label_predict))\n",
    "train_acc = sum(label_predict == y_val[:n_sample]) / len(y_val[:n_sample])\n",
    "print(\"Val accuracy: {:.1f}%\".format(train_acc*100))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "QarRNeu7r7Ep"
   },
   "source": [
    "# Inception-ResNet-V2\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "r9cdhvErGkPr"
   },
   "source": [
    "## Method 1\n",
    "* Resize the data patches to match required input size of the model (at least 75x75)\n",
    "* Freeze up to layer 40\n",
    "* Add data augmentation\n",
    "* Batch size 512"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "y88acskXr7Er"
   },
   "source": [
    "### Load InceptionV3 and visualize the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 11678,
     "status": "ok",
     "timestamp": 1587647590638,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "MilcGG7Lr7Er",
    "outputId": "1b282ded-31eb-4b34-d6ba-9bc9ccd08f7f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 10.3 s\n"
     ]
    }
   ],
   "source": [
    "# Load the base model\n",
    "base_model = tf.keras.applications.InceptionResNetV2(input_shape=[75, 75, 3], include_top=False, weights='imagenet')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "hmAQoEqmFKtF"
   },
   "outputs": [],
   "source": [
    "base_model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "PGPT1KS1r7Ev"
   },
   "source": [
    "### Resize the data and construct the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 7968,
     "status": "ok",
     "timestamp": 1587646590268,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "FfmRdzjar7Ew",
    "outputId": "1b95867d-0654-41c2-cf5b-02036e0fa645"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 7.12 s\n"
     ]
    }
   ],
   "source": [
    "# Resize the data patches to match required input size of InceptionV3\n",
    "X_train_resize = tf.image.resize(\n",
    "                        X_train,\n",
    "                        [75, 75],\n",
    "                        method='bilinear',\n",
    "                        antialias=True)\n",
    "\n",
    "X_val_resize = tf.image.resize(\n",
    "                        X_val,\n",
    "                        [75, 75],\n",
    "                        method='bilinear',\n",
    "                        antialias=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 8161,
     "status": "ok",
     "timestamp": 1587647598804,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "jnCIPMePr7Ez",
    "outputId": "57d7a6f0-b639-4909-e5df-5e0107b2317a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 6.98 s\n"
     ]
    }
   ],
   "source": [
    "# Freeze all weights\n",
    "base_model.trainable = False \n",
    "\n",
    "# Add last layers\n",
    "global_average_layer = tf.keras.layers.GlobalAveragePooling2D()\n",
    "prediction_layer = tf.keras.layers.Dense(1, activation='sigmoid')\n",
    "\n",
    "# Build a new model reusing the pretrained base\n",
    "model = tf.keras.Sequential([\n",
    "  base_model,\n",
    "  global_average_layer,\n",
    "  prediction_layer\n",
    "])\n",
    "\n",
    "# Compile the model\n",
    "early_stopping_cb = keras.callbacks.EarlyStopping(patience=50, restore_best_weights=True, monitor='val_loss')\n",
    "optimizer = keras.optimizers.Adam(learning_rate=0.005)\n",
    "model.compile(optimizer=optimizer,\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "xY0RPkcVr7E1"
   },
   "outputs": [],
   "source": [
    "# Check the model\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mdElVsv0r7E3"
   },
   "source": [
    "### Train the last layer first"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "bJ_PXNkbr7E3"
   },
   "outputs": [],
   "source": [
    "# Train the last layer first\n",
    "EPOCH = 20\n",
    "BATCH_SIZE = 512\n",
    "history = model.fit(X_train_resize, y_train, validation_data=(X_val_resize, y_val), \n",
    "                    epochs=EPOCH, callbacks=[early_stopping_cb], batch_size=BATCH_SIZE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "DbJLLr7Rr7E5"
   },
   "source": [
    "### Unfreeze last layers and train the model again"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1213,
     "status": "ok",
     "timestamp": 1587647055112,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "GlrIvEA1r7E5",
    "outputId": "a0b692cd-9a4b-477c-9189-4f03ba82b566"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "40:mixed_5b\n",
      "time: 4.89 ms\n"
     ]
    }
   ],
   "source": [
    "# Find the layer at which we will freeze\n",
    "layer_freeze = \"mixed_5b\"\n",
    "for index, layer in enumerate(base_model.layers):\n",
    "    if layer.name == layer_freeze:\n",
    "        print(str(index) + \":\" + str(layer.name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1463,
     "status": "ok",
     "timestamp": 1587647938212,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "Lf_BOipOr7E7",
    "outputId": "1cb84d39-fee8-4448-ade1-7fbf74be89b1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 21.8 ms\n"
     ]
    }
   ],
   "source": [
    "# Unfreeze some layers\n",
    "model.trainable = True\n",
    "max_layer_freeze = 40\n",
    "for layer in base_model.layers[:max_layer_freeze]:\n",
    "  layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 67
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1454,
     "status": "ok",
     "timestamp": 1587647938213,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "Ps5VOMQjr7E9",
    "outputId": "9a034b7b-9444-4216-a8b5-c85e5bd227e2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checkpoints directory is /content/drive/My Drive/Checkpoints/InceptionResnetV3-DataAug-Freeze40\n",
      "Checkpoints folder already exists\n",
      "time: 5.13 ms\n"
     ]
    }
   ],
   "source": [
    "# Make checkpoint\n",
    "drive_root = '/content/drive/My Drive/'\n",
    "checkpoint_dir = os.path.join(drive_root, \"Checkpoints\", \"InceptionResnetV3-DataAug-Freeze40\")\n",
    "\n",
    "# Create checkpoint folder\n",
    "print(\"Checkpoints directory is\", checkpoint_dir)\n",
    "if os.path.exists(checkpoint_dir):\n",
    "  print(\"Checkpoints folder already exists\")\n",
    "else:\n",
    "  print(\"Creating a checkpoints directory\")\n",
    "  os.makedirs(checkpoint_dir)\n",
    "\n",
    "# Make checkpoint callbacks\n",
    "checkpoint_path = os.path.join(checkpoint_dir, \"best_model.hdf5\")\n",
    "checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
    "                    filepath=checkpoint_path, monitor='val_loss', verbose=0, save_best_only=True,\n",
    "                    save_weights_only=False, mode='auto')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1959,
     "status": "ok",
     "timestamp": 1587647938727,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "4fsWns8Cr7FA",
    "outputId": "26f20a82-9b37-45ff-96aa-7a9950608ca0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 1.28 s\n"
     ]
    }
   ],
   "source": [
    "# Data augmentation\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "# Create the augmentation generator\n",
    "train_datagen = ImageDataGenerator(\n",
    "    shear_range=0.2,    \n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2,  \n",
    "    horizontal_flip=True,\n",
    "    vertical_flip=True)\n",
    "\n",
    "train_datagen.fit(X_train_resize)\n",
    "\n",
    "# Note that the validation data should not be augmented!\n",
    "BATCH_SIZE = 512\n",
    "train_generator = train_datagen.flow(\n",
    "        X_train_resize, y_train,\n",
    "        batch_size=BATCH_SIZE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "f2YUueyzr7FC"
   },
   "outputs": [],
   "source": [
    "# Recompile the model (to set lr after initial training on 1 batch)\n",
    "optimizer = keras.optimizers.Adam(learning_rate=0.002)\n",
    "model.compile(optimizer=optimizer,\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "# Train the whole model again\n",
    "EPOCH = 200\n",
    "history = model.fit(\n",
    "      train_generator,\n",
    "      steps_per_epoch=X_train_resize.shape[0] / BATCH_SIZE,\n",
    "      epochs=EPOCH,\n",
    "      validation_data=(X_val_resize, y_val), \n",
    "      callbacks=[checkpoint_callback, early_stopping_cb])   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 386
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2483,
     "status": "ok",
     "timestamp": 1587656057060,
     "user": {
      "displayName": "Long Luu",
      "photoUrl": "",
      "userId": "03414912596698907112"
     },
     "user_tz": 240
    },
    "id": "wpfOUIElr7FE",
    "outputId": "1acbe1e2-bdab-4733-f843-861da75fc6e1"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time: 1.26 s\n"
     ]
    }
   ],
   "source": [
    "fig_save_path = \"/content/drive/My Drive/Train_figs/InceptionResnetV3_method1\"\n",
    "plot_training(history)"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [
    "S-2YALtORhPR",
    "Oqfb507nMU7j",
    "eyn4hWezJwqw",
    "iu7fyoXZNhWv",
    "CiwsP_dSKTIu",
    "Fa6HESjDorDC",
    "JJrGJbWKorDE",
    "O3XDzHPJorDK",
    "BYe83-J2orDU",
    "dVVgJVPPorDc",
    "QarRNeu7r7Ep",
    "r9cdhvErGkPr",
    "y88acskXr7Er",
    "PGPT1KS1r7Ev"
   ],
   "machine_shape": "hm",
   "name": "CancerDetection_model_training_1level_transfer.ipynb",
   "provenance": []
  },
  "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}