task2_template.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "1d234e3f-af8b-4757-83f3-241b04b7a511"
    }
   },
   "source": [
    "# Class Challenge: Image Classification of COVID-19 X-rays\n",
    "# Task 2 [Total points: 30]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "b07b9992-b592-4871-9f9a-3b0ae5fa8b5f"
    }
   },
   "source": [
    "## Setup\n",
    "\n",
    "* This assignment involves the following packages: 'matplotlib', 'numpy', and 'sklearn'. \n",
    "\n",
    "* If you are using conda, use the following commands to install the above packages:<br>\n",
    "```shell\n",
    "conda install matplotlib\n",
    "conda install numpy\n",
    "conda install -c anaconda scikit-learn\n",
    "```\n",
    "\n",
    "* If you are using pip, use use the following commands to install the above packages: <br> \n",
    "```shell\n",
    "pip install matplotlib\n",
    "pip install numpy\n",
    "pip install sklearn\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "5516f610-f00c-43c3-bf19-4cf7ab9c089f"
    }
   },
   "source": [
    "## Data\n",
    "\n",
    "Please download the data using the following link: [COVID-19](https://drive.google.com/file/d/1Y88tgqpQ1Pjko_7rntcPowOJs_QNOrJ-/view). \n",
    "\n",
    "* After downloading 'Covid_Data_GradientCrescent.zip', unzip the file and you should see the following data structure:\n",
    "\n",
    "\n",
    "|--all<br>\n",
    "|--------train<br>\n",
    "|--------test<br>\n",
    "|--two<br>\n",
    "|--------train<br>\n",
    "|--------test<br>\n",
    "\n",
    "\n",
    "* Put the 'all' folder, the 'two' folder and this python notebook in the **same directory** so that the following code can correctly locate the data.  \n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "## [20 points] Multi-class Classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "nbpresent": {
     "id": "5e05f980-3d14-4367-b3d1-664249145b13"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'2.1.0'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "os.environ['OMP_NUM_THREADS'] = '1'\n",
    "os.environ['CUDA_VISIBLE_DEVICES'] = '-1'\n",
    "tf.__version__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "#### Load Image Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "nbpresent": {
     "id": "58317664-8da3-4283-b3e5-35721687e7ab"
    }
   },
   "outputs": [],
   "source": [
    "DATA_LIST = os.listdir('all/train')\n",
    "DATASET_PATH  = 'all/train'\n",
    "TEST_DIR =  'all/test'\n",
    "IMAGE_SIZE    = (224, 224)\n",
    "NUM_CLASSES   = len(DATA_LIST)\n",
    "BATCH_SIZE    = 10  # try reducing batch size or freeze more layers if your GPU runs out of memory\n",
    "NUM_EPOCHS    = 100\n",
    "LEARNING_RATE = 0.0001 # start off with high rate first 0.001 and experiment with reducing it gradually "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "#### Generate Training and Validation Batches"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "nbpresent": {
     "id": "71760e29-2b08-4259-93f1-ebe68cd74a6d"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "H:\\Anaconda3\\envs\\tf\\lib\\site-packages\\keras_preprocessing\\image\\image_data_generator.py:341: UserWarning: This ImageDataGenerator specifies `zca_whitening` which overrides setting of`featurewise_std_normalization`.\n",
      "  warnings.warn('This ImageDataGenerator specifies '\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 216 images belonging to 4 classes.\n",
      "Found 54 images belonging to 4 classes.\n"
     ]
    }
   ],
   "source": [
    "train_datagen = ImageDataGenerator(rescale=1./255,rotation_range=50,featurewise_center = True,\n",
    "                                   featurewise_std_normalization = True,width_shift_range=0.2,\n",
    "                                   height_shift_range=0.2,shear_range=0.25,zoom_range=0.1,\n",
    "                                   zca_whitening = True,channel_shift_range = 20,\n",
    "                                   horizontal_flip = True,vertical_flip = True,\n",
    "                                   validation_split = 0.2,fill_mode='constant')\n",
    "\n",
    "\n",
    "train_batches = train_datagen.flow_from_directory(DATASET_PATH,target_size=IMAGE_SIZE,\n",
    "                                                  shuffle=True,batch_size=BATCH_SIZE,\n",
    "                                                  subset = \"training\",seed=42,\n",
    "                                                  class_mode=\"categorical\")\n",
    "\n",
    "valid_batches = train_datagen.flow_from_directory(DATASET_PATH,target_size=IMAGE_SIZE,\n",
    "                                                  shuffle=True,batch_size=BATCH_SIZE,\n",
    "                                                  subset = \"validation\",\n",
    "                                                  seed=42,class_mode=\"categorical\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "#### [10 points] Build Model\n",
    "Hint: Starting from a pre-trained model typically helps performance on a new task, e.g. starting with weights obtained by training on ImageNet. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "nbpresent": {
     "id": "5c52eeb2-1092-4d45-9e16-02219c82cb5e"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "vgg16 (Model)                (None, 7, 7, 512)         14714688  \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 25088)             0         \n",
      "_________________________________________________________________\n",
      "feature_dense (Dense)        (None, 256)               6422784   \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 4)                 1028      \n",
      "=================================================================\n",
      "Total params: 21,138,500\n",
      "Trainable params: 6,423,812\n",
      "Non-trainable params: 14,714,688\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "raise NotImplementedError(\"Build your model based on an architecture of your choice \"\n",
    "                          \"A sample model summary is shown below\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "#### [5 points] Train Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "nbpresent": {
     "id": "cdb90d24-6a23-4969-8138-f37e7050062d"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "22\n",
      "6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "H:\\Anaconda3\\envs\\tf\\lib\\site-packages\\keras_preprocessing\\image\\image_data_generator.py:716: UserWarning: This ImageDataGenerator specifies `featurewise_center`, but it hasn't been fit on any training data. Fit it first by calling `.fit(numpy_data)`.\n",
      "  warnings.warn('This ImageDataGenerator specifies '\n",
      "H:\\Anaconda3\\envs\\tf\\lib\\site-packages\\keras_preprocessing\\image\\image_data_generator.py:735: UserWarning: This ImageDataGenerator specifies `zca_whitening`, but it hasn't been fit on any training data. Fit it first by calling `.fit(numpy_data)`.\n",
      "  warnings.warn('This ImageDataGenerator specifies '\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:sample_weight modes were coerced from\n",
      "  ...\n",
      "    to  \n",
      "  ['...']\n",
      "WARNING:tensorflow:sample_weight modes were coerced from\n",
      "  ...\n",
      "    to  \n",
      "  ['...']\n",
      "Train for 21 steps, validate for 5 steps\n",
      "Epoch 1/100\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "H:\\Anaconda3\\envs\\tf\\lib\\site-packages\\keras_preprocessing\\image\\image_data_generator.py:716: UserWarning: This ImageDataGenerator specifies `featurewise_center`, but it hasn't been fit on any training data. Fit it first by calling `.fit(numpy_data)`.\n",
      "  warnings.warn('This ImageDataGenerator specifies '\n",
      "H:\\Anaconda3\\envs\\tf\\lib\\site-packages\\keras_preprocessing\\image\\image_data_generator.py:735: UserWarning: This ImageDataGenerator specifies `zca_whitening`, but it hasn't been fit on any training data. Fit it first by calling `.fit(numpy_data)`.\n",
      "  warnings.warn('This ImageDataGenerator specifies '\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "21/21 [==============================] - 66s 3s/step - loss: 1.6074 - acc: 0.2427 - val_loss: 1.3532 - val_acc: 0.3600\n",
      "Epoch 2/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 1.2078 - acc: 0.4272 - val_loss: 1.1174 - val_acc: 0.4800\n",
      "Epoch 3/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 1.0575 - acc: 0.5048 - val_loss: 0.9547 - val_acc: 0.6600\n",
      "Epoch 4/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 1.0413 - acc: 0.5680 - val_loss: 0.9135 - val_acc: 0.5000\n",
      "Epoch 5/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.9270 - acc: 0.5971 - val_loss: 0.9027 - val_acc: 0.6800\n",
      "Epoch 6/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.9233 - acc: 0.5777 - val_loss: 0.8939 - val_acc: 0.5800\n",
      "Epoch 7/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.9431 - acc: 0.5728 - val_loss: 0.8314 - val_acc: 0.6000\n",
      "Epoch 8/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.9077 - acc: 0.6262 - val_loss: 0.8964 - val_acc: 0.6000\n",
      "Epoch 9/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.8537 - acc: 0.6456 - val_loss: 0.8214 - val_acc: 0.6800\n",
      "Epoch 10/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.8151 - acc: 0.6699 - val_loss: 0.7294 - val_acc: 0.6600\n",
      "Epoch 11/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.7999 - acc: 0.6650 - val_loss: 0.8830 - val_acc: 0.6600\n",
      "Epoch 12/100\n",
      "21/21 [==============================] - 68s 3s/step - loss: 0.7547 - acc: 0.6650 - val_loss: 0.7593 - val_acc: 0.6600\n",
      "Epoch 13/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.7219 - acc: 0.6505 - val_loss: 0.7540 - val_acc: 0.6200\n",
      "Epoch 14/100\n",
      "21/21 [==============================] - 72s 3s/step - loss: 0.7014 - acc: 0.6845 - val_loss: 0.8700 - val_acc: 0.5600\n",
      "Epoch 15/100\n",
      "21/21 [==============================] - 70s 3s/step - loss: 0.7170 - acc: 0.7039 - val_loss: 0.7459 - val_acc: 0.6800\n",
      "Epoch 16/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.7106 - acc: 0.6990 - val_loss: 0.7080 - val_acc: 0.6800\n",
      "Epoch 17/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.6886 - acc: 0.6942 - val_loss: 0.7756 - val_acc: 0.6800\n",
      "Epoch 18/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.7079 - acc: 0.6990 - val_loss: 0.6881 - val_acc: 0.6200\n",
      "Epoch 19/100\n",
      "21/21 [==============================] - 68s 3s/step - loss: 0.7108 - acc: 0.6845 - val_loss: 0.8930 - val_acc: 0.6000\n",
      "Epoch 20/100\n",
      "21/21 [==============================] - 72s 3s/step - loss: 0.6758 - acc: 0.7330 - val_loss: 0.8015 - val_acc: 0.6000\n",
      "Epoch 21/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.7053 - acc: 0.6748 - val_loss: 0.7972 - val_acc: 0.7200\n",
      "Epoch 22/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.7212 - acc: 0.6845 - val_loss: 0.6422 - val_acc: 0.6800\n",
      "Epoch 23/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.7102 - acc: 0.7184 - val_loss: 0.7059 - val_acc: 0.6600\n",
      "Epoch 24/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.6430 - acc: 0.7136 - val_loss: 0.6389 - val_acc: 0.7200\n",
      "Epoch 25/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.6289 - acc: 0.7233 - val_loss: 0.7491 - val_acc: 0.6000\n",
      "Epoch 26/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.6824 - acc: 0.7136 - val_loss: 0.7847 - val_acc: 0.6200\n",
      "Epoch 27/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.6452 - acc: 0.7136 - val_loss: 0.7972 - val_acc: 0.5800\n",
      "Epoch 28/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.6968 - acc: 0.7330 - val_loss: 0.8064 - val_acc: 0.6800\n",
      "Epoch 29/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.6546 - acc: 0.7282 - val_loss: 0.8066 - val_acc: 0.6400\n",
      "Epoch 30/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.6388 - acc: 0.7330 - val_loss: 0.6779 - val_acc: 0.6000\n",
      "Epoch 31/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.6296 - acc: 0.7330 - val_loss: 0.6667 - val_acc: 0.7000\n",
      "Epoch 32/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.6013 - acc: 0.7427 - val_loss: 0.5956 - val_acc: 0.6600\n",
      "Epoch 33/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.6152 - acc: 0.7476 - val_loss: 0.7639 - val_acc: 0.6200\n",
      "Epoch 34/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.6560 - acc: 0.7136 - val_loss: 0.9225 - val_acc: 0.5800\n",
      "Epoch 35/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.5997 - acc: 0.7767 - val_loss: 0.8162 - val_acc: 0.6800\n",
      "Epoch 36/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.5723 - acc: 0.7524 - val_loss: 0.7213 - val_acc: 0.6800\n",
      "Epoch 37/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.6523 - acc: 0.7039 - val_loss: 0.7317 - val_acc: 0.7400\n",
      "Epoch 38/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.6422 - acc: 0.7282 - val_loss: 0.6942 - val_acc: 0.6800\n",
      "Epoch 39/100\n",
      "21/21 [==============================] - 68s 3s/step - loss: 0.6045 - acc: 0.7621 - val_loss: 0.7990 - val_acc: 0.6000\n",
      "Epoch 40/100\n",
      "21/21 [==============================] - 69s 3s/step - loss: 0.6027 - acc: 0.7330 - val_loss: 0.7935 - val_acc: 0.6400\n",
      "Epoch 41/100\n",
      "21/21 [==============================] - 74s 4s/step - loss: 0.5507 - acc: 0.7913 - val_loss: 0.7630 - val_acc: 0.6600\n",
      "Epoch 42/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.6141 - acc: 0.7233 - val_loss: 0.7112 - val_acc: 0.7000\n",
      "Epoch 43/100\n",
      "21/21 [==============================] - 77s 4s/step - loss: 0.5431 - acc: 0.7718 - val_loss: 0.7325 - val_acc: 0.7000\n",
      "Epoch 44/100\n",
      "21/21 [==============================] - 69s 3s/step - loss: 0.5976 - acc: 0.7476 - val_loss: 0.7057 - val_acc: 0.6400\n",
      "Epoch 45/100\n",
      "21/21 [==============================] - 77s 4s/step - loss: 0.5342 - acc: 0.7621 - val_loss: 0.7234 - val_acc: 0.7000\n",
      "Epoch 46/100\n",
      "21/21 [==============================] - 74s 4s/step - loss: 0.5404 - acc: 0.7864 - val_loss: 0.5719 - val_acc: 0.7000\n",
      "Epoch 47/100\n",
      "21/21 [==============================] - 66s 3s/step - loss: 0.5525 - acc: 0.7670 - val_loss: 0.8004 - val_acc: 0.6400\n",
      "Epoch 48/100\n",
      "21/21 [==============================] - 66s 3s/step - loss: 0.6155 - acc: 0.7573 - val_loss: 0.6358 - val_acc: 0.7200\n",
      "Epoch 49/100\n",
      "21/21 [==============================] - 70s 3s/step - loss: 0.6035 - acc: 0.7233 - val_loss: 0.9068 - val_acc: 0.5400\n",
      "Epoch 50/100\n",
      "21/21 [==============================] - 72s 3s/step - loss: 0.5507 - acc: 0.7670 - val_loss: 0.7026 - val_acc: 0.7000\n",
      "Epoch 51/100\n",
      "21/21 [==============================] - 68s 3s/step - loss: 0.5346 - acc: 0.7718 - val_loss: 0.7747 - val_acc: 0.6200\n",
      "Epoch 52/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.5631 - acc: 0.7667 - val_loss: 0.6148 - val_acc: 0.6800\n",
      "Epoch 53/100\n",
      "21/21 [==============================] - 73s 3s/step - loss: 0.5081 - acc: 0.7816 - val_loss: 0.8564 - val_acc: 0.6600\n",
      "Epoch 54/100\n",
      "21/21 [==============================] - 76s 4s/step - loss: 0.5436 - acc: 0.7913 - val_loss: 0.7360 - val_acc: 0.6800\n",
      "Epoch 55/100\n",
      "21/21 [==============================] - 68s 3s/step - loss: 0.5652 - acc: 0.7476 - val_loss: 0.8324 - val_acc: 0.6800\n",
      "Epoch 56/100\n",
      "21/21 [==============================] - 73s 3s/step - loss: 0.5786 - acc: 0.7427 - val_loss: 0.8425 - val_acc: 0.6400\n",
      "Epoch 57/100\n",
      "21/21 [==============================] - 68s 3s/step - loss: 0.5638 - acc: 0.7282 - val_loss: 0.5901 - val_acc: 0.7600\n",
      "Epoch 58/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.5494 - acc: 0.7476 - val_loss: 0.5829 - val_acc: 0.7600\n",
      "Epoch 59/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.5598 - acc: 0.7573 - val_loss: 0.6610 - val_acc: 0.6000\n",
      "Epoch 60/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.5162 - acc: 0.7573 - val_loss: 0.6419 - val_acc: 0.6400\n",
      "Epoch 61/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.5650 - acc: 0.7282 - val_loss: 0.5991 - val_acc: 0.7600\n",
      "Epoch 62/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.5455 - acc: 0.7670 - val_loss: 0.5701 - val_acc: 0.7400\n",
      "Epoch 63/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.5706 - acc: 0.7427 - val_loss: 0.6488 - val_acc: 0.7000\n",
      "Epoch 64/100\n",
      "21/21 [==============================] - 72s 3s/step - loss: 0.5495 - acc: 0.7767 - val_loss: 0.6517 - val_acc: 0.7200\n",
      "Epoch 65/100\n",
      "21/21 [==============================] - 66s 3s/step - loss: 0.5267 - acc: 0.7864 - val_loss: 0.6769 - val_acc: 0.7000\n",
      "Epoch 66/100\n",
      "21/21 [==============================] - 76s 4s/step - loss: 0.5111 - acc: 0.7857 - val_loss: 0.7348 - val_acc: 0.6600\n",
      "Epoch 67/100\n",
      "21/21 [==============================] - 68s 3s/step - loss: 0.4734 - acc: 0.8252 - val_loss: 0.5517 - val_acc: 0.7400\n",
      "Epoch 68/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.5352 - acc: 0.7961 - val_loss: 0.8462 - val_acc: 0.6800\n",
      "Epoch 69/100\n",
      "21/21 [==============================] - 66s 3s/step - loss: 0.5727 - acc: 0.7670 - val_loss: 0.6732 - val_acc: 0.7600\n",
      "Epoch 70/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.5697 - acc: 0.7670 - val_loss: 0.6238 - val_acc: 0.6600\n",
      "Epoch 71/100\n",
      "21/21 [==============================] - 70s 3s/step - loss: 0.5606 - acc: 0.7670 - val_loss: 0.8438 - val_acc: 0.6000\n",
      "Epoch 72/100\n",
      "21/21 [==============================] - 68s 3s/step - loss: 0.5461 - acc: 0.7718 - val_loss: 0.6238 - val_acc: 0.7400\n",
      "Epoch 73/100\n",
      "21/21 [==============================] - 74s 4s/step - loss: 0.5965 - acc: 0.7476 - val_loss: 0.4831 - val_acc: 0.7600\n",
      "Epoch 74/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.4972 - acc: 0.7816 - val_loss: 0.6426 - val_acc: 0.7200\n",
      "Epoch 75/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.5378 - acc: 0.8107 - val_loss: 0.7613 - val_acc: 0.6800\n",
      "Epoch 76/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.5068 - acc: 0.8058 - val_loss: 0.8575 - val_acc: 0.6600\n",
      "Epoch 77/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.5610 - acc: 0.7476 - val_loss: 0.7318 - val_acc: 0.7200\n",
      "Epoch 78/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.5256 - acc: 0.7670 - val_loss: 0.6099 - val_acc: 0.7200\n",
      "Epoch 79/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.4928 - acc: 0.8204 - val_loss: 0.5245 - val_acc: 0.7400\n",
      "Epoch 80/100\n",
      "21/21 [==============================] - 68s 3s/step - loss: 0.5503 - acc: 0.7524 - val_loss: 0.6653 - val_acc: 0.6200\n",
      "Epoch 81/100\n",
      "21/21 [==============================] - 66s 3s/step - loss: 0.5619 - acc: 0.7573 - val_loss: 0.6939 - val_acc: 0.6400\n",
      "Epoch 82/100\n",
      "21/21 [==============================] - 67s 3s/step - loss: 0.5407 - acc: 0.7379 - val_loss: 0.6574 - val_acc: 0.6600\n",
      "Epoch 83/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.4767 - acc: 0.7913 - val_loss: 0.7607 - val_acc: 0.5400\n",
      "Epoch 84/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.4603 - acc: 0.8155 - val_loss: 0.6997 - val_acc: 0.7200\n",
      "Epoch 85/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.4556 - acc: 0.7961 - val_loss: 0.7430 - val_acc: 0.6200\n",
      "Epoch 86/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.5102 - acc: 0.8204 - val_loss: 0.7158 - val_acc: 0.6600\n",
      "Epoch 87/100\n",
      "21/21 [==============================] - 65s 3s/step - loss: 0.5002 - acc: 0.7810 - val_loss: 0.6548 - val_acc: 0.6000\n",
      "Epoch 88/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.4876 - acc: 0.7767 - val_loss: 0.7578 - val_acc: 0.6400\n",
      "Epoch 89/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.5028 - acc: 0.7913 - val_loss: 0.7843 - val_acc: 0.7000\n",
      "Epoch 90/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.4556 - acc: 0.8058 - val_loss: 0.8094 - val_acc: 0.6800\n",
      "Epoch 91/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.5060 - acc: 0.7816 - val_loss: 0.7303 - val_acc: 0.6600\n",
      "Epoch 92/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.4337 - acc: 0.8107 - val_loss: 0.8861 - val_acc: 0.5800\n",
      "Epoch 93/100\n",
      "21/21 [==============================] - 62s 3s/step - loss: 0.4943 - acc: 0.7816 - val_loss: 0.8048 - val_acc: 0.6400\n",
      "Epoch 94/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.4923 - acc: 0.7961 - val_loss: 1.0071 - val_acc: 0.6400\n",
      "Epoch 95/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.5634 - acc: 0.7427 - val_loss: 0.7142 - val_acc: 0.6800\n",
      "Epoch 96/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.5167 - acc: 0.7961 - val_loss: 0.6059 - val_acc: 0.6800\n",
      "Epoch 97/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.4929 - acc: 0.8010 - val_loss: 0.5782 - val_acc: 0.7000\n",
      "Epoch 98/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.4356 - acc: 0.8252 - val_loss: 0.5561 - val_acc: 0.7200\n",
      "Epoch 99/100\n",
      "21/21 [==============================] - 63s 3s/step - loss: 0.4297 - acc: 0.8058 - val_loss: 0.6671 - val_acc: 0.6400\n",
      "Epoch 100/100\n",
      "21/21 [==============================] - 64s 3s/step - loss: 0.4960 - acc: 0.8204 - val_loss: 0.5952 - val_acc: 0.7200\n"
     ]
    }
   ],
   "source": [
    "#FIT MODEL\n",
    "print(len(train_batches))\n",
    "print(len(valid_batches))\n",
    "\n",
    "STEP_SIZE_TRAIN=train_batches.n//train_batches.batch_size\n",
    "STEP_SIZE_VALID=valid_batches.n//valid_batches.batch_size\n",
    "\n",
    "raise NotImplementedError(\"Use the model.fit function to train your network\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "#### [5 points] Plot Accuracy and Loss During Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "nbpresent": {
     "id": "ff342098-784a-4e20-ac34-b74ca8ebe839"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "raise NotImplementedError(\"Plot the accuracy and the loss during training\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "#### Testing Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 36 images belonging to 4 classes.\n",
      "36\n",
      "WARNING:tensorflow:sample_weight modes were coerced from\n",
      "  ...\n",
      "    to  \n",
      "  ['...']\n",
      "36/36 [==============================] - 10s 274ms/step - loss: 0.7570 - acc: 0.6944\n",
      "Test loss: 0.757028494571235\n",
      "Test accuracy: 0.6944444\n"
     ]
    }
   ],
   "source": [
    "test_datagen = ImageDataGenerator(rescale=1. / 255)\n",
    "\n",
    "eval_generator = test_datagen.flow_from_directory(TEST_DIR,target_size=IMAGE_SIZE,\n",
    "                                                  batch_size=1,shuffle=True,seed=42,class_mode=\"categorical\")\n",
    "eval_generator.reset()\n",
    "print(len(eval_generator))\n",
    "x = model.evaluate_generator(eval_generator,steps = np.ceil(len(eval_generator)),\n",
    "                           use_multiprocessing = False,verbose = 1,workers=1)\n",
    "print('Test loss:' , x[0])\n",
    "print('Test accuracy:',x[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "## [10 points] TSNE Plot\n",
    "t-Distributed Stochastic Neighbor Embedding (t-SNE) is a widely used technique for dimensionality reduction that is particularly well suited for the visualization of high-dimensional datasets. After training is complete, extract features from a specific deep layer of your choice, use t-SNE to reduce the dimensionality of your extracted features to 2 dimensions and plot the resulting 2D features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 270 images belonging to 4 classes.\n",
      "{'covid': 0, 'normal': 1, 'pneumonia_bac': 2, 'pneumonia_vir': 3}\n",
      "Extracting features for 270 images.\n",
      "270/270 [==============================] - 71s 265ms/step\n",
      "Training TSNE model.\n"
     ]
    },
    {
     "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": [
    "from sklearn.manifold import TSNE\n",
    "\n",
    "intermediate_layer_model = models.Model(inputs=model.input,\n",
    "                                        outputs=model.get_layer('feature_dense').output)\n",
    "\n",
    "tsne_eval_generator = test_datagen.flow_from_directory(DATASET_PATH,target_size=IMAGE_SIZE,\n",
    "                                                  batch_size=1,shuffle=True,seed=42,class_mode=\"categorical\")\n",
    "\n",
    "raise NotImplementedError(\"Extract features from the tsne_data_generator and fit a t-SNE model for the features,\"\n",
    "                          \"and plot the resulting 2D features of the four classes.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<!-- ## Task 2: COVID-19 vs Normal vs Tertiary Pneumonia (Bacterial and Viral) -->"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "<!-- #### Renew Training Batch and Validation Batch -->"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "66140980-2aa4-457f-b1df-74c10c234cc2"
    }
   },
   "source": [
    "<!-- #### Renew Model -->"
   ]
  }
 ],
 "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": "3.7.4"
  },
  "latex_envs": {
   "LaTeX_envs_menu_present": true,
   "autoclose": false,
   "autocomplete": true,
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 1,
   "hotkeys": {
    "equation": "Ctrl-E",
    "itemize": "Ctrl-I"
   },
   "labels_anchors": false,
   "latex_user_defs": false,
   "report_style_numbering": false,
   "user_envs_cfg": false
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}