HW4.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Deep Learning in Medicine\n",
    "### BMSC-GA 4493, BMIN-GA 3007 \n",
    "### Homework 4\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note:** If you need to write mathematical terms, you can type your answeres in a Markdown Cell via LaTex \n",
    "\n",
    "See: <a href=\"https://stackoverflow.com/questions/13208286/how-to-write-latex-in-ipython-notebook\">here</a> if you have issues. To see basic LaTex notation see: <a href=\"https://en.wikibooks.org/wiki/LaTeX/Mathematics\"> here </a>.\n",
    "\n",
    "**Submission instruction**: Upload and Submit your final jupyter notebook with necessary files in <a href='http://newclasses.nyu.edu'>newclasses.nyu.edu</a>. If you use code or script from web, please give a link to the code in your answers.\n",
    "\n",
    "**Submission deadline:** Saturday May 12th 2018 (5:00 PM)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Question 1: Literature Review: DeepPatient (Total 25 points + 10 points for Bonus question)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Read this paper:\n",
    "\n",
    "Riccardo Miotto, Li Li, Brian A. Kidd & Joel T. Dudley, \"Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the Electronic Health Records\" Scientific Reports, 2016 https://www.nature.com/articles/srep26094.pdf\n",
    "\n",
    "We are interested in understanding the task, the methods that is proposed in this publication, technical aspects of the implementation, and possible future work. After you read the full article answer the following questions. Describe you answers in your own words.  \n",
    "\n",
    "### 1.1) (5 points) \n",
    "What type of learning algorithm is used (supervised, semi-supervised or unsupervised)? What is the reason for selecting this type of learning algorithm?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The learning algithm is unsupervised. The reason this was chosen is that generally an expert is required to select features and patterns are identified within the data. This approach does not allow the algorithm to identify new patterns or features within the data. An alternative approach is data driven feature selection which results in sparse vectors which are not suitable for the approach taken by the authors."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.2) (5 points)\n",
    "What type of neural network architecture is used in the paper? What is the reason for selecting this type of network?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A multi-layer neural network in a deep learning architecture which consists of a stack of denoising autoencoders\n",
    "\n",
    "By combining multiple layers the model is able to achieve higher levels of abstraction and use this higher level of abstraction to identify patterns within the data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.3) (5 points) \n",
    "What is the loss function? Whay kind of approaches are used to improve model generalization?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A reconstruction cross-entropy loss function was used.\n",
    "Masking noise algorithm which randomly sets some input features to zero. This similates a missed input within an EHR."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.4) (5 points) \n",
    "How many hidden layers and units does DeepPatient has? What type of activation function is used?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Three layers of auto encoders using Sigmoid activation functions."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.5) (5 points) \n",
    "What are the evaluation metrics used for model comparison? Explain why those metrics were chosen?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "AUC-ROC, accuracy and F score. These metrics cover a wide range of methods of evaluating a model. AUROC compares the true positive prediction rate to the false positive prediction rate. Accuracy reflects the proprotion of true results and the f score is the models accuracy in terms of precision and recall."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.6) (Bonus: maximum 10 points) \n",
    "How did the authors decide on using specific number of hidden layers in the article? What other architectures would you try?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The authors compared the evaluation metrics for models of varying depth and observed that after 3 layers there was little to no improvement in the performance of the model.\n",
    "\n",
    "I would try to incorperate a recursive nerual network into the model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Question 2: Transfer learning for disease classification (Total 75 points + 20 points in bonus question)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this part of the howework, we will revisit the disease classification task we worked on HW2. As opposed to developing our own CNN network as we did on HW2, in this HW we are interested in using transfer learning for the disease classification task.  we will use ResNet50 model to achieve this goal. Here is the link for the ResNet paper: https://arxiv.org/pdf/1512.03385.pdf You will use ResNet50 model as a fixed feature extractor in this question. \n",
    "\n",
    "As a reminder: we focused on classifiying the lung disease using chest x-ray dataset provided by NIH (https://www.nih.gov/news-events/news-releases/nih-clinical-center-provides-one-largest-publicly-available-chest-x-ray-datasets-scientific-community). Please go over the following paper for the details of the dataset: https://arxiv.org/pdf/1705.02315.pdf \n",
    "\n",
    "You need to use HPC for training part of this question, as your computer's CPU will not be fast enough to compute learning iterations. In case you use HPC, please have your code/scripts uploaded under the questions and provide the required plots and tables there as well. Data is available in HPC under /beegfs/ga4493/data/HW2 folder. We are interested in classifying infiltration, pneumothorax, cardiomegaly and *not*(infiltration OR pneumothorax OR cardiomegaly) cases. By saying so we have 4 classes that we want to identify by modelling a deep CNN.\n",
    "\n",
    "Due to time limitations you do not need to your models using the whole dataset, we will use HW2_RandomtrainSet.csv, HW2_testSet.csv and HW2_RandomvalidationSet.csv provided under /beegfs/ga4493/data/HW2 folder for defining train, test and validation set samples that are generated in HW2 Q3.1.\n",
    "In these .csv files 4 classes were defined as :\n",
    "- 1 infiltration\n",
    "- 2 pneumothorax\n",
    "- 3 cardiomegaly\n",
    "- 0 for all other diseases (doesnt have infiltration OR pneumothorax OR cardiomegaly) or NoFinding\n",
    "\n",
    "## 2.1) Define dataloaders for train, test and validations sets\n",
    "We used Dataset class defined in HW2. Here, we have a copy of it. You need to make necessary changes to make the dataset class is capable of feeding the X-ray dataset into resnet model for transfer learning. Remember X-ray data has grayscale images of 1024x1024 pixels and resnet50 model requires RGB images of size 224x224. Add necessary code to the specified locations: \n",
    "\n",
    "### 2.1.a) (10 points) \n",
    "Define data transforms required for train and validation/test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.utils.data import Dataset, DataLoader\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import os\n",
    "from skimage import io\n",
    "import torch\n",
    "from torchvision import transforms\n",
    "import torchvision\n",
    "from skimage import color, transform\n",
    "from skimage.transform import resize\n",
    "\n",
    "# Torchvision transforms are buggy in the version of pytorch I am using\n",
    "# for this reason I am using skimage to resize\n",
    "train_transform = True\n",
    "\n",
    "validation_transform = True\n",
    "\n",
    "class ChestXrayDataset(Dataset):\n",
    "    \"\"\"Chest X-ray dataset from https://nihcc.app.box.com/v/ChestXray-NIHCC.\"\"\"\n",
    "\n",
    "    def __init__(self, csv_file, root_dir, transform=None):\n",
    "        \"\"\"\n",
    "        Args:\n",
    "            csv_file (string): Path to the csv file filename information.\n",
    "            root_dir (string): Directory with all the images.\n",
    "            transform (callable, optional): Optional transform to be applied\n",
    "                on a sample.\n",
    "        \"\"\"\n",
    "        self.data_frame = pd.read_csv(csv_file)\n",
    "        self.root_dir = root_dir\n",
    "        self.transform = transform\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.data_frame)\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        img_name = os.path.join(self.root_dir,\n",
    "                                self.data_frame.iloc[idx, 0])\n",
    "        \n",
    "        #some cases io.imread brings more channels than 1 due to bitsize issues\n",
    "        # for this reason load the image as gray\n",
    "        image = io.imread(img_name, as_grey=True)\n",
    "\n",
    "        image_class = self.data_frame.iloc[idx, -1]\n",
    "        \n",
    "        # normalize the image - I think that resnet does this\n",
    "#         image = (image - image.mean()) / image.std()\n",
    "        \n",
    "        # this will resize the grayscale image\n",
    "        if self.transform:\n",
    "            image = transform.resize(image, (224, 224), mode='reflect')\n",
    "            \n",
    "        # replicate the image into 3 RGB channels\n",
    "        image=np.repeat(image[None,...],3,axis=0)\n",
    "            \n",
    "        sample = {'x': image, 'y': image_class}\n",
    "\n",
    "        return sample\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.1.b) (5 points)\n",
    "Define train, validation and test dataloaders loaders using the dataset class defined in Q.3.1.a and .csv files: HW2_RandomtrainSet.csv, HW2_testSet.csv and HW2_RandomvalidationSet.csv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'train': 256, 'val': 64}\n"
     ]
    }
   ],
   "source": [
    "## Defining the data loaders with emprical batch size \n",
    "BATCH_SIZE = 16\n",
    "## data loaders\n",
    "chestXray_TrainData = ChestXrayDataset('HW2_randomTrainSet.csv', '/beegfs/ga4493/data/HW2/images', train_transform)\n",
    "train_loader = DataLoader(chestXray_TrainData, batch_size=BATCH_SIZE,\n",
    "                        shuffle=True, num_workers=4)\n",
    "\n",
    "\n",
    "chestXray_ValidationData = ChestXrayDataset('HW2_randomValidationSet.csv', '/beegfs/ga4493/data/HW2/images', validation_transform)\n",
    "validation_loader = DataLoader(chestXray_ValidationData, batch_size=BATCH_SIZE,\n",
    "                        shuffle=False, num_workers=4)\n",
    "\n",
    "chestXray_TestData = ChestXrayDataset('HW2_testSet.csv', '/beegfs/ga4493/data/HW2/images', True)\n",
    "test_loader = DataLoader(chestXray_TestData, batch_size=BATCH_SIZE,\n",
    "                        shuffle=False, num_workers=4)\n",
    "\n",
    "dataset_sizes = {'train': len(chestXray_TrainData), 'val': len(chestXray_ValidationData)}\n",
    "print(dataset_sizes)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Helper functions --> You dont need to add/remove anything here. Functions implemented on HW2 will be used here. if you want, you can use your own functions as well"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "import copy\n",
    "import numpy as np\n",
    "from torch.autograd import Variable\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "use_gpu = torch.cuda.is_available()\n",
    "\n",
    "def train_model(model, criterion, optimizer, num_epochs=25, trainVal=['train','val'],verbose=True):\n",
    "    since = time.time()\n",
    "    \n",
    "    best_model_wts = copy.deepcopy(model.state_dict())\n",
    "    best_acc = 0.0\n",
    "\n",
    "    loss2plot = np.zeros([2,num_epochs])\n",
    "    acc2plot  = np.zeros([2,num_epochs])\n",
    "\n",
    "    for epoch in range(num_epochs):\n",
    "        if verbose:\n",
    "            print('Epoch {}/{}'.format(epoch, num_epochs - 1))\n",
    "            print('-' * 10)\n",
    "\n",
    "        # Each epoch has a training and validation phase\n",
    "        for phase in trainVal:\n",
    "            if phase == 'train':\n",
    "                imageLoader = train_loader\n",
    "            else:\n",
    "                imageLoader = validation_loader\n",
    "\n",
    "            running_loss = 0.0\n",
    "            running_corrects = 0\n",
    "\n",
    "            # Iterate over data.\n",
    "            for sample_batched in imageLoader:\n",
    "                # get the inputs\n",
    "                #print(sample_batched)\n",
    "                inputs = sample_batched['x']\n",
    "                labels = sample_batched['y']\n",
    "\n",
    "                # wrap them in Variable\n",
    "                if use_gpu:\n",
    "                    inputs = Variable(inputs).type(torch.FloatTensor).cuda()\n",
    "                    labels = Variable(labels).type(torch.LongTensor).cuda()\n",
    "                else:\n",
    "                    inputs, labels = Variable(inputs).type(torch.FloatTensor), Variable(labels).type(torch.LongTensor)\n",
    "\n",
    "                # zero the parameter gradients\n",
    "                optimizer.zero_grad()\n",
    "\n",
    "                # forward\n",
    "                outputs = model(inputs)\n",
    "                _, preds = torch.max(outputs.data, 1)\n",
    "                loss = criterion(outputs, labels)\n",
    "\n",
    "                # backward + optimize only if in training phase\n",
    "                if phase == 'train':\n",
    "                    loss.backward()\n",
    "                    optimizer.step()\n",
    "\n",
    "                # statistics\n",
    "                running_loss += loss.data[0] * inputs.size(0)\n",
    "                running_corrects += torch.sum(preds == labels.data)\n",
    "#                 print(labels.data,preds,preds == labels.data,outputs.data)\n",
    "\n",
    "            epoch_loss = running_loss / dataset_sizes[phase]\n",
    "            epoch_acc = running_corrects / dataset_sizes[phase]\n",
    "\n",
    "            if verbose:\n",
    "                print('{} Loss: {:.4f} Acc: {:.4f}'.format(\n",
    "                    phase, epoch_loss, epoch_acc))\n",
    "            \n",
    "            if phase == 'train':\n",
    "                loss2plot[0,epoch] = epoch_loss\n",
    "                acc2plot[0,epoch] = epoch_acc\n",
    "            else:\n",
    "                loss2plot[1,epoch] = epoch_loss\n",
    "                acc2plot[1,epoch] = epoch_acc\n",
    "                    \n",
    "\n",
    "            # deep copy the model\n",
    "            if phase == 'val' and epoch_acc > best_acc:\n",
    "                best_acc = epoch_acc\n",
    "                best_model_wts = copy.deepcopy(model.state_dict())\n",
    "\n",
    "        if verbose:\n",
    "            print()\n",
    "\n",
    "    time_elapsed = time.time() - since\n",
    "    print('Training complete in {:.0f}m {:.0f}s'.format(\n",
    "        time_elapsed // 60, time_elapsed % 60))\n",
    "    print('Best val Acc: {:4f}'.format(best_acc))\n",
    "    \n",
    "    for phase in trainVal:\n",
    "        if phase == 'train':\n",
    "            idx=0\n",
    "        else:\n",
    "            idx=1\n",
    "            \n",
    "        fig = plt.figure()\n",
    "        \n",
    "    \n",
    "        a = fig.add_subplot(2,2,2*idx+1)\n",
    "        plt.plot(loss2plot[idx,:])\n",
    "        plt.title('Loss per epoch for ' + phase)\n",
    "        #plt.suptitle('Curves for ' + phase)\n",
    "\n",
    "        a = fig.add_subplot(2,2,2*idx+2)\n",
    "        plt.plot(acc2plot[idx,:])\n",
    "        plt.title('Accuracy per epoch for ' + phase)\n",
    "        plt.show()\n",
    "\n",
    "        #plt.plot(loss2plot[idx,:]);plt.title('Loss per epoch for ' + phase); plt.show()\n",
    "        #plt.plot(acc2plot[idx,:]);plt.title('Accuracy per epoch for ' + phase); plt.show()\n",
    "\n",
    "    # load best model weights\n",
    "    model.load_state_dict(best_model_wts)\n",
    "    \n",
    "    return model\n",
    "\n",
    "# This is the place we predict the disease from a model trained, output for this function is \n",
    "#the target values and probabilty of each image having a disease \n",
    "import torch.nn.functional as F\n",
    "def inference(model_ft,loader):\n",
    "    use_gpu = 1\n",
    "    model_ft.eval()\n",
    "    whole_output =[]\n",
    "    whole_target = []\n",
    "\n",
    "    for valData in loader:\n",
    "        data = valData['x']\n",
    "        target = valData['y']\n",
    "        if use_gpu:\n",
    "            data = Variable(data,volatile=True).type(torch.FloatTensor).cuda()\n",
    "            target = Variable(target,volatile=True).type(torch.LongTensor).cuda()\n",
    "        else:\n",
    "            data, target = Variable(data,volatile=True).type(torch.FloatTensor), Variable(target,volatile=True).type(torch.LongTensor)\n",
    "\n",
    "        output = F.softmax(model_ft(data))\n",
    "        whole_output.append( output.cpu().data.numpy())\n",
    "        whole_target.append( valData['y'].numpy())\n",
    "\n",
    "    whole_output = np.concatenate(whole_output)\n",
    "    whole_target = np.concatenate(whole_target)\n",
    "\n",
    "    y_score = whole_output\n",
    "    y_target = label_binarize(whole_target, classes=[0, 1, 2, 3])\n",
    "    \n",
    "    return y_score, y_target\n",
    "\n",
    "# this function AUC of ROC for each disease seperately and also macro and micro averages,\n",
    "# we will use macro average to compare different models we will train. \n",
    "from sklearn.metrics import roc_curve, auc\n",
    "from sklearn.preprocessing import label_binarize\n",
    "from scipy import interp\n",
    "from itertools import cycle\n",
    "\n",
    "def get_AUC(y_score, y_target,plotROC=False):\n",
    "    n_classes = y_score.shape[1]\n",
    "\n",
    "    # Compute ROC curve and ROC area for each class\n",
    "    fpr = dict()\n",
    "    tpr = dict()\n",
    "    roc_auc = dict()\n",
    "    for i in range(n_classes):\n",
    "        fpr[i], tpr[i], _ = roc_curve(y_target[:, i], y_score[:, i])\n",
    "        roc_auc[i] = auc(fpr[i], tpr[i])\n",
    "\n",
    "    # Compute micro-average ROC curve and ROC area\n",
    "    fpr[\"micro\"], tpr[\"micro\"], _ = roc_curve(y_target.ravel(), y_score.ravel())\n",
    "    roc_auc[\"micro\"] = auc(fpr[\"micro\"], tpr[\"micro\"])\n",
    "\n",
    "\n",
    "    # Compute macro-average ROC curve and ROC area\n",
    "\n",
    "    # First aggregate all false positive rates\n",
    "    all_fpr = np.unique(np.concatenate([fpr[i] for i in range(n_classes)]))\n",
    "\n",
    "    # Then interpolate all ROC curves at this points\n",
    "    mean_tpr = np.zeros_like(all_fpr)\n",
    "    for i in range(n_classes):\n",
    "        mean_tpr += interp(all_fpr, fpr[i], tpr[i])\n",
    "\n",
    "    # Finally average it and compute AUC\n",
    "    mean_tpr /= n_classes\n",
    "\n",
    "    fpr[\"macro\"] = all_fpr\n",
    "    tpr[\"macro\"] = mean_tpr\n",
    "    roc_auc[\"macro\"] = auc(fpr[\"macro\"], tpr[\"macro\"])\n",
    "    \n",
    "    if plotROC:\n",
    "        lw = 2\n",
    "        # Plot all ROC curves\n",
    "        plt.figure()\n",
    "        plt.plot(fpr[\"micro\"], tpr[\"micro\"],\n",
    "                 label='micro-average ROC curve (area = {0:0.2f})'\n",
    "                       ''.format(roc_auc[\"micro\"]),\n",
    "                 color='deeppink', linestyle=':', linewidth=4)\n",
    "\n",
    "        plt.plot(fpr[\"macro\"], tpr[\"macro\"],\n",
    "                 label='macro-average ROC curve (area = {0:0.2f})'\n",
    "                       ''.format(roc_auc[\"macro\"]),\n",
    "                 color='navy', linestyle=':', linewidth=4)\n",
    "\n",
    "        colors = cycle(['aqua', 'darkorange', 'cornflowerblue'])\n",
    "        for i, color in zip(range(n_classes), colors):\n",
    "            plt.plot(fpr[i], tpr[i], color=color, lw=lw,\n",
    "                     label='ROC curve of class {0} (area = {1:0.2f})'\n",
    "                     ''.format(i, roc_auc[i]))\n",
    "\n",
    "        plt.plot([0, 1], [0, 1], 'k--', lw=lw)\n",
    "        plt.xlim([0.0, 1.0])\n",
    "        plt.ylim([0.0, 1.05])\n",
    "        plt.xlabel('False Positive Rate')\n",
    "        plt.ylabel('True Positive Rate')\n",
    "        plt.title('Some extension of Receiver operating characteristic to multi-class')\n",
    "        plt.legend(loc=\"lower right\")\n",
    "        plt.show()\n",
    "\n",
    "    return roc_auc\n",
    "\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import itertools\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Blues):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        print(\"Normalized confusion matrix\")\n",
    "    else:\n",
    "        print('Confusion matrix, without normalization')\n",
    "\n",
    "    print(cm)\n",
    "\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    fmt = '.2f' if normalize else 'd'\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt),\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.2) Loading the pre-trained CNN model, training the model and analyzing results (TL as a feature extractor) \n",
    "\n",
    "### 2.2.a) (6 points)\n",
    "Since now we can import images using dataloaders, next step is to load the pretrained resnet50 network from torchvision.models and prepare it for transfer learning for using resnet model as a feature extractor (see slide #12 for Lecture 12). Write the code for designing resnet50 architecture and loading the weights from ImageNet trained network. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ResNet (\n",
       "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
       "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "  (relu): ReLU (inplace)\n",
       "  (maxpool): MaxPool2d (size=(3, 3), stride=(2, 2), padding=(1, 1), dilation=(1, 1))\n",
       "  (layer1): Sequential (\n",
       "    (0): Bottleneck (\n",
       "      (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "      (downsample): Sequential (\n",
       "        (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck (\n",
       "      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (2): Bottleneck (\n",
       "      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "  )\n",
       "  (layer2): Sequential (\n",
       "    (0): Bottleneck (\n",
       "      (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "      (downsample): Sequential (\n",
       "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck (\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (2): Bottleneck (\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (3): Bottleneck (\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "  )\n",
       "  (layer3): Sequential (\n",
       "    (0): Bottleneck (\n",
       "      (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "      (downsample): Sequential (\n",
       "        (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (2): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (3): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (4): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (5): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "  )\n",
       "  (layer4): Sequential (\n",
       "    (0): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "      (downsample): Sequential (\n",
       "        (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck (\n",
       "      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (2): Bottleneck (\n",
       "      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "  )\n",
       "  (avgpool): AvgPool2d (size=7, stride=7, padding=0, ceil_mode=False, count_include_pad=True)\n",
       "  (fc): Linear (2048 -> 4)\n",
       ")"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resnet50 = torchvision.models.resnet50(pretrained=True)\n",
    "# from https://pytorch.org/tutorials/beginner/transfer_learning_tutorial.html#convnet-as-fixed-feature-extractor\n",
    "for param in resnet50.parameters():\n",
    "    param.requires_grad = False\n",
    "num_ftrs = resnet50.fc.in_features\n",
    "resnet50.fc = nn.Linear(num_ftrs, 4)\n",
    "resnet50.cuda()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2.b) (4 points)\n",
    "Define the loss and the optimizer you want to use "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.SGD(resnet50.fc.parameters(), lr=0.001, momentum=0.9)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2.c) (4 points)\n",
    "Train the model for 100 epochs and save the weights of the best model using the validation loss. Plot train and validation loss curves"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training complete in 8m 1s\n",
      "Best val Acc: 0.437500\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_FE = train_model(resnet50, criterion, optimizer,\n",
    "                       num_epochs=100, trainVal=['train','val'],verbose=False) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2.d) (4 points)\n",
    "Plot ROC curve of the best model using the dataloader of the test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "macro AUC for the model trained on full data is :  0.503595601868\n"
     ]
    }
   ],
   "source": [
    "y_s, y_t=inference(model_FE,test_loader)\n",
    "r_AUC_wholeSet = get_AUC(y_s, y_t,True)\n",
    "print('macro AUC for the model trained on full data is : ', r_AUC_wholeSet['macro'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2.e) (4 points)\n",
    "Compute the confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix, without normalization\n",
      "[[1684    2    0    0]\n",
      " [1932    7    0    0]\n",
      " [ 438    0    0    0]\n",
      " [ 217    1    0    0]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Compute confusion matrix\n",
    "cnf_matrix = confusion_matrix(y_t.dot([0,1,2,3]), np.argmax(y_s,axis=1))\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "# Plot non-normalized confusion matrix\n",
    "plt.figure()\n",
    "class_names = ['0: None Class','1: Infiltration','2: Pneumothorax','3: Cardiomegaly']\n",
    "plot_confusion_matrix(cnf_matrix, classes=class_names,\n",
    "                      title='Confusion matrix')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.3) Loading the pre-trained CNN model, training the model and analyzing results (TL with fine-tuning) \n",
    "\n",
    "### 2.3.a) (6 points)\n",
    "Perform a transfer learning using resnet50 model as a fine-tuning base model (see slide #13 for Lecture 12).\n",
    "Write the code for designing resnet50 architecture and loading the weights from ImageNet trained network.                                                                       "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ResNet (\n",
       "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
       "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "  (relu): ReLU (inplace)\n",
       "  (maxpool): MaxPool2d (size=(3, 3), stride=(2, 2), padding=(1, 1), dilation=(1, 1))\n",
       "  (layer1): Sequential (\n",
       "    (0): Bottleneck (\n",
       "      (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "      (downsample): Sequential (\n",
       "        (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck (\n",
       "      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (2): Bottleneck (\n",
       "      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "  )\n",
       "  (layer2): Sequential (\n",
       "    (0): Bottleneck (\n",
       "      (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "      (downsample): Sequential (\n",
       "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck (\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (2): Bottleneck (\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (3): Bottleneck (\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "  )\n",
       "  (layer3): Sequential (\n",
       "    (0): Bottleneck (\n",
       "      (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "      (downsample): Sequential (\n",
       "        (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (2): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (3): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (4): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (5): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "  )\n",
       "  (layer4): Sequential (\n",
       "    (0): Bottleneck (\n",
       "      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "      (downsample): Sequential (\n",
       "        (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck (\n",
       "      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "    (2): Bottleneck (\n",
       "      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True)\n",
       "      (relu): ReLU (inplace)\n",
       "    )\n",
       "  )\n",
       "  (avgpool): AvgPool2d (size=7, stride=7, padding=0, ceil_mode=False, count_include_pad=True)\n",
       "  (fc): Linear (2048 -> 4)\n",
       ")"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resnet50 = torchvision.models.resnet50(pretrained=True)\n",
    "# from https://pytorch.org/tutorials/beginner/transfer_learning_tutorial.html#finetuning-the-convnet\n",
    "num_ftrs = resnet50.fc.in_features\n",
    "resnet50.fc = nn.Linear(num_ftrs, 4)\n",
    "resnet50.cuda()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.3.b) (4 points)\n",
    "Define the loss and the optimizer you want to use"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.optim import lr_scheduler\n",
    "import torch.nn as nn\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.SGD(resnet50.parameters(), lr=0.001, momentum=0.9)\n",
    "\n",
    "# Decay LR by a factor of 0.1 every 7 epochs\n",
    "exp_lr_scheduler = lr_scheduler.StepLR(optimizer, step_size=7, gamma=0.1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.3.c) (4 points)\n",
    "Train the model for 100 epochs and save the weights of the best model using the validation loss. Plot train and validation loss curves"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training complete in 8m 54s\n",
      "Best val Acc: 0.453125\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAACSCAYAAACpHBqyAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4wLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvFvnyVgAAIABJREFUeJztnXmcHVWZ97/PXXpfk+7sG0tAAiMYI6AT0FFAYAR8XUFF4wbOiOCCvjivCx+3UUbx1RFHURg2ZRFc4sgIjoIoGkjAsAsmkJVO0ul0uju93uWZP865SeWmu+/t5d66t/v5fj796VtVp8556tSpXz31nFOnRFUxDMMwpgeRsA0wDMMwioeJvmEYxjTCRN8wDGMaYaJvGIYxjTDRNwzDmEaY6BuGYUwjTPSnMCJyv4h8YAzp/0lEdorIPhGZWUjb8kVEvicinw3bDqP8EZFVIvLHMaQ/WkTWi0iPiFxaSNvyRUROEZFnJ5JHyYm+iGwSkdPCtmO6ISJx4GrgDFWtU9WOSchzwudSVT+kql+cqC2Tjb+hdopIZdi2GAXjU8B9qlqvqt+eaGYicqWI3DKRPFT1D6p69ETyKDnRnwqISCxsG8bBbKAKeGqsO4pjzG2pTOsJEVkCnAIocG6Ryy67OitHmz2LGcf1AOM75vFeR2NGVUvqD9gEnDbCtg8CG4A9wGpgnl8vwDeBXUA38ARwnN92NvA00ANsBy4fIe9VwIPAd4Au4K/A6wLbG4HrgDafz5eAaNa+3wQ6gC8Nk38EuALY6NPcAczw25bgBOQi4EVfxuWBfSuB/++3veh/Vwa2nwes98e+ETjTr78f+KK3rQe4F2gZxrajgF5vwz7gd379q4C1vj7WAq8K7HM/8GWfdz9wZFaeNwNpv20fzmvKHOf7gS3AAz7tT4AdvpwHgGMD+dyQqU/gNcA24BP+XLcB7w2hjX7OH/fVwH9lbasGvgFs9sfzR6Dab1sJ/AnYC2wFVgXq8gNZbfGPgWUFPgz8DXjBr/uWz6MbeAQ4JZA+CvyLbws9fvtC4BrgG1n2rgY+NsJxKnAp8DywG/g3IBLY/j7gGaATuAdYPJrNw+R/cqA+HgNek9W+/hV42B/jL/DXi99+Lk6Q9/q0xwS2LQR+CrTjrrXvBOsV+Lq3+QXgrBFs+x2QAgZw7fconAbc5PPdDHwmUx/k0ADgTGAISPj8HhvpOgLe6+u1x9f9xYF8XgNsy9LLy4HHce3tdqBq1PZb7AsmjwtqE8OIPvBa3/CW40Tw3zkgGq/3DbsJdwM4Bpjrt7XhLwigGVg+QrmrgCTwMSAOvN1XYkaYfwZ8H6gFZvnGeHHWvh8BYviLPCv/y4A1wAJv//eBW/22JbiL5Faf/9/5hnWa3/4Fv+8soBV3oXzRbzvR23k67sYyH3hJoEFt9A222i9/dYTjz9gQ88szcBfGhf6YLvDLMwN5bwGO9dvjuc5loIyb/HFmxPB9QD0Hbm7rA/vcwMGin/T1Ecfd0PuA5iK30Q3APwMvx13EswPbrvF1Mx8nvq/yx7UYdxFf4G2fCZwQqMtcov8bf04ydfYun0cMdxPcgb/YgU/iHJ+jcdfD8T7tiTinISNULb7+Zo9wnArc58tdBDyXsRPnaGzAXWsxnAD+aTSbs/KejxPHs3Ht9nS/3Bqok+3Acb6t3AXc4rdlnJTTfV1+yttS4ev8MZz41uKeXlcG6jWBcx6jwD/5+pARjj/7vNyEu/nU49ryc8D7x6ABV2aOIauMg64j4B+BI/y5e7U/R8sD10C26D8MzPN1/QzwoVHbbzEvljwvqE0ML/rXAVcFluv8CVyCuyE8h/McIln7bQEuBhpylLsquwH4yrwQF/oYDJ5I3MV7X2DfLTnyf4aDnxzmevtjHBDDlwS2XwVc539vBM4ObHs9sMn//j7wzVEa7WcCy/8M/HqEtBkbMqJ/IfBwVpo/c7B3+oWxnMtAGYePsk+TT9Pol2/gYNHvz9jo1+0CTi5i+1zpz1uLX/4r3lPGiVc/cPww+30a+Nko5ymX6L82h12dmXKBZ4HzRmmHp/vflwB3j5Kn4p8aA+3nt/73f+MFL3DsfXhvP5fNwP8Fbs5adw/wnkCdfDWwbRnOU44CnwXuyCp7u28fr8Q5TLFhylwFbAgs13g75+Q6L77cIWBZYPvFwP2BvHNpwJUML/q5rqOfA5cFroFs0X9XYPkq4Huj5VdOMf15uEcqAFR1H84zmK+qv8OFZa4BdonItSLS4JO+GedNbBaR34vIK0cpY7v6mvNs9uUuxt2B20Rkr4jsxYntrEDarTnsXwz8LLD/M7jHx9kj5JEp+5Bjz9q2EHdTGIkdgd99uJtlPmSXmSl3/gj2joX9+4lIVES+KiIbRaQb14jBeaHD0aGqycDyWI5pMngPcK+q7vbLP/brwNlcxfDnI9d5ysVBdS0il4vIMyLS5dtTIwfqbLSybsQ9JeD/3zyGcoPtbjHwrUB73oPzTPNtH4uBt2b293msxDlDI5Udxx1jthakfdr5uGPfnNVGguwI7Nfnf+bTflp8+dnX4aReDwAicpaIrBGRPb5ezmbk6wHGeI2Xk+i/iGsoAIhILe6RdTuAqn5bVV+O8wiOwj3ioqprVfU8nED/HBdLH4n5IiKB5UW+3K04T79FVZv8X4OqHhtIG7xZDMdWXPywKfBXparbA2kWDlP2IceetW0r7lFwsskuM1Nu0N5cxzzS9uD6d+BCBafhhGuJXy+UGCJSDbwNeLWI7BCRHbhw4PEicjwu/DjA8OdjtPPUi/M6M8wZJs3+OhORU3AhjbfhQltNuBBfps5GK+sW4Dxv7zG4a2I0RmqTW3HhzWB7rlbVPw1n8zBsxXn6wf1rVfWro5SdwNVxthaIT7vd57uoAJ3Hu3352dfhpF4PfjTYXbh+h9n+3N7NJF4PpSr6cRGpCvzFcPHu94rICb5ivgI8pKqbROQVInKSH3bYi7vw0iJSISLvFJFGVU3gOoTSo5Q7C7hUROIi8lbcRXG3qrbhOkG/ISINIhIRkSNE5NVjOKbvAV8WkcUAItIqIudlpfmsiNSIyLG4zpzb/fpbgc/4fVpwHYmZoV/X+Xp5nbdrvoi8ZAx2jcTdwFEi8g4RiYnI23E31P8aQx47gcNzpKnH3VA7cML3lfEYWyTeiHs6Wwac4P+OAf4AvNt7nNcDV4vIPP8U80rfXn8EnCYib/P1OVNETvD5rgfe5M/9kbiO7tGox8WP24GYiHwOaAhs/yHwRRFZ6keEvDTz3oWqbsN1yt8M3KWq/TnK+qSINIvIQly/VKZNfg/4tG+riEijv2by5RbgHBF5va+nKhF5jYgsCKR5l4gsE5EaXD/Onaqawjlu/+jbfBzXpzGI6+t6GNeP91URqfX5/v0Y7BqWQLlfFpF6fx1/nAPXYT7sBJbkGKFTgesDageSInIWcMY4zR6WUhX9u3Gx0czflar6P7hY3l24k3oEcL5P3wD8ABfX3IwTkH/z2y4ENvnQwYeAd45S7kPAUtxd/cvAW/TAePV3407I076cOzn4UTQX38KNlLhXRHpwHbMnZaX5Pa5D6rfA11X1Xr/+S8A6XA/9E8Cjfh2q+jDuBvFNnLf3ew710MeMP+434C6oDpxn+YZAWCMf/hV3s9orIpePkOYm3DnbjqvbNeO3uuC8B/hPVd2iqjsyf7jQ4ju9c3I57hytxYU8vobrZ9qCe0z/hF+/HtfBCu7cDeFE4UbcDWI07gF+jevH2oxzcoIhgqtxAnUvztG5DteRn+FG3GCBXKEdcB2Xj3h7f+XzQlV/5o/tNn9tPQmclUd++P234p7w/gUncFtxT+dBTboZ16ezAxc2u9Tv+ywuNPXvuGv1HOAcVR3y4nwObhTMFtxor7fna1cOPoJzKp/HjQL6Me4mny8/8f87ROTR4RKoag/uOO/A6cw7cLoxacjBIezpi4iswnXarAyh7CW44WPxUWKRhjEpiMipOA91sY4iACKiwFJV3VA04w6UfT+u0/OHxS57qlOqnr5hGAXAh0MuA344muAbUxcTfcOYJojIMbiXmebi3ocwpiEW3jEMw5hGmKdvGIYxjTDRNwzDmEaENvtdS0uLLlmyJKzijSnOI488sltVW8Mo29q2UUgm2rZDE/0lS5awbt26sIo3pjgikj2FxHBprse9i7BLVY8bZrvg3q/ITOy2SlWHHV8dxNq2UUjyadujYeEdYzpzA27K25E4C/ey3lLctNf/UQSbDKOglNzHDf6ypZNUWlmxZEbYphhTHFV9wL8YNxLnATf58exrRKRJROb6aTkKQjKV5k8bO0illePmN9JUE2ftpj0MJFKFKtIYJxXRKCceNoOK2AHfeVfPAE9u7xp1v1gkwoolzaTSyiObO0kPM4JyRm0lJyxsmnSboQRF/yt3P0NEhNsvHm0yTMMoCvM5eHqDbX7dIaIvIhfhngZYtGjRuAv8wR9e4Gu//isAsYjQWB2no3do3PkZheUVS5p5/bFzeOBvuxkYSvHolk6S6dzD4OsrY6RU6Rsa/mb+D0e38p/vPXGyzQVKUPRb6yt5dkdP2GYYxphQ1WuBawFWrFgxrpdfegeTXPvARlYe2cJlpy3lnid30NY9wDkvncfcxqpJtdeYOH/d0c3nVz/F2k2dHDW7joaqOO9beRhnLJtNPDpy5LyrP8Hqx14kHhXe8NJ51FUeKsP1VYWT5pIT/Za6Sv7YM5Y5vQyjYGzn4Ol9F3DwVLqTgqpyy0NbWLOxg86+BJ844yhetqiZV1iIs6Q5fmETpyxtpXcwydLZ9WPa99SjQhlYBpSo6HcPJBlMpqiMRcM2x5jerAYuEZHbcDOidhUinv/czn189udPAnDWcXN42aLmyS7CKBDzmqpzJyoxSk70W+srAejYN1SWFWqUDyJyK+7zcy0isg34PO7rSKjq93BTfJ+Nm+66DzeF9aSzo3sAgNsuOpmTDjPv3igsJSf6LXVO9Nt7Bk30jYKiqhfk2K7AhwttR3vPIABzG6s4+MNthjH5lNw4/Yynv3vfYMiWGEZx2NXjPP1M2zeMQlJyot9SVwEc8H4MY6rT3jNIXWWMmoqSe/A2piAlKPrm6RvTi109g+blG0Wj5ES/Kh6lvirG7n32QooxPWjvNtE3ikfJiT5Aa12lhXeMaUP7PhN9o3iUpOi31FfSbuEdY5qwq3uAWSb6RpEoSdFvratkt3n6xjSgdzBJ71CKWfU2zYJRHHKKvohcLyK7ROTJEbaLiHxbRDaIyOMisnyiRrWap29MEzJhTAvvGMUiH0//Boo853hLXQU9A0mbTtaY8mScGwvvGMUip+ir6gPAnlGS7J9zXFXXAE0iMnciRmW8HuvMNaY6u7rN0zeKy2TE9Eeac3zczG5w8c2dfk4Sw5iqtPu3cc3TN4pFUTtyReQiEVknIuva29tHTJeZc2f73v5imWYYobCrZ5BoRGiuqQjbFGOaMBmin/ec46p6raquUNUVra0jzyed+WBEW5d5+sbUZkfXALPrK4lEbKI1ozhMhuivBt7tR/GczCTMOV5fFae+KkabefrGFKeta4C5NpusUURyzvAU1pzj8xqr2b7XPH1jatPW1c9x8xvDNsOYRuQU/bDmHJ/XVEVbl3n6xtRFVWnrGuCMY+eEbYoxjSjJN3IB5jZV86KFd4wpzJ7eIQaTafvouVFUSlb05zVW0dmXoH/IXtAypiaZgQpzGy2mbxSP0hV937n1ooV4jClK5kl2XpN5+kbxKFnRz3g/bdaZaxQQETlTRJ71c0ddMcz2RSJyn4j8xc8tdfZklZ3x9OdYeMcoIiUr+vPN0zcKjIhEgWtw80ctAy4QkWVZyT4D3KGqLwPOB747WeW/2NVPPCq01NrbuEbxKFnRn93oLgTrzDUKyInABlV9XlWHgNtwc0kFUaDB/24EXpyswtv2DjCnscpezDKKSsl+ibkyFqW5Jm7fyjUKyXDzRp2UleZK4F4R+QhQC5w2WYW3dfVbJ65RdErW0wdorq1gT699K9cIlQuAG1R1Ae4lxJtF5JDrJt95pYK0dQ0wz+L5RpEpadGfWVtBh30g3Sgc+cwb9X7gDgBV/TNQBbRkZ5TvvFKB9OzsHmCOefpGkSlp0Z9RW0Fnn4m+UTDWAktF5DARqcB11K7OSrMFeB2AiByDE/38XPlR2NuXIJFSm1LZKDolL/oW3jEKhaomgUuAe4BncKN0nhKRL4jIuT7ZJ4APishjwK3AKj/1yITYZZ9JNEKiZDtyIePpJ0in1UY4GAVBVe/GTRoYXPe5wO+ngb+f7HIzX4UzT98oNiXt6TfXVJBKK90DibBNMYxJZZf/YpZ5+kaxKWnRn1nnviZkIR5jqrHf02+w0TtGcSlp0Z/h31Q00TemGrt6BqmOR6mtiIZtijHNKG3RrzFP35ia7OoZZFZDJSLWV2UUl9IWfQvvGFOU9p4B68Q1QqG0RT/j6dtYfWOKsatn0DpxjVAoadGvrohSHY+yx97KNaYY7T2DzKq3Tlyj+JS06IO9oGVMPQYSKXoGkubpG6FQHqJv4R1jCtFub+MaIVIeom+evjGFsBezjDAx0TeMIrPf068z0TeKj4m+YRSZDt+eW0z0jRAoC9HvG0oxkEiFbYphTAp7+9xcUk018ZAtMaYjZSH6gM2rb0wZ9vQOUVMRpSpuUzAYxafkRb/Zv6BlX9AypgqdfUP727VhFJuSF33z9I2pxt6+hIV2jNDIS/RF5EwReVZENojIFcNsXyUi7SKy3v99YLIMzIi+deYaU4U9vebpG+GR88tZIhIFrgFOB7YBa0Vktf+iUJDbVfWSyTbQRN+YauztG2LhjJqwzTCmKfl4+icCG1T1eVUdAm4DziusWQdorI4jAp0m+sYUobMvQbOFd4yQyEf05wNbA8vb/Lps3iwij4vInSKycLiMROQiEVknIuva29vzMjAaEZprbCoGY2qQTKXp6k/QZOEdIyQmqyP3l8ASVX0p8BvgxuESqeq1qrpCVVe0trbmnXlzTZzOXvtOrjG55Oqr8mneJiJPi8hTIvLjiZbZ1e/a8Qzz9I2QyBnTB7YDQc99gV+3H1XtCCz+ELhq4qYdYEZtBR29g5OZpTHNyaevSkSWAp8G/l5VO0Vk1kTL7fQvZjXXmqdvhEM+nv5aYKmIHCYiFcD5wOpgAhGZG1g8F3hm8kx0om+evjHJ5NNX9UHgGlXtBFDVXRMtNDP02MI7RljkFH1VTQKXAPfgxPwOVX1KRL4gIuf6ZJf6x9/HgEuBVZNppE2vbBSAfPqqjgKOEpEHRWSNiJw50UIzAxKsI9cIi3zCO6jq3cDdWes+F/j9adxjcEForqmgs3cIVbUPSRvFJAYsBV6DC2s+ICJ/p6p7sxOKyEXARQCLFi0aMcPMvDs2Tt8Ii5J/Ixecp59MK90DybBNMaYOOfuqcN7/alVNqOoLwHO4m8Ah5DtIIRPesZi+ERZlI/pgY/WNSSVnXxXwc5yXj4i04MI9z0+k0D19Q8SjQm2FTbZmhENZiH7GK+ow0TcmiTz7qu4BOkTkaeA+4JNZI9XGzN7eBM01FRamNEIjr5h+2My0qRiMApBHX5UCH/d/k4LNsGmETVl4+rMbqgDY0dUfsiWGMTH29idotJE7RoiUhei31lVSEYuwrdNE3yhv9g0kqa8siwdsY4pSFqIfiQgLmqrZ2tkXtimGMSF6h5LUmugbIVIWog+wYEaNefpG2dM7aKJvhEv5iH5zNVv3mKdvlDf7BpPUV5noG+FRNqK/sLmGzr4E+wbtBS2jPEmm0gwk0tRWmOgb4VE2or+guRqAbRbXN8qU3sEUALWV9mKWER5lI/qZz8tt22NxfaM82TfknlItvGOESdmIfsbTtxE8RrnS60OT1pFrhEnZiP7M2gqq41EbwWOULT0DJvpG+JSN6IsIC5qr2dzRG7YphjEuMp5+nYm+ESJlI/oAK5bM4MENHTaCxyhLTPSNUqCsRP8tL19AfyLF3U+0hW2KYYyZfSb6RglQVqK/fFETh7fWcue6bWGbYhhjZp915BolQFmJvojwlpcv4OFNe2y8vlF2HBi9Y+P0jfAoK9EHOGPZbAAeeG53yJYYxtjYN5iiIhqhMmaib4RH2Yn+Ea11zGus4oHn2sM2xTDGhJtszQTfCJeyE30R4ZSlrTy4cTfJVDpscwwjb/bZDJtGCVB2og9wylEt9AwkeWxbV9imGEbe7BtM2sgdI3TKUvRXHtlCROBXj9vQTaN86DXRN0qAshT9ppoK3rR8ATev2cTz7fvCNscoY0TkTBF5VkQ2iMgVo6R7s4ioiKwYb1n2ARWjFChL0Qf41JlHUxmLcsVPn6C9ZzBsc4wyRESiwDXAWcAy4AIRWTZMunrgMuChiZTXM5ikzmbYNEKmbEV/Vn0Vnz9nGX/Z0slrv34/a57vCNsko/w4Edigqs+r6hBwG3DeMOm+CHwNGJhIYb2DSersAypGyJSt6AO8dcVCfv3RU5nVUMmHbnmEz//iSd54zYO0ddlMnEZezAe2Bpa3+XX7EZHlwEJV/dVEC+sdTFl4xwidvEQ/V9xTRCpF5Ha//SERWTLZho7EEa11XL/qFQDcvGYzT7d187Hb17N1Tx93PbKNL//qaTbttpk5w2YgkWJv3xCqGrYpeSMiEeBq4BN5pL1IRNaJyLr29kPfIUmnld6hJHU2Tt8ImZxuRyDueTrOE1orIqtV9elAsvcDnap6pIicj3sUfnshDB6OxTNr+eUlKwH488YOPnXX45xy1X37t9/16HbOeelc2roGOH5hE3t6h1CFNy2fTyKVpq4yxhGtdaRU2bS7l8FkmqPn1NPZO0RlLEpDdYy9fQl6h5I0VMdpqIoX69CmBN+9fwNfv+dZ0gpHza7jHScu4uWLZzC7oZKZdZVEIzLsfm1d/Tz8wh7OOm4uFbGCPJRuBxYGlhf4dRnqgeOA+0UEYA6wWkTOVdV1wYxU9VrgWoAVK1YccmfrS6RQxWL6Rujk0wL3xz0BRCQT9wyK/nnAlf73ncB3RES0iG5d5nOKC5qrGUqlUVVOWNhMTWWUD960jtvXbWVOQxX3Pr2TylgEBa5/8IX9+8ciQjJ9wFwRyFhfUxGlb8h93zQicNz8RkSEiqjQUBUnkVZq4lFqK2NEBAaTaSpiEWoqouztS9CfSCFAPBphR/cA1fEoS1pqaKyOk0pDKp0mGomQSqeJRISICP1DKfqGkvQOphhIpJjTWEVTTZxoJIIAiVSaZFoRoCIW4dEtnWzr7GflkS0011SQVkWBRDJNZTxCLBKhqz8BPn004sqIR4XqCmd3PBoh4sSNWERIqZJKKzFvU1qV3sEkO7sHqYhFaKiOIQixqNsn7etPRIhHhUhEeHZHD79Y/yJnLJvN8sXN/GL9i1z5ywNNp6WuklOXttA3lOK5XT0MJtKceNgMAH795A76Eym+M2sDi2fWMpBIceSsOhqqYrTvG+K5nT386AMnURUft/e8FlgqIofhxP584B2ZjaraBbQcaBNyP3B5tuDng301yygV8mmBw8U9TxopjaomRaQLmAkcNEGOiFwEXASwaNGicZo8OiLCu05efNC633781STTSjwaYW/fELWVMbr7E/z2mV0011awt2+Ije29VMejLJpZTSwS4dkdPbTWV9KfSLGze4D5TdU0VMXZ1tnHus2dRCPCUDLNju4BYtEIbXv76R1MklaojEcYTKTpT6RoqolT7UVpKJlmVkMl3QMJfvV4G139CWIRJ8CptBKJQFpBVamKR6mtiFFbGaUiFmXd5k56B5P7b0yxiBNbVXeTOby1liNb6/jpo9sZTKb2e8/xaITBZJpUWvd/mzWRcstV8SiJVJqBRP5vNos4oU6k0nT7m0h6lFt7Q1WMC09ezJXnHks0Ilx86uFs3dPPE9u76Ogd5M8bO/jDht00VLmnrVhE+NPG3UREOG3ZbP7h6Fa+e/9GNnf0Ul0R5c5HtrFvMElTTZyjZ9fT1Z8Yt+j7tnoJcA8QBa5X1adE5AvAOlVdPa6Mh8GmVTZKBcnljIvIW4AzVfUDfvlC4CRVvSSQ5kmfZptf3ujTjDgr2ooVK3TdujE7TFMKVcWHDSa0X3B5uDxVlbQyYhgFnJeeTDvPHiCZThP1Hn4q7Z4aBKiKRw/JJ+PhRyIHbEikXF7VFZMfw8602dHqTkQeUdVxj6mfCMO17XRa6Uu4JyubcM2YCBNt2/m4HbninsE020QkBjQCNoYyB+MR/OH2Cy4Pl6eIEM1RVCQiVBwk5vkLUyRyqD0VsfEdWz6Mt97CJBIR8/KNkiCf3rH9cU8RqcDFPbMfe1cD7/G/3wL8rpjxfMMwDCM/croeecY9rwNuFpENwB7cjcEwDMMoMXLG9AtWsEg7sHmEzS1kdQKHiNkyPKVuy2JVbQ3DGGvbY6ZU7IDysGVCbTs00R8NEVkXVidcNmbL8Jgt46OUbC0VW0rFDpgetpT1NAyGYRjG2DDRNwzDmEaUquhfG7YBAcyW4TFbxkcp2VoqtpSKHTANbCnJmL5hGIZRGErV0zcMwzAKQEmJfr6fritQ2QtF5D4ReVpEnhKRy/z6K0Vku4is939nF8meTSLyhC9znV83Q0R+IyJ/8/+bi2DH0YFjXy8i3SLy0WLVi4hcLyK7/FQfmXXD1oM4vu3bz+N+LvySwNr2QfZY2ybEtq2qJfGHe/FrI3A4UAE8BiwrYvlzgeX+dz3wHO4TelfiZlYsdn1sAlqy1l0FXOF/XwF8LYRztANYXKx6AU4FlgNP5qoH4Gzgv3HTBJ0MPFTs8zZKvVnbPmCPtW0Nr22Xkqef76frCoKqtqnqo/53D/AMWV9RKgHOA270v28E3ljk8l8HbFTVkV48mnRU9QHcW95BRqqH84Cb1LEGaBKRucWxdFSsbefG2raj4G27lEQ/56frioW4L3+9jAMfwr7EP1JdX4zHTo8C94rII+KmpAaYrapt/vcOYHaRbMlwPnBrYDmMeoGR66Fk2lAWJWOXte0RmTZtu5REvyQQkTrgLuCjqtoN/AdwBHAC0AZ8o0imrFTV5cBZwIdF5NTgRnXPfEUbeiVusr1zgZ/4VWHVy0EUux7KGWvbwzPd2nYpiX4+UzgXFBHbjnLRAAABUUlEQVSJ4y6KH6nqTwFUdaeqplQ1DfwA96hecFR1u/+/C/iZL3dn5pHO/99VDFs8ZwGPqupOb1co9eIZqR5Cb0MjELpd1rZHZVq17VIS/XymcC4YIiK42UKfUdWrA+uDcbP/AzyZvW8BbKkVkfrMb+AMX25wCuv3AL8otC0BLiDw+BtGvQQYqR5WA+/2Ix1OBroCj8phYm37QJnWtken8G270D3UY+zNPhs3smAj8P+KXPZK3KPU48B6/3c2cDPwhF+/GphbBFsOx43weAx4KlMXuE9Q/hb4G/A/wIwi1U0t7qM4jYF1RakX3MXYBiRwccz3j1QPuJEN1/j28wSwophtKMdxWNtWa9tZZYfStu2NXMMwjGlEKYV3DMMwjAJjom8YhjGNMNE3DMOYRpjoG4ZhTCNM9A3DMKYRJvqGYRjTCBN9wzCMaYSJvmEYxjTifwF2LMzOQcB3YwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_FT = train_model(resnet50, criterion, optimizer,\n",
    "                       num_epochs=100, trainVal=['train','val'],verbose=False) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.3.d) (4 points)\n",
    "Plot ROC Curve and confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "macro AUC for the model trained on full data is :  0.561220798736\n",
      "Confusion matrix, without normalization\n",
      "[[ 918  765    3    0]\n",
      " [ 935 1002    2    0]\n",
      " [ 251  182    5    0]\n",
      " [  97  121    0    0]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWQAAAEmCAYAAABVi+pHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4wLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvFvnyVgAAIABJREFUeJzsnXm8V9P6x9+f02lOJRUa0KQUSSpRMkSSJnPGiitc48U1XuTimq6Zy0XIcFVmMpaZH0VKJBRCiSSludPp+f2x1qlvdabOtL/n9Ly99uu799prr/35fnWevfaznvUsmRmO4zhO8mQkLcBxHMcJuEF2HMdJE9wgO47jpAlukB3HcdIEN8iO4zhpghtkx3GcNMENslMukVRd0ouSFkl6shjtHCfp9ZLUlhSS9pb0ddI6nKIjj0N2ShNJxwLnAW2AxcAU4Foze7+Y7Z4AnAXsZWariy00zZFkQCszm5m0Fqf08B6yU2pIOg+4DfgXsDWwHfAfYEAJNL898M3mYIwLg6TMpDU4JYCZ+eZbiW9AHWAJcGQ+daoSDPbPcbsNqBrP7QvMBs4H5gFzgaHx3FXAKiAr3uNkYDjwWErbOwAGZMbjIcB3hF7698BxKeXvp1y3F/AxsCh+7pVy7m3gauCD2M7rQP08vluO/gtT9A8E+gDfAAuAS1PqdwE+BBbGuncBVeK5d+N3WRq/79Ep7V8E/AI8mlMWr2kR79ExHjcCfgP2Tfrfhm95b95DdkqLPYFqwLP51LkM6Ap0AHYlGKV/pJzfhmDYGxOM7t2StjSzKwm97tFmVsvMRuQnRFJN4A7gYDPbgmB0p+RSrx7wUqy7FXAL8JKkrVKqHQsMBRoCVYAL8rn1NoTfoDFwBXA/cDywO7A3cLmkZrFuNvA3oD7ht+sJ/BXAzHrEOrvG7zs6pf16hLeFYak3NrNvCcb6MUk1gIeAkWb2dj56nYRxg+yUFlsB8y1/l8JxwD/NbJ6Z/Ubo+Z6Qcj4rns8ys5cJvcPWRdSzBthZUnUzm2tm03Kpcwgww8weNbPVZvYE8BXQL6XOQ2b2jZktB8YQHiZ5kUXwl2cBowjG9nYzWxzv/yXhQYSZTTKzj+J9ZwH/BfYpxHe60sxWRj3rYWb3AzOBCcC2hAegk8a4QXZKi9+B+gX4NhsBP6Qc/xDL1raxgUFfBtTaVCFmtpTwmn8aMFfSS5LaFEJPjqbGKce/bIKe380sO+7nGMxfU84vz7le0o6Sxkr6RdKfhDeA+vm0DfCbma0ooM79wM7AnWa2soC6TsK4QXZKiw+BlQS/aV78THjdzmG7WFYUlgI1Uo63ST1pZq+Z2YGEnuJXBENVkJ4cTXOKqGlTuIegq5WZ1QYuBVTANfmGSEmqRfDLjwCGR5eMk8a4QXZKBTNbRPCb3i1poKQakipLOljSjbHaE8A/JDWQVD/Wf6yIt5wC9JC0naQ6wCU5JyRtLWlA9CWvJLg+1uTSxsvAjpKOlZQp6WigLTC2iJo2hS2AP4Elsfd++gbnfwWab2KbtwOfmNlfCL7xe4ut0ilV3CA7pYaZ3UyIQf4HYYT/J+BM4LlY5RrgE2Aq8DnwaSwryr3GAaNjW5NY34hmRB0/EyIP9mFjg4eZ/Q70JUR2/E6IkOhrZvOLomkTuYAwYLiY0HsfvcH54cBISQslHVVQY5IGAL1Z9z3PAzpKOq7EFDsljk8McRzHSRO8h+w4jpMmuEF2HMdJE9wgO47jpCDpQUnzJH2RUlZP0jhJM+LnlrFcku6QNFPSVEkdU64ZHOvPkDS4MPd2g+w4jrM+DxMGRFO5GHjDzFoBb8RjgIOBVnEbRghfzJn1eSWwB2EG6pU5Rjw/PCFJBUFVapqqlZ8w08rVqyUtYZPYuUmdpCVUaD79dNJ8M2tQ3HYq1d7ebPVGkxbXw5b/9pqZbWhw1503e1fSDhsUDyDkCgEYSchrclEsf8RCdMRHkupK2jbWHWdmCwAkjSMY+Sfy0+YGuYKgavWo2vWcpGUUmoZtijoDOhk+uKVfwZWcIlO9sjacIVkkbPVyqrbOPypwxZS720j6JKXoPjO7r4CmtzazuXH/F0L2QgizOH9KqTc7luVVni9ukB3HqThIkFGpoFrzzaxTUW9hZhbzU5c47kN2HKdioYz8t6Lxa3RFED/nxfI5QNOUek1iWV7l+eIG2XGcCkTsIee3FY0XgJxIicHA8ynlJ8Zoi67AoujaeA3oJWnLOJjXK5bli7ssHMepWKignEwFXa4nCINy9SXNJkRLXA+MkXQyIQNgjqP6ZcKiAzMJ2f+GApjZAklXExY5gJBGdkFB93aD7DhOxaFwPuR8MbNj8jjVM5e6BpyRRzsPAg9uyr3dIDuOU7Eoup84cdwgO45TgSh+DzlJ3CA7jlNxEMX2ISeJG2THcSoW7rJwHMdJBwSV3GXhOI6TPMJ7yI7jOOmBD+o5juOkDz6o55RXzhjYkaF92iPgoVemctezn3LY3jty2Ql70Wa7rdj7rMf4dMavAGRWyuCe8w6iQ8uGZFbK4PHx0/j3qIllprV5w5rcNWT3tcfb1a/BLS9/zYNvf8+QHjtwwt7NWLPGeHPar1z3wnSa1KvOG5fux7fzlgAwedYfXDbm8zLTmxcrVqzggP16sGrlSlZnr+bQw47g8iuvSlpWvrz+2qtccN45ZGdnM+Skv/D3Cy8u+KIkKIGJIUniBnkzpu0O9Rnapz17n/UYq7KyeeFfR/DyhO+YNms+g/75PHed02u9+of32JGqlSvR+dSRVK+ayeT7hzLmra/48dc/y0Tvd/OW0ufGdwHIEEy4+kBe++wX9my1FQfusg0H3/AOq1avYataVdZe88P8ddekC1WrVuXVcW9Sq1YtsrKy2H+f7vQ66GD26No1aWm5kp2dzblnn8FLr4yjcZMmdO/amb59+7NT27ZJS8udcuxDLr/KnWLTpmk9Pv5qLstXriZ7jfHe5z8xsFsrvv5pATNm/7FRfTOoUa0ylTJE9SqZrFqdzeJlqxJQDt1aN+DH+cuY88dyju++A/8ZN5NVq9cA8PuSZDQVFknUqlULgKysLFZnZaE0fs3+eOJEWrRoSbPmzalSpQpHHj2IsS8+X/CFiVBqyYXKBDfImzHTZs2n286NqbdFNapXzaR35+Y0abBFnvWfee8blq3I4vtRp/PN46dy21Of8MfiFWWoeB39OzbihUkhm2GzBjXp0qIez53XndFn70X77dat7tF0qxq8fGEPRp+9F52bp8+KKtnZ2eyxewe2a9SQ/Q84kC577JG0pDz5+ec5NGmyLpNk48ZNmDOnwEySySHlv6UxaW+QJfWW9HVcRLBAx5WkIZLWSGqfUvZFLkuylLTOLpLejVonS3pAUo2o567SvHdR+fqnBdw8ZiIvXn8EL/zrcD77dh7Za/LOu9259TZkr1lD82PuZacT7+ecwzuxwzZlv7RR5UrigJ234aUpPwOQmSHq1qjCwFve51/Pfcl/hobc4/P+XMmeV46nz43vcvWz07hjcEdqVUsPL12lSpWYMGkKM2fN5pOPJzLtiy8KvsgpGAkyMvPf0pi0NsiSKgF3ExYSbAscI6kwjqvZwGWlqS0VSVsDTwIXmVlrM9sNeBXIu7uZJox89Qu6nfEYB54/moVLVjBjzsauihyO2n8nXv94Fquz1/DbwmV8OG0Ou++4TRmqDezbtiFfzF7E/MXBNTF30QpenRpW1/nsx4WsMaNerSqsWr2GhcuyAPjip0X8MH8pzRrULHO9+VG3bl322Xc/Xn/91aSl5EmjRo2ZPXvdakRz5symceMCVyNKDu8hlxpdgJlm9p2ZrQJGERYVLIixQDtJGy3cJukYSZ/HXvMNKeVLJF0r6TNJH0Uji6QGkp6W9HHcuuVyvzOAkWb2YU6BmT1lZr9ucO9+kibEHvT4lHvsI2lK3CZL2kLStrHHPSVq3bsQ33uTaVC3BgBNG2zBgO6tGP3m9Dzrzp63mH07bAcEX3KXnRrx9U+/l4asfOnfsfFadwXA61N/Yc9W9YHgvqhcKYMFS1ZRr1YVMuLfX9OtatCsQU1+/H1ZmevdkN9++42FCxcCsHz5ct4YP47WrdskrCpvOnXuzMyZM5j1/fesWrWKJ0eP4pC+/ZOWlTfl2Iec3v333BcK3ANA0j+BT8zshVyuWwPcCFzKuiz/SGoE3ADsDvwBvC5poJk9B9QEPjKzyyTdCJwCXAPcDtxqZu9L2o6Q9X+nDe63M2El2oJ4H+ga1+T6C3AhcD5wAXCGmX0gqRawgrCk+Gtmdm18U6ixYWOShsV6UK1uIW6/MU9c3p96tauTtTqbc+98g0VLV9K/W0tu+WtP6tepzjPXHMbUb+fR/9KnufeFydx3QW8m3TcESTz6+hd88f38It23qFSvUom92zTg0tFT15aN+ehHbjq2A69fvA9Z2cb5j00GYI8WW3Fen9ZkZa/BDC4d8zmLYo85SX6ZO5dTThpMdnY2a2wNhx9xFH0O6Zu0rDzJzMzk1tvvot8hB5Gdnc3gISfRtl27pGXljlSuoyzS3SDniZldUUCV/wGXSWqWUtYZeNvMfgOQ9DjQA3gOWEXoWQNMAg6M+wcAbVNGwWtLqmVmS4oguwkwOq7JVQX4PpZ/ANwS9TxjZrMlfQw8KKky8JyZTdmwsbhS7n0AGbWbFmnRxQPOH7VR2QsfzOSFD2ZuVL50RRbHXfNiUW5TYixflU2HS9ZfCScr2zj30ckb1X3ls7m88tncjcqTZpf27fnok431pjO9D+5D74P7JC2jcKS5WyI/0v1RUqSFAgHMbDVwM3BRIe+VFbP/A2Sz7mGVQejVdohb41yM8TRCr7sg7gTuMrNdgFOBalHr9cBfgOrAB5LamNm7hIfFHOBhSScW8ns4zmaLgIyMjHy3dCa91YX1qFpJaiapCjCIsKhgYXmY0MNtEI8nAvtIqh/dAMcA7xTQxuvAWTkHkjrkUucuYLCkPVLqHZbjI06hDuseKKmulBZm9rmZ3UD4zm0kbQ/8amb3Aw8AHQvQ6TiOCrGlMWltkGMv90yC33Y6MMbMpkHwIUvKd2QhDgTeATSMx3OBi4G3gM+ASWZWUIT72UAnSVMlfQmclst9fiU8LP4dw96mAwcBizeoOhx4UtIkINX5em4cuJsKZAGvEBZZ/EzSZOBogi/bcZx8UbnuIWvdW7pTnsmo3dSqdj0naRmFpmGbjQJg0pqvb+mXtIQKTfXKmmRmnYrbTqV6zaxmr/zzgiwePbhE7lUalNtBPcdxnI0QKCPN/RL54AbZcZwKg1Ba5wUpCDfIjuNUKNLdT5wfbpAdx6lQeA/ZcRwnHSjnPuTy27d3HMfZgBwfcn5bgW1If5M0LYaiPiGpWpwLMSFmnRwd50UgqWo8nhnP71Ac/W6QHcepUBTHIEtqTJx7YGY7A5UIcwxuIOS0aUnIg3NyvORk4I9YfmusV2TcIDuOU3GILov8tkKQCVSXlElI6jUX2B94Kp4fCQyM+wNYl1jsKaCniuHEdoPsOE6FohA95PqSPknZhuVca2ZzgH8DPxIM8SJCsrGFceYwhKyTOQmh12akjOcXAVsVVbsP6jmOU2FQnDpdAPPzmqknaUtCr7cZsJCw8ETvEhWZD95DdhynYlG85EIHAN+b2W9mlgU8A3QD6kYXBqyfdXJtRsp4vg5Q5FUb3CA7jlNxULHTb/4IdFVYD1NAT+BLQkKyI2KdwUBOUrIXWJe58QjgTStGgiB3WTiOU6EozsQQM5sg6SngU2A1MJmwCMRLwChJ18SyEfGSEcCjkmYCCwgRGUXGDbLjOBWK4k4MMbMrgSs3KP6OsMbnhnVXAEcW64YpuEF2HKfCUNjJH+mKG2THcSoUnlzISQPWwOpVSYsoNPPee63gSmmFJ6gvN5TfDrIbZMdxKhbusnAcx0kDJMgox9ne3CA7jlOB8EE9x3GctMF7yI7jOOmAgtuivOIG2XGcCoOASpXKr0V2g+w4ToXCfciO4zhpgEdZOI7jpA0eZeE4jpM2eA/ZcRwnHfAoC8dxnPRA+KCe4zhO2uAuC8dxnDShHHeQ3SBv7pxxWGeGHtIBCR56aQp3Pf0xVwztQd+9dmSNGb8tXMqwG8Yy9/cl7L3rdjx59RHM+mURAM+/9zXXPfp+qeq798rjOLjHzvy2YDGdjvwXAFvWrsGjN5zE9o3q8cPPCzj+whEsXLwcgJsvPIKDurVj2YpVDLvyUaZ8NZv2OzbmjssGsUXNamRnr+HGEa/x1Ouflqru/Pjpp5/4y9ATmTfvVyRx0snDOPPscxLTUxhef+1VLjjvHLKzsxly0l/4+4UXJy0pV8p72Fv5zeTsFJu2OzRg6CEd2PuvD9HlLw9wcNeWNG+0JbeO/ogupzxA12EjeOXDmVxyQve113zw+U90HTaCrsNGlLoxBnj0xY8YcMbd65VdMPRA3p74NbsM+CdvT/yaC4b2AuCg7m1psV0Ddh5wFWde8wR3XBqWN1u2IouTL3+E3Y+4lgFn/ocbLzicOrWql7r2vMjMzOT6G29m8tQveef9j/jvvXcz/csvE9NTENnZ2Zx79hk8/+IrTJ76JU+OeiKN9WrtqiF5bemMG+TNmDbbb8XH0+ewfOVqstcY7332IwP3bs3iZesS3deoVpkiL6FbAnzw6bcsWLRsvbK++7bnsRcnAPDYixPot1/7UL5Pe/43diIAEz+fRZ0tqrNN/drM/HEe3/74GwBzf1vEb38spn69WmX4LdZn2223ZbeOHQHYYostaNNmJ37+eU4BVyXHxxMn0qJFS5o1b06VKlU48uhBjH3x+YIvTIiMDOW7pTNukDdjpn3/G912aUq92tWpXjWT3nu0oEnD2gAMP2kfZow6k0EH7MzVD7279po92jZmwv0n89x1R7PTDvUT0d1wqy34Zf6fAPwy/08abrUFAI0a1mX2L3+srTfn14U0alh3vWs7tdueKpmZfPfT/LITnA8/zJrFlCmT6dxlj6Sl5MnPP8+hSZOma48bN27CnDlp+gCJYW/5belMuTLIkh6UNE/SF4Wsv6+ksYWod7ak6ZIel9Rf0sWxfLikC+L+EEmNiqB5oKS2Kcf/lHTAprZTGnz94+/cPOojXrxxEC/cMIjPvp1H9po1AAx/8B1aDbqLUeO/4LSBuwMwZcYvtD7mbvY4ZQT3PPcJY/55RJLy12KF7MJvU782I645kVOHP4YV9qJSZMmSJRxz1OHcdPNt1K5dO2k5FQIR1tTLb0tn0lvdxjwM9C6Fdv8KHGhmx5nZC2Z2fS51hgC5GmRJlfJpeyCw1iCb2RVmNr44YkuSka98RrfTHuLAcx9j4eLlzPhpwXrnR78xjYE92gCweNkqlq7IAuC1Cd9SOTODrWqXvS923u+L2aZ+MGDb1K/NbwsWA/DzvIU02WbLtfUab12Xn+ctBGCLmtV45o7TGX73i0z8fFaZa96QrKwsjjnqcI4+5jgGHnpY0nLypVGjxsye/dPa4zlzZtO4ceMEFeVPhewhS6qd31aWInMws3eBBQVWzIXY231Q0tuSvpN0diy/F2gOvCLpb7EnfNcG1x4BdAIelzRFUnVJsyTdIOlT4EhJp0j6WNJnkp6WVEPSXkB/4KZ4XQtJD8f2kNRT0mRJn0dtVWP5LElXSfo0nmtT1N+sIBrUrQFA04a1GbB3G0a/MY0WjdcZtb7dduSbH38HYOsta64t79RmWzIkfv9zeWlJy5OX3vmc4/uFV/zj++3B2Lenri0/tm8XALrssgN/LlnOL/P/pHJmJUbffAr/GzuBZ8dPKXO9G2JmnHbKybRusxPn/O28pOUUSKfOnZk5cwazvv+eVatW8eToURzSt3/SsnJH5duHnF/Y2zTAWH8N15xjA7YrRV2bhKTTAMzs3gKqtgH2A7YAvpZ0j5mdJqk3sJ+ZzZc0ZMOLzOwpSWcCF5jZJ/GeAL+bWcd4vJWZ3R/3rwFONrM7Jb0AjDWzp1KuQ1I1Qo+/p5l9I+kR4HTgtnjb+WbWUdJfgQuAv+TyvYcBwwCoWqfgHyoXnhh+OPVqVycrO5tzb3+NRUtXcu/fD6FV061Ys8b4cd4izr71FQAO3acNp/TvyOrsNaxYuZoTr3muSPfcFEZeN4S9d29F/bq1mPnq1Vx978v8+6FxPHbDSQweuCc/zl3A8Rc+CMCr70/joO7tmPbClSxbkcWpwx8D4PBeHenesSX16tbk+P5dARh2xaNM/SYZP+j/ffAB/3v8UXbeeRf22L0DAFdd8y96H9wnET0FkZmZya2330W/Qw4iOzubwUNOom27dknLyhVV1ORCZtY0r3PpRiEMcQ4vmdlKYKWkecDWwOxi3Hp0yv7O0RDXBWoBBa1z3xr43sy+iccjgTNYZ5CfiZ+TgFzfac3sPuA+gIzajYvkFD3g3Ec3Kjtm+DO51IR7n5vEvc9NKsptiszgSx7OtbzPaXfmWv6368dsVDbq5Y8Z9fLHJSmrWHTr3p3lWcn7sDeF3gf3SdsHxoaUY3tcOB+ypEGSLo37TSTtXrqySo2VKfvZFH9izNKU/YeBM81sF+AqoFox287RWhI6HWezoVKG8t0Kg6S6kp6S9FUc8N9TUj1J4yTNiJ9bxrqSdIekmZKmSupYVO0FGuToT90POCEWLQMK2yOtSCwmuDryYgtgrqTKwHGFuO5rYAdJLePxCcA7JSHUcTZXwsBdiUwMuR141czaALsC04GLgTfMrBXwRjwGOBhoFbdhwD1F1V+YHvJeZnYqsALAzBYAVYp6w+Ig6QngQ6C1pNmSTo7lp+X4kUuRh4F7cwb1cjl/OTAB+AD4KqV8FPD3OHjXIqfQzFYAQ4EnJX0OrGHzfNA5TolS3B6ypDpAD2AEgJmtMrOFwACCa5H4OTDuDwAescBHQF1J2xZFe2FehbMkZRAG8pC0FcF4lDlmdkwe5bkaMjN7G3g77g/f4NzOKfs7pOw/TDC+611jZk8DT6c0sUPKPmZ2D7k8Gc3sA1LC3gjhcznn3gB2y+WaVD2fAPtuWMdxnNwpRCe4vqRPUo7vi+MxOTQDfgMekrQrYRznHGBrM5sb6/xCGIMCaAz8lHL97Fg2l02kMAb5boIhaiDpKuAogo/UcRwnrRBQqWCLPN/MOuVzPhPoCJxlZhMk3c469wQAZmaSSnxktkCDbGaPSJoE5MwuO9LMCjVTznEcp0wpmQRCs4HZZjYhHj9FMMi/StrWzOZGl8S8eH4OkBqV1iSWbTKFnalXCcgCVm3CNY7jOGWKKL4P2cx+AX6S1DoW9QS+BF4ABseywUBOhqUXgBNjtEVXYFGKa2OTKLCHLOky4FjgWcL3/Z+kx83suqLc0HEcpzQpoTjkswgzc6sA3xEG4DOAMTGY4AeC+xbgZaAPMJMQhTa0qDctjA/5RGA3M1sGIOlaYDLgBtlxnLSipBLUm9kUQrqEDemZS10jTOoqNoUxyHM3qJdJEUYPHcdxyoKMcjxVL0+DLOlWQqjbAmCapNficS8gfeahOo7jpFAhDTKQE0kxDXgppfyj0pPjOI5TdASkeUK3fMkvudCIshTiOI5TbMrBunn5UZgoixbAtYTZZmsT5pjZjqWoy3Ecp0ike87j/ChMTPHDwEOEt4GDgTGsn3bScRwnLchxWeS3pTOFMcg1zOw1ADP71sz+QTDMjuM4aUeGlO+WzhQm7G1lTC70bcyoNof801A6juMkglRxoyxy+BtQEzib4EuuA5xUmqIcx3GKSnn2IRcmuVBOgo3FrEtS7ziOk5aU4w5yvhNDniXmQM4NM0vvtcsdx9nskAq/TFM6kl8P+a4yU+EUmx23b8iD/z09aRmFpnb1yklL2CRCuoLyQ3mOxS0u5fm75zcx5I2yFOI4jlNcCpmgPm3x1Ywdx6lQlGOPhRtkx3EqFpuFQZZU1cxWlqYYx3Gc4iBRrgf1CpypJ6lLXKZ+RjzeVdKdpa7McRynCEj5b+lMYaZO3wH0BX4HMLPPgP1KU5TjOE5REJAp5bulM4VxWWSY2Q8bhJJkl5Iex3GcYpHmNjdfCmOQf5LUBTBJlQiL/31TurIcx3E2nYo8MSSH0wlui+2AX4HxscxxHCftKMf2uFC5LOYBg8pAi+M4TrEQ5TvKojArhtxPLjktzGxYqShyHMcpKuUgCX1+FMZlMT5lvxpwKPBT6chxHMcpOhV+6rSZrbdck6RHgfdLTZHjOE4xqOg95A1pBmxd0kIcx3FKgvKc7a0wM/X+kLQgbguBccAlpS/NKW1+nTubM0/oz3EHd+W4PnsyZuS9AIy443oGdG/H4P49GNy/B//39jgAFv2xgDNP6M8BHZpy81UXlrney88/nX06NOPQnl3Wln01bSrH9d+PIw7ai6P79ODzyZ8AMPbZ0Rx2YFcOPWAPjh/Yk6+//LzM9eZHm1bN6Lxbe/botBvdunZOWk6BvP7aq7Rv15p2bVpy043XJy0nT8LU6fy3wrWjSpImSxobj5tJmiBppqTRkqrE8qrxeGY8v0Nx9OfbQ1Z41OxKWEcPYI2Vt8SwTp5UqpTJWRdfTet2u7J0yWJOPmx/OnfbF4Cjh57GsSeftV79KlWrcso5l/LdjOl89830Mtc74MjjOGbIqVx27rrx5FuuvZzT/nYJe+/Xi3fffI1b/nU5Dz35Ck2abs9DT75Cnbpb8t5br3PVRWfzvxffKnPN+fHKuDepX79+0jIKJDs7m3PPPoOXXhlH4yZN6N61M3379mentm2TlpYrJbSm3jnAdKB2PL4BuNXMRkm6FzgZuCd+/mFmLSUNivWOLupN831eROP7spllx82NcQWifsNtaN1uVwBq1tqC7VvsyG+/zs2zfvUaNdm1U1eqVK1aVhLXo1PX7tSpu+V6ZZJYungxAEv+/JMGW28LQIdOXdfWbb9bZ36dOwenaHw8cSItWrSkWfPmVKlShSOPHsTYF59PWlauhLC34vWQJTUBDgEeiMcC9geeilVGAgPj/oB4TDzfU8XwmRSmAz9F0m5FvYFTPpg7+0dmfDmVdrvuDsDTjz3Aif26869LzuTPRQsTVpc3Fw2/npuv/QcHdGnDzddcxrkXD9+ozrOjHqH7fgeWvbh8kES/Pgex1x6dGPHAfUnLyZeff55DkyZN1x4yiSFMAAAgAElEQVQ3btyEOXPS9QEnMgrYgPqSPknZNgzhvQ24EFgTj7cCFprZ6ng8G2gc9xsTo87i+UWxfpHI0yBLynFn7AZ8LOlrSZ9Gv8qn+TUqqamktyR9KWmapHMKEiJpX0mLJE2RNF3SlZv2VUofSedKqpFyvCRJPSXFsqVLuOyswZx96b+oWas2hx57EmPGf8rDz7/LVg224a7r/5G0xDwZ/egILrzyesZP/Iq/X3k9V/z9jPXOT/y/d3lm9CP87dJ/JqQwd8a/9R4fTpzEcy++zH33/If333s3aUkVgkL6kOebWaeU7b5116svMM/MJiWhP78e8sT42R9oDfQBjgSOiJ/5sRo438zaAl2BMyQVxuH0npl1ADoBx0vqWIhrypJzgRoF1ioEKQ+8RFmdlcVlZw2mV78j2PegfgDUq9+QSpUqkZGRQf+jTuTLqfk+fxPlhaf+xwEH9wfgoL6H8sWUdX9HX0//giv/fiZ3jBhF3S2L3GkpFRo3Dh2shg0b0m/AQD75eGIBVyRHo0aNmT173dSDOXNmr9WfjmRI+W4F0A3oL2kWMIrgqrgdqJvyN9uEdeNqc4CmsPZvug4xM2aRtOdzTgBm9m1uW36NmtlcM/s07i8mOMcL/X/QzJYCk4CWkoZIekbSq5JmSLpxrUCpl6QPY8/9SUm1YvksSfXjfidJb8f94ZJGSnpP0g+SDpN0o6TPY/uVY72e8U3gc0kPxpHUs4FGwFuS3krRcK2kzyR9JGnrWLaDpDclTZX0hqTtYvnDku6VNAG4Meaa/jDe6/8ktY71/ibpwbi/i6QvUnvmJYWZcd2lZ7N9ix0ZdNK6nuX8eb+s3X9n3Fiat9qppG9dYjTYehs++SiExU/44B22a9YCgLlzfuJvpxzHdbffxw7NWyUpcSOWLl3K4uj3Xrp0KW+MH0fbdjsnrCpvOnXuzMyZM5j1/fesWrWKJ0eP4pC+/ZOWlSs5U6fz2/LDzC4xsyZmtgMhZcSbZnYc8BahMwowGMhxor8Qj4nn3yzOWFt+vbQGks7LR/gthblBDAPZDZgQj0+L19+bzzVbEXrWVwOdgQ6xjZXA1zFB/nLgH8ABZrZU0kXAeUBB76YtCPmc2wIfAoeb2YWSngUOkfQq8DDQ08y+kfQIcLqZ3RZ/j/3MbH5sqybwkZldFh8UpwDXAHcCI81spKSTCMmZcgYBmgB7mVm2pNrA3ma2WtIBwL+AwwlP5LclHQpcBpxqZsty+Z2GAcMAtm7UpICvvTFTJ03g1edH06J1Wwb37wHAqeddzvixTzPjq8+RxDaNt+PCf677X334fiEiY3VWFu+Nf4lbH3qaZi3bbPK9i8KFZwzl44/eY+GC3+nZuTVnnH8pw2+4k+uHX0T26tVUrVqNK6+/A4B7b7uehQsXcM1l4Z9wpUqZjH45PdwC8379lUFHHgbA6tWrOWrQMfQ6qHfCqvImMzOTW2+/i36HHER2djaDh5xE23btkpaVJ6UUhnwRMErSNcBkYEQsHwE8KmkmsIBi5v1RXsZc0lxCWEeuX8/Mriqw8dBjfQe41syeKaDuvoSnzncEZ/r9ZnavpCFANzM7JdZ7BbgWqEswnLNjE1WAD83s5Pi60cnM5kvqBPzbzPaVNBzIMrNrJWUQjHo1MzNJ/yT8oG8Bd5pZj3i/nsAZZnZYarvx3MqU648GDjSzv0iaD2xrZlmx1z3XzOpLehh4y8xGxuubEox1K0K+kMpm1iaeaw5MBf5rZucX9Fu32WU3e/CZNwuqljbUrl45aQmbRIutayYtYZMoxkB/IlSvrElm1qm47TRr296GP/JSvnWGdN6uRO5VGuTXQ55rZkUeCYmG6Gng8YKMcQrvmVnfXMpT1/LLJugWMM7Mjsml/mrWuWOq5daWma2RlJXyerGGTZ+5mHp9jq6CWJqyfzXBQB8a3yTeTjnXClhCcJM4jlNIytejaH0K9CEXBYXH8whgemFdG0XgI6CbpJbxnjUl7RjPzQJ2j/uHb2K7XwM75LQLnEDo5QMsBrYoRBv/x7pXl+OA9/KoV4d1gwNDcgol1SH0nHsAW0k6YuNLHcfZEFHsQb1Eyc8g9yxGu90Ihmz/GMY2RVIfCD7kHD9ycTCz3whG7AlJUwn+4Bxn5lXA7ZI+YROXmzKzFcBQ4EmFxV3XADn+7vuAV1MH9fLgLGBo1HUCYdZPbtwIXCdpMuv3rm8F7jazbwgzga6X1HBTvofjbK5kKP8tncnTh+yUL9yHXLq4D7l0KSkfcvO2u9q1j7+cb51jOzYplz5kx3GcckWFz4fsOI5Tnii/5tgNsuM4FQjJe8iO4zhpQ3nzn6fiBtlxnApFukdS5IcbZMdxKgyCnBSb5RI3yI7jVCDSf/JHfrhBdhynQlGO7bEbZMdxKg4eZeE4jpNGlGN77AbZcZyKhXxQz3EcJ3l86rTjOE4aUY7tsRtkx3EqDt5DdhzHSRvkPmTHcZy0oBwkoc8PN8gVhKqZGWxfv0bSMgpNVnb5WhihPCes2ZzIWcKpvOIG2XGcCkU5tsdukB3HqVi4D9lxHCdNcB+y4zhOuuAG2XEcJ3lE+XZZZCQtwHEcp8SIYW/5bQU2ITWV9JakLyVNk3ROLK8naZykGfFzy1guSXdImilpqqSORZXvBtlxnIqFCtgKZjVwvpm1BboCZ0hqC1wMvGFmrYA34jHAwUCruA0D7imqdDfIjuNUIMKKIfltBWFmc83s07i/GJgONAYGACNjtZHAwLg/AHjEAh8BdSVtWxT1bpAdx6kwFNQ5jua4vqRPUrZhebYn7QDsBkwAtjazufHUL8DWcb8x8FPKZbNj2Sbjg3qO41QoCjGrcr6ZdSpEO7WAp4FzzezP1HbNzCSV+HRT7yE7jlOhkPLfCteGKhOM8eNm9kws/jXHFRE/58XyOUDTlMubxLJNxg2y4zgVhwKMcWEMskJXeAQw3cxuSTn1AjA47g8Gnk8pPzFGW3QFFqW4NjYJd1k4jlOhKIE45G7ACcDnkqbEskuB64Exkk4GfgCOiudeBvoAM4FlwNCi3tgNsuM4FYaQ7a14bZjZ++QdINczl/oGnFG8uwbcZeGs5YF776LnXh3Zf8/deOCeOwE4/aTj6dWjC716dKHrrjvSq0eXxPRdePapdNppOw7ae/e1Zf8afgk999yV3vt05tTBR/HnooUA/LHgd44ZeBDttq/PFRedm5TkPHn9tVdp36417dq05KYbr09aToGUK73Fj0NODDfIDgBffTmNJx55kLHj3+f19z5m/Osv8/1333LPg4/x+rsTef3difTpdygH9x2QmMbDB53Aw6OeX6+s+z49ee29Sbz6zsc0a9GK/9x+EwBVq1bjvIuv4NKrrktCar5kZ2dz7tln8PyLrzB56pc8OeoJpn/5ZdKy8qS86VUB/6UzbpAdAGZ+8xUddu9M9Ro1yMzMpOtee/PK2OfWnjczXnzuKQYcfnRiGvfYqzt1t6y3XlmP/Q4gMzN43nbbvQu//BwGt2vUrEnnrt2oWrVamessiI8nTqRFi5Y0a96cKlWqcOTRgxj74vMFX5gQ5U1vcadOJ4kbZAeA1ju1Y+JHH/DHgt9ZvmwZb457jZ/nzF57fsKH79Og4dY0b9EyQZX5M+Z/j7BPz4OSllEgP/88hyZN1kVJNW7chDlzihQlVSaUN73usigASdUkTZT0WUzWcVUhrzs4zqT5UtJkSTcXU8fbkjrF/Zcl1S1OeyVNqr6yplXrNvz17PM59vC+HH9kP9rt0p5KGZXWnn/+6TEMOOyofFpIlrtuuYHMzEoMPGJQ0lKcBJEo9tTpJCmrHvJKYH8z2xXoAPSO8Xp5Imln4C7g+JjkoxMhrKRQSMo3gsTM+pjZwsK2tzlwzAlDeeWtD3n6pTeoU7cuzVu2AmD16tW8MvZ5+h16RMIKc+epJx7lzXEvc9s9D5eLte8aNWrM7NnrZtrOmTObxo2LNNO2TChvestxB7lsDHJMurEkHlaOW0HTDi8ErjWzr2Ib2WZ2D4CkfpImxF7zeElbx/Lhkh6V9AHwqKTqkkZJmi7pWaB6TuOSZkmqH/fPk/RF3M6NZTtI+krSw5K+kfS4pAMkfRDT73WJ9WpKejC+AUyWNCCW15A0Jvbun416c3rn98Sef65vC5JOknRbyvEpkm7d5B9+E5n/W5h4NGf2j7wy9nkGHhH8xe+9/SYtWu1Io8ZNSlvCJvPOG6/z37tu4f5Hn6J6jfKxyGunzp2ZOXMGs77/nlWrVvHk6FEc0rd/0rLypHzpFVL+WzpTZnHIkioBk4CWwN1mNiGW/xP4xMxe2OCSnYG8XBTvA13jfPK/EIz3+fFcW6C7mS2XdB6wzMx2ktQe+DQXXbsTArn3IDxAJ0h6B/gjaj0SOAn4GDgW6A70JwSKDwQuA940s5OiC2SipPHA6cAfZtY29vanpNz2MjNbEH+TNyS1N7OpKefHAJdJ+ruZZUV9p+aifRgh3R+NmzTd8PQmM2zwIP5YsIDMypW59sbbqFMneHReeHYMAxMczMvh7GEn8tEH7/HHgvns2b4F5154OffcfhOrVq3khCP6ArBbpy5c++8Qste9Y2uWLF5M1qpVjHvlRR55ciytWu+U5FcAIDMzk1tvv4t+hxxEdnY2g4ecRNt27ZKWlSflTW+a29x8KTODbGbZQIdotJ6VtLOZfWFmVxShuSbA6DifvArwfcq5F8xsedzvAdwR7z9V0lQ2pjvwrJktBZD0DLA3YTrk92b2eSyfRsiFapI+B3aI1/cC+ku6IB5XA7aL7d4e7/3FBvc+KhrTTGBbwkNk7XkzWyLpTaCvpOlA5RwdqZjZfcB9ALvutnuxE5088/KbuZbfevcDxW26RLjjvkc2Kjv6+CF51n//069LUU3x6H1wH3of3CdpGYWmvOgV5dsgl3mURfTbvgX0LqDqNGD3PM7dCdxlZrsQeo6psU1Liy1yHStT9tekHK9h3cNMwOFm1iFu25nZ9LwalNQMuADoaWbtgZdYX38ODwBDCL3jh4r1LRxnM8LjkAtAUoOciAZJ1YEDga8KuOwm4FJJO8brMiSdFs/VYV02pcG5XRx5l+BmyBkkbJ9LnfeAgdHnWxM4NJYVlteAs2JCEiTtFss/IM51V1htYJdYXpvw0FgUfd8H59ZodOk0jfqf2AQ9jrNZU57jkMvKZbEtMDL6TDOAMWY2FvL2IUcXw7nAE5JqEAYBx8bTw4EnJf0BvAk0y+O+9wAPxdf+6QQf9nqY2aeSHgYmxqIHzGyyQmLqwnA1cBswVVIGwX3SF/hP/M5fEh4+0whZoGZImhzLfiIY7rwYA3Qwsz8KqcVxNm82IcVmOqKQF8MpaeLDp7KZrZDUAhgPtDazVZvQxljgVjN7o6C6u+62u7385v8VXXAZk5Vdvv7dbVM3/Wb8VSSqV9akwiSNL4j2u+1uL7/5Yb51mtarWiL3Kg0821vpUQN4SyHRtYC/FtYY50RrAJ8Vxhg7jhMoiWxvSeIGuZSIiyMW6SkcBz53LFlFjrN5UJ5dFm6QHcepUKT75I/8cIPsOE6FovyaYzfIjuNUIHKSC5VX3CA7jlOxKL/22A2y4zgVC4+ycBzHSQvSf3p0frhBdhynwlDekwu5QXYcp0LhBtlxHCdNcJeF4zhOGqBykNEtP9wgO45TsSjHBrnME9Q7juOUJsVddVpSb0lfS5op6eIykLwWN8iO41QoirPqdEybezdh4Yi2wDFxgYkywQ2y4zgVimKuOt0FmGlm38V0uaOAAaUuOuI+5ArC1Cmfzm9Sr9oPpdB0fWB+KbRbWrje0qW09G5fEo1M/nTSazWqqH4B1apJ+iTl+L64YDBAY8JKPjnMJqxIXya4Qa4gmFmD0mhX0ifpurpCbrje0iXd9ZpZQYsnpzXusnAcx1nHHMLiwjk0Yd2CyqWOG2THcZx1fAy0ktRMUhVgEPBCAdeUGO6ycArivoKrpBWut3Qpb3o3CTNbLelM4DWgEvCgmU0rq/v7qtOO4zhpgrssHMdx0gQ3yI7jOGmCG2THqeCoPC/DvJnhBtkpE8qLUZBU4f4mrJwNFElqmLSGpKhw//ic9EOScoyCpC7p/AdnZmsAJHWV1FJSnaQ1lQSSBkoakbSOgpC0LXCVpG3Ky0O8JHGD7JQ6Kcb4HELilqrJKtqY1D9+SacDTwOXAP+VtE1iwkqOucDCpEXkxgaGtyZhYsa2Zmabm1F2g+yUCZL6AScA+5vZT5JaS9opaV05pDw0egHbAXsClwEzgPvKq1GWtL+kI4G6QLt0+h6SMiH89rFnjJnNBF4BrpNUt7y5W4qLxyE7pUKqmyIe7wX0A/4EagCHAF8B/zWzd5JRuY6YdrE28BnwHSH94ipgK+AsYG/gaDP7NTGRhSDnd489y/rAqcCBwBfA6cBzwDigMjDCzJYmpHMb4O/AHcAa4EJgB+CvQHXgSOAJM/tOUkaOK6mi4z1kp8TZwGc8TNKewALCK3MXYDxwGCFHQPUkdaYcZpjZH8A+QB3gTDPLNrN5BDfLG0CVBGQWmg0egpXM7Dczu8bM9jGzM4AngHeALYFuQJK+/N+BZsD5QC3gn4SHxoXA5YSHx+mwzq+/OeA9ZKfUkHQ2MAQ4zsymx7KcHtzhwKWEXufMBGUiaQiwP6Fn/Bowi9CLHGlmN8U65aaXFn/33YGVwKNm9l58A3gcGGtmjyWsr5KZZUuqDDwMLAauMLN5khoBrYELYvV/mNnkhKSWOd5DdkoFSU2Bw4G+wPeS+ks6Bmgu6SDgPGBIEsZ4gwG8ocC5hFf5JcDFQAeCy+ICSWdB+emlRWN8KHANIbfvg5J6m1k28CbQSFKlpAbL4gM5W1J1M8siPLCrEnzGTYG5ZvYWcArwI9AyCZ1J4T1kp0TIxWe8LcE/OJfgh61EGCy7E3gJqBbdAWWtc2egpplNiMeXAp+Z2UuStiD4io8ysyFx0HGVmX1b1joLywbuoUZAf8IqFycDexEGyC4kGLi5hO8zK0mtkg4EDgKmR60rgf8SfPY3ArNivZsIfvC/AGs2hwE+7yE7xWYDo7BHNAzzgbuAmcB1ZjYIGEPw0S5OwhhH9gC+k9QgTgIRcImkKma2GPgUqC+pqZlNL0fGeBhhAO9pYGvCAOpQ4CHgN+B2gqGblYzatdEUvaKWcYS3kZuBFsAwQiTIP1g3rrAauDX68iu8MQZPv+mUIPH1/njCoF1T4NKcCApJgwk9nSOT+OOSVNnMssxshKQdgeHAA4Qe+5bAw5JOI4S71Sb02tKaFGO8F2GQ9HAzWxp7+j8SepeHA+8Dt8U14hIjTrI5FjgC2BZYQYj2uIQwqHc80NbMlgGY2SUJSU0M7yE7RSYldtQkDSCEKu1LMHA7ASMktYgz8w4iDOBNT0DnlsDOcf8QwqvxNwTDsAchiuJPQg/+PEKERVI9+EKjQEtCVEINQg8TgjHOBq6I26NmNjcpjfGztpktIvy+SwkPxE6EwbuDCREV1c3s8yR0pgvuQ3aKRHRLXA28b2YPSeoI/EoYxDsKOI5g6BoSDPVCM1uRkNbWBJ/qdkBnYCczWyXpcqARId71XUlVgcykYnMLw4a++li2NyF87GngdTP7NUZV1CM8LxNZRDXFZ7wHoRd8k5l9EN9QRgM9CP8+/gP83cymJqEznXCXhbPJSKpNCFX6EOgiaamZjYnn2hH+uH6R9B3Bl1w5KWMMYGZfS5oPnEaIPlgTy6+Og3p/jWFtb5PmrooUN8VfCL3+bOBWwmDYWYBJGm9mvxB8x4kRjXFvgm97e+ABSUPMbIKk1wmurTrABW6MA95DdjaJOEJ+CsEIzCLMAusOvGNmj0t6hBA+9jVhqnS/JF6Xc4n6aEjwD/cDJgIvmdmc+HA5mdBL/qWsdRYFSScT/h/cAOxGGBDrSvDb/50Qb/xk0qF6krYnrEc31Mw+lXQlIYrlH8AnQEdC1MeUBGWmFd5DdgpN9L/eQPD7/WJmcyWNjaf3k7QAOJOQA2Jn4OSkjXGMM25MyEnxFLCIYMBWSGpB8HdfkPSA1ybSErjZzJ4Fno29/0fNbG9J9YEJSRvjyO8EX/1qADO7StJI4EHCAOTEJMWlIz6o5xSK+Id+GXCOmb1K6B1jZguAscC7hAkJvczsImCYmX2WhNYUY3wuMJgwyHUmoVc/FbgHaEvorT2YrsY4DtpVyuVUBmHqcw73Ad9KqmZmz5rZz2WjcH1SBvC2lNTAzJYQjHIXSQ1itYeA5YSETbl9t80a7yE7hSUDWEYwvBDidw2CUZY0HqgG9JD0upn9WdYCN+gZ7wQ0J7hUzopaVwFXAVfH6cQ1ckKs0pTKOQ8LSUcRwsS+IbylTJL0ByGOdyDhAVMr1kmElGibs4AsSa8SBhrPBnaStJIQUXEiYbJKE+CHpPSmI26QnUIR8wysIuR8eM3M1kjKtLBseltCyNXjhCiFpI3xIMKg442E2Wr9CRNSDiPEu66WdGG6GuPY02wFjI8RIt2B6whTn6sDIwnRIqMIGdJ2IvhpE4mmyEFSZ4IPuz/BCA81sw6SfiL4unchGON6BP9xWv7+SeIG2SmQaCAygMmEHvAMM/vOzFbHKnsTBpVOSypkLMUY70nIj3CkmS2WtDvwfey9VSNkbbvOQh6FtCR+l28kvULoET9L+H0XAgMIM/AyzeyA+NpfJ7qOkqYqYbJNH0Lc+WGxfLmZPQE8IalHrHOYmSUaBZKOeJSFU2gkNSe8Ls8APiZkRhtEGOQ7wsy+TEBTJQuJc4jGdxzwHzP7RyyrR8i7/D5hIkLvJHQWFkk1Ux9q0Q/+b6BLjFRoQOjtH0+IpHg8Iam5RbJ0JmitBBxjYSGCPoRe81Fm9luMaqmX5BTudMYNsrMReUw+yEmZ2IxgDLoRYmBrEma2fZGAzvrAdtFQHUMIsRpO8Bv3ypltF6fsdgRmmtlPZa2zsEiqQkjQ/i0hMmQ/Mzta0m2EKdC7R9dRTgjfx0kN4OUgqS/h38JcQu6SSwkulhsJU7evAC42s7EqRylMk8INsrMeG/hiaxISty+Obgul/kEp5ExYk4SbIrofahGMgAgGt4uZ/SHpbkIKzUOtHEyBTiXOYptEmFDT1eIKJZL+TRi82zuGGyZu3CRtR4iweRDYlTB9exAhoqUVIaTwMTN7LbeHvLMxbpCdXJF0HmHgqBHBBTA65VyixkBSG8L07FsIfso7gWvN7MaUOncD+wH7pLuvMmWKcUYcLL2TMAX9MjP7X0q9W4FeQHsSTkcZB3JbAlua2cj4gPwP4Y3pBAtT06ub2fKkNJZH3CA7GyHpBIJb4hDCgFKWmR2W/1Vlg8LCmLsQJnf8QBhoXEl4VX4KGJUT5aGQq+LRdPZXbvBG0hmYR5i8sgUhquIWM7sn+mLfIORyTmQAL+XB0ZMQy72MsAzX+Wb2lUIukAcJ+SkOItiX7CS0llfcIDsbIelEwvpmPePWj+AvbmJmPyaoqzlhKvbtCglrjiJEHtxE6DXeSBjBb0Yw2oOSfq0vLJL+Rnjdn0uIJb6fMKHlZYJh7gf0sOSXu+pG8HNfRdB5JSES5BkzmxGN8o62mWdtKyo+U28zJ2d21QZsCfwP6GBmvWOI2F+BC+PAU5kTDcGewCPxdXkGYT22+oRR/K/iZzfCQqrXlSNj3AI4hvDwO5Uwm+1swpTj3sDbQPekjXGkF+HBkREfzncR/MXHSWptZivdGBcdj0PejNngdflEwqrKX8YeaK9Y3ozw+nkaIba3zKcZx97wAwQXSnVCnHEm65LMn0yYHfagmZ2e7r7LlFf/nN+/GpAdpxovkTSRMKOtu5k9Kum7pPzFKVqrm9lyM7syhq6NlDTAzCZLuod1DxCnGHgPeTMmxRgfQshX2x44TdJJhEGl5cBFhDCyoyyZ5PIZBBfEK/HzyLi/ipBb40fWuSmOU1gZJO2NcTxsDmBm04AZcSASM/uD8NsnusBnijHuA9wt6TGFbH83Ai8CoyU1MrNJhNDHtF3uqrzgPuTNHEmnECYaXGhmPyvkrx0EvGdmI2KdRHuccSBvNmEmWDMzWxhdGP0Jvu0bCdEgv+eEiaU7ks4g9Oo/IMRPf08YqOxIGEg9meAvn5GYSEBh4YExhOW39ie8ofxAcBddQ1hxZR9CD98H8IqJ95A3M3LxGVcirHO2Zzx+n5AjoZekc2JZmSesydEZjXE1wsDWLwRfNmb2AfAcYf27c4Hp6WyMY08/Z7854ffuC3xOmEjRkZBw5yFChrSBSRhjSS3jG1MObYDxZva2mV1BeID0IkR7nEvIV7HKjXHJ4D7kzYgNfMbNgF/N7F5JiwmLfM62sJrDBwR/4DRY59pIQichI9jvZnaspFrA/0nawswuMbMPJWUTl40vS42bQvw+a+L+scBWwAozmynpv4RE/h0IE1xGJuGnT2EL4A9JW0bXyXTgZEndzOwDM3tOIcd0e2CcmX2VoNYKh/eQNyNSjPG5hLCq52Ks7kuEFSjGStrLzBYDb1gyyeUzUnTmrBP3haTBcdCrN3BgnDyBmU1M99l4Kd/nWEKYWHXgJEknR1fQg8B3QOt4rsyR1FTSQWY2GfgUmKKwTNRU4C2gj6RjJe1CiKpIZNHUio73kDczFJaMH0rw++1BWNnjKjM7R9LWwOMKuYQTWVsupSe5L2Gm3Z5R603RWD8kaSAhc1jDdDbGG7yR7E2Imz7JwkKf7wMvx+90v6T/ALUsrMycBN2Ai6LmV+PA7n+ABYQHRi/Cv5slhBmEZZ67ZLPAzHyrwBshbeNlKcd9gKdSjjsAzwO7xeMtE9K5C3BN3N+BMJD0csr5Awk9t9PjcaWkf9tN/H79CDPtbgEaxrI9CQuuDklaX9RzCWHh0cPi8T6EnvtR8bgqUDfuK2m9FXFzl0XF53fgeEkXx+MJwJYx7hgLC0wuJCQ5hzBtNwlqEFaV+IeFqc5jgGWSTlRY2WMccDnhu9Qhrjg5BJAAAAveSURBVByd7kjaX9LDZvYiIcl8FeD/2zvzWLvqIo5/vpSt0GINQqW4FClQkAKWRQVEthQrBREwgIAghZaqCBKJqBUUG0QQRKVaBBEXooJQQ0QFQhAoFgFbqyCbSyTRGosIKouyfP1j5raXmwKPJe+c+958kpfenvt758x97ZszZ34z3zlA0rq2FxI3zIWNGglImkrcIB4DvpY1xjcQNd/nSTrK0fTxEAz+vsJwocrehig9j8ubAt8lcsVzCCnHqcTj5+1ErfFU242N08lqijcBJwK/tX165lx3JLSXf2j7EbV87FLPhiSSJhANLP+0fVxWMOxFlLl9yy0Qlpe0LnANMS9xQaYrjga+YPsKSW8nOvOub9TQYUBFyEMQSbsBt0uaLWmy7XuJR+Z3ENKI1xJC4qsRZU0HNemMARzTRxYTg0gnSfq4Q+lsAVH/+q5c2tqmD3jGBt5GeegPhCbwWpLm2r6KaIUeR4NRfldZ4VZEWeNvgNXzhnIR4aC/I2ma7RtsX7+SksniZaYi5CFG/tIcAJxBdLM9BPya0H64CzgX+KLt83N947q63WSkPBn4CLDY9pmSDgQW2P5bs9Y9P4qRSq8nnjxm2P5hHpsAnE/US89Sz2SQhmzdhRBmmkU0Az1IqOP9RdJ2xM3xeNuLGjRzWFFVFkMM25b0E6KmdSJRRnURoc41BliPyBGub/uzTTrj3sd7iEhZ0iJi8+tUSU/Y/mIzFg6M7s/haJD4o2LQ6lxJT9meD9wjaQmwoaSxbriJRTE89VhC63pR1qKfAWwk6UliQ++D5YwHl3LIQxDbj0q6hujC2w3Y2vahAOnstiPqeweVXgecN4/RRNfX37qOPylpMVGz2/qouCtNMZ3Ig/8Z+AHh8OZJGkMo6E0kxNsbK9Xr+jfYGhgL7CnpZw7pzOOIho+NgUtzU68YRCplMYRRqHJNJWQdb7N9QR5vJE3R2ZDTiskYaxCt0DfbvnWw7Xk5UWhTHEBsms4h2o1PUYjxTCPSGKfaXtKQfR2hoFd3bn5Z630QkT++zPYDTdhWrKAi5CGM7X9l+sLAvvn4f3FDzngf4DRJM23fmg7iv5Iud4Oi9y8Hmfdej9g4PQz4N/BZxVij621fq1Che6Ih+zo3wKlEGug64L+2T8ub4t5EOeG321D1MZypKoshQNeO+ciMipfjaIO+GrgC+FkD5iFpS2LzaDGRv96hK+d6f9e6SZI2aMLGF0JvtUFWiIwhhpNOs71XOt8jgHenQxx0Z6wcJpDO+G2EKt50orrmEIWO8TX5NZEYGls0SDnkPkXSZpLeKmk1IlcM0Wk1rnetox13foNVCsuA020fRWgXXyBpB1ihgqYYbf9OWi5y3lPfvbOk3fOtC4muthvzvSOA44ElDT2RbAJ8XlJHxW8MkZ4YR5QRHkPoMc8l6tNn9/uTylCgcsh9iKT9gdOJAZN/IUqs5gGvsv3XJm17NiStmpEkko4lSq1mONTlNrL9p35q+pA0i/gMI4gnkG8SynQfIcoN1yekKe9swM7NiRFc3wDusP1zSSOJuud5xGSVmyTNI8rxTnBpU7SCyiH3GRkRHwRMd4jUHEB0s80Gzu5a9wZCtrKpVuhn0HHG+XpePvWfK2khUWp1hHNadFvpcsZ7EKmJrTJPfAEhHPR1ovlmDPC0s814MMmU1TeBr2SDR4fHCS2K0cDEfDKZCMy0fc9g21msnEpZ9CfrEBKIENMlriRE3A+RtIpiaOb7mzJuINieR0T3hwOfbrMzlvRmSTMlbZc/27cDb5S0le3HiXbv1xEjpV5j+8EmnHEykrgRX5S2j4C4maSt5xFdj6cBc8sZt4tyyH1Gbg6dA+wv6W2Zn1xAdOPtbPtpx2yzs9sSHa+MjDK3AHZvqhRsIEjai4g4tye6HHcihI+uIDSNt7C9jJj2IRqYrtLDQ0QL9HYQjSpawWTgftvTiAj/st4NyqJZyiH3JzcRO+OHS9rF9lMO3Ydx+UtHgxHaQPkdIWjU2pHxufE4n5DHPJrQB/4YsUn5DcL5HSNpUnbezWqy6SMxMfh1z0xvdaJjAxsCJ0saldU3pdrWMsoh9yH56HkJsAT4uKQZuas/lkgDtB7bS92woNEAuR7YFSBver8H1svNuh8BTxDTrlcnnGGjOMY/nUtM5z4xW6RRDIU9k6i2+U+DJhbPQVVZ9DHpBHYCZhKPyl9yjOApXga0QhL0w8DdhGbz1sCBeVNEMdJoaVu63LqaQCYRG72jiM37scAptq9s1MDiOSmHPATIjRs3Ue861NEK9blTgW1sb5jH1+w45QZtE/E7/HTP8Y5THkWU5W0APGr7/u7SvaJ9lEMuiuchb3iTgROIRo8zG7ZneU13/n0VIq1yS5vruIvnp3LIRdHFyqoOHJKai4nc7M6SPjHohiWSJgLnSDpF0hZ5eA9gg3LG/U9FyEXRg1YiCZrHVyVyyEub6IjMDrxLiAqPXYDHbR+xknUj8iZS9BkVIRfDGkkTsuFjzazVHUHoPLyud63tJ23/qiFnvAYxweMi23MJ2dK1Jb03P8PIXLcpsHfePIo+oyLkYtgiaRqhCfIPQgh/ju07JY13TL5uFZJeZfuBdLaLiLFcDwKPAL+wfamkdwN/ckwTL/qMcsjFsETSjsSj/3ttL5b0VWAt20f2rJsEPGB7aQNmPiuSDrb9/Xz9KWCs7Q81bFbxEqmURTGc+XxX3fapwJhMDQAgaSwxcaU1kqCdTceOM05uJdIXa1UrdH9TDrkYrvyS0KPolLWtQYxZWiePrZft0OelVkUr6K0hTk2Qs4gRTI9WjXF/Uw65GJak/kdHYU6ELsWDtpdJOhSYI2lkW0vJJI3IErjTCHH5n1R03P9UDrkoEkkXA0uBKYSgUGuFj2B5ZL++7aXVgTc0KIdcDHsyslwNuCv/3MP2fc1aVQxHyiEXRSLpSOC2JsYuFQWUQy6K5dRjf9E05ZCLoihaQlVZFEVRtIRyyEVRFC2hHHJRFEVLKIdcFEXREsohF61D0lOSfi3pDkmXSVrrJZxrV0k/ztf7Sjr5OdaOkfSBF3GNT0v66ECP96y5WNKBL+Ba4yXd8UJtLPqDcshFG3nM9ja2twT+Bxzb/WbqFr/g/7u2r7R9xnMsGUPoDBdFI5RDLtrOTcCEjAzvkfRt4A7gtZKmSFooaVFG0qMAJL1D0t2SFgH7d04k6UhJ5+XrsZLmS1qSXzsCZwAbZ3R+Vq47SdJtkn4j6TNd5/qkpHslLQA2e74PIemYPM8SSZf3RP17Sro9zzct14+QdFbXtWe+1B9k0X7KIRetJYXYpwIdTYlNgK/afiMhyj4b2NP2ZOB24ERJawIXAPsA2wKvfpbTfxm4wfbWxADTO4GTgT9kdH6SpCl5zR2AbYBtJe0iaVvg4Dz2TmD7AXycK2xvn9e7C5je9d74vMbewLz8DNOBh21vn+c/RtJGA7hO0cfUmJeijYyU1Jl4cRMhJD8O+LPtW/L4W4AtgJtT5Gx1YCEwkZiYcR+ApO8CM1Zyjd2B98HyIaYPS3plz5op+dXRTB5FOOjRwPyOEpykKwfwmbaUNIdIi4wCru5671LbTwP3SfpjfoYpwFZd+eVX5LXvHcC1ij6lHHLRRh6zvU33gXS6j3QfAq61fUjPumd830tEwOdsn99zjRNexLkuBvazvSQ1M3bteq+3XdZ57eNsdztuJI1/Edcu+oRKWRT9yi3ATpImAEhaOwd83g2Ml7RxrjvkWb7/OmBWfu8ISa8A/k1Evx2uBo7qyk1vKGl94EZgP0kjFROq9xmAvaOBpZJWAw7tee89klZJm98A3JPXnpXrkbSppLUHcJ2ij6kIuehLUkj+SOB7XWOXZtu+V9IM4CpJjxIpj9ErOcXxwNclTQeeAmbZXijp5iwr+2nmkTcHFmaE/h/gMNuLJP0AWAL8HbhtACZ/iphSsiz/7LbpfmIM0zrAsbYfl3QhkVtelPKgy4D9BvbTKfqVEhcqiqJoCZWyKIqiaAnlkIuiKFpCOeSiKIqWUA65KIqiJZRDLoqiaAnlkIuiKFpCOeSiKIqW8H+AhG1aoAUfUwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_s, y_t=inference(model_FT,test_loader)\n",
    "r_AUC_wholeSet = get_AUC(y_s, y_t,True)\n",
    "print('macro AUC for the model trained on full data is : ', r_AUC_wholeSet['macro'])\n",
    "\n",
    "# Compute confusion matrix\n",
    "cnf_matrix = confusion_matrix(y_t.dot([0,1,2,3]), np.argmax(y_s,axis=1))\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "# Plot non-normalized confusion matrix\n",
    "plt.figure()\n",
    "class_names = ['0: None Class','1: Infiltration','2: Pneumothorax','3: Cardiomegaly']\n",
    "plot_confusion_matrix(cnf_matrix, classes=class_names,\n",
    "                      title='Confusion matrix')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.4) (10 points)\n",
    "Descibe your findings using two different TL approaches that you implemented in Q2.2. and Q2.3. for example what are the differences in terms of speed, accuracy, and so on. Was TL successfull classifying lung diseases?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fine tuning (FT) was almost a minute slower than feature extraction (FE) (8m:1s vs 8m:54s). As I only ran the training once for each model there could be some variation in the times. You would expect the fine tuning model to be slightly slower as the weights for the entire model are being updated versus the feature extraction model in which only a single dense layer is being updated.\n",
    "\n",
    "The best accuracy was better in the FT model (0.453125 vs 0.437500). The FT was able to overfit to the training data very quickly while the FE model did not.\n",
    "\n",
    "The FE model predicted the None class for almost all of the test samples while the FT model predicted some Infiltration as well as None Class and had a betro macro AUC. The FE did not predict any pneumothorax while the FT model did. Neither model predicted any Cardiomegaly.\n",
    "\n",
    "While FT at least created a model that predicted more than one class, neither model performed well at classifying lung disease."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.5) (10 points) \n",
    "In the view of your findings, describe transfer learning approaches we used in this HW in terms of data size and emprical observation domain specifics (Hint: Checkslide # 16 on Lecture #12). What can you do better to improve transfer learning? Propose changes to the current TL strategies. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The data size of the new data set is small and the domain is very different from the domain used to train the model.\n",
    "\n",
    "Guidelines indicate that when using a small dataset it is better to only train the classification layer of the model.\n",
    "\n",
    "As the dataset is very different it would be better to only use lower levels within the existing model as the higher levels will be specialized at identifying specific features found in the dataset used to train the model.\n",
    "\n",
    "The changes I propose are the only use the first few sequential layers within resnet and freeze the weights of these layers. A final classification layer will be trained to make predictions."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.6) Bonus Question (maximum 20 points) \n",
    "Implement the changes you proposed in Q2.5 and analyze your results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MyModel (\n",
      "  (model): Sequential (\n",
      "    (0): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "    (1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "    (2): ReLU (inplace)\n",
      "    (3): MaxPool2d (size=(3, 3), stride=(2, 2), padding=(1, 1), dilation=(1, 1))\n",
      "    (4): Sequential (\n",
      "      (0): Bottleneck (\n",
      "        (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "        (downsample): Sequential (\n",
      "          (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "          (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        )\n",
      "      )\n",
      "      (1): Bottleneck (\n",
      "        (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "      (2): Bottleneck (\n",
      "        (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "    )\n",
      "    (5): AvgPool2d (size=7, stride=7, padding=0, ceil_mode=False, count_include_pad=True)\n",
      "  )\n",
      "  (fc): Linear (16384 -> 4)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "from torch.optim import lr_scheduler\n",
    "import torch.nn as nn\n",
    "\n",
    "class MyModel(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(MyModel, self).__init__()\n",
    "        resnet50 = torchvision.models.resnet50(pretrained=True)\n",
    "        for param in resnet50.parameters():\n",
    "            param.requires_grad = False\n",
    "        num_ftrs = resnet50.fc.in_features\n",
    "        front = list(resnet50.children())[:5]\n",
    "        back = list(resnet50.children())[-2:]\n",
    "        self.model = nn.Sequential(*(front + [back[0]]))\n",
    "        self.fc = nn.Linear(16384, 4)\n",
    "        \n",
    "    def forward(self, image):\n",
    "        x = self.model(image)\n",
    "        x = x.view(-1, 16384)\n",
    "#         print(x.size())\n",
    "        x = self.fc(x)\n",
    "        return x\n",
    "\n",
    "resnet50 = MyModel()\n",
    "print(resnet50)\n",
    "resnet50.cuda()\n",
    "\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.SGD(resnet50.fc.parameters(), lr=0.001, momentum=0.9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([16, 4])\n"
     ]
    }
   ],
   "source": [
    "for sample_batched in train_loader:\n",
    "    # get the inputs\n",
    "    #print(sample_batched)\n",
    "    inputs = sample_batched['x']\n",
    "    labels = sample_batched['y']\n",
    "\n",
    "    # wrap them in Variable\n",
    "    if use_gpu:\n",
    "        inputs = Variable(inputs).type(torch.FloatTensor).cuda()\n",
    "        labels = Variable(labels).type(torch.LongTensor).cuda()\n",
    "    else:\n",
    "        inputs, labels = Variable(inputs).type(torch.FloatTensor), Variable(labels).type(torch.LongTensor)\n",
    "\n",
    "    # zero the parameter gradients\n",
    "    optimizer.zero_grad()\n",
    "\n",
    "    # forward\n",
    "    outputs = resnet50(inputs)\n",
    "    print(outputs.size())\n",
    "#     _, preds = torch.max(outputs.data, 1)\n",
    "#     loss = criterion(outputs, labels)\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training complete in 7m 24s\n",
      "Best val Acc: 0.484375\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_M = train_model(resnet50, criterion, optimizer,\n",
    "                       num_epochs=100, trainVal=['train','val'],verbose=False) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "macro AUC for the model trained on full data is :  0.593756582391\n",
      "Confusion matrix, without normalization\n",
      "[[ 240 1446    0    0]\n",
      " [ 121 1818    0    0]\n",
      " [  31  407    0    0]\n",
      " [  23  195    0    0]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_s, y_t=inference(model_M,test_loader)\n",
    "r_AUC_wholeSet = get_AUC(y_s, y_t,True)\n",
    "print('macro AUC for the model trained on full data is : ', r_AUC_wholeSet['macro'])\n",
    "\n",
    "# Compute confusion matrix\n",
    "cnf_matrix = confusion_matrix(y_t.dot([0,1,2,3]), np.argmax(y_s,axis=1))\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "# Plot non-normalized confusion matrix\n",
    "plt.figure()\n",
    "class_names = ['0: None Class','1: Infiltration','2: Pneumothorax','3: Cardiomegaly']\n",
    "plot_confusion_matrix(cnf_matrix, classes=class_names,\n",
    "                      title='Confusion matrix')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.optim import lr_scheduler\n",
    "import torch.nn as nn\n",
    "\n",
    "class MyModel(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(MyModel, self).__init__()\n",
    "        resnet50 = torchvision.models.resnet50(pretrained=True)\n",
    "        for param in resnet50.parameters():\n",
    "            param.requires_grad = False\n",
    "        num_ftrs = resnet50.fc.in_features\n",
    "        front = list(resnet50.children())[:6]\n",
    "        back = list(resnet50.children())[-2:]\n",
    "        self.model = nn.Sequential(*(front + [back[0]]))\n",
    "        self.fc = nn.Linear(8192, 4)\n",
    "        \n",
    "    def forward(self, image):\n",
    "        x = self.model(image)\n",
    "        x = x.view(-1, 8192)\n",
    "#         print(x.size())\n",
    "        x = self.fc(x)\n",
    "        return x\n",
    "\n",
    "resnet50 = MyModel()\n",
    "\n",
    "resnet50.cuda()\n",
    "\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.SGD(resnet50.fc.parameters(), lr=0.001, momentum=0.9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([16, 4])\n"
     ]
    }
   ],
   "source": [
    "for sample_batched in train_loader:\n",
    "    # get the inputs\n",
    "    #print(sample_batched)\n",
    "    inputs = sample_batched['x']\n",
    "    labels = sample_batched['y']\n",
    "\n",
    "    # wrap them in Variable\n",
    "    if use_gpu:\n",
    "        inputs = Variable(inputs).type(torch.FloatTensor).cuda()\n",
    "        labels = Variable(labels).type(torch.LongTensor).cuda()\n",
    "    else:\n",
    "        inputs, labels = Variable(inputs).type(torch.FloatTensor), Variable(labels).type(torch.LongTensor)\n",
    "\n",
    "    # zero the parameter gradients\n",
    "    optimizer.zero_grad()\n",
    "\n",
    "    # forward\n",
    "    outputs = resnet50(inputs)\n",
    "    print(outputs.size())\n",
    "#     _, preds = torch.max(outputs.data, 1)\n",
    "#     loss = criterion(outputs, labels)\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training complete in 7m 35s\n",
      "Best val Acc: 0.468750\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_M = train_model(resnet50, criterion, optimizer,\n",
    "                       num_epochs=100, trainVal=['train','val'],verbose=False) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "macro AUC for the model trained on full data is :  0.576052119851\n",
      "Confusion matrix, without normalization\n",
      "[[ 333 1353    0    0]\n",
      " [ 169 1770    0    0]\n",
      " [  43  395    0    0]\n",
      " [  26  192    0    0]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_s, y_t=inference(model_M,test_loader)\n",
    "r_AUC_wholeSet = get_AUC(y_s, y_t,True)\n",
    "print('macro AUC for the model trained on full data is : ', r_AUC_wholeSet['macro'])\n",
    "\n",
    "# Compute confusion matrix\n",
    "cnf_matrix = confusion_matrix(y_t.dot([0,1,2,3]), np.argmax(y_s,axis=1))\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "# Plot non-normalized confusion matrix\n",
    "plt.figure()\n",
    "class_names = ['0: None Class','1: Infiltration','2: Pneumothorax','3: Cardiomegaly']\n",
    "plot_confusion_matrix(cnf_matrix, classes=class_names,\n",
    "                      title='Confusion matrix')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MyModel (\n",
      "  (model): Sequential (\n",
      "    (0): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "    (1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "    (2): ReLU (inplace)\n",
      "    (3): MaxPool2d (size=(3, 3), stride=(2, 2), padding=(1, 1), dilation=(1, 1))\n",
      "    (4): Sequential (\n",
      "      (0): Bottleneck (\n",
      "        (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "        (downsample): Sequential (\n",
      "          (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "          (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        )\n",
      "      )\n",
      "      (1): Bottleneck (\n",
      "        (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "      (2): Bottleneck (\n",
      "        (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential (\n",
      "      (0): Bottleneck (\n",
      "        (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "        (downsample): Sequential (\n",
      "          (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "          (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "        )\n",
      "      )\n",
      "      (1): Bottleneck (\n",
      "        (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "      (2): Bottleneck (\n",
      "        (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "      (3): Bottleneck (\n",
      "        (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential (\n",
      "      (0): Bottleneck (\n",
      "        (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "        (downsample): Sequential (\n",
      "          (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "          (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
      "        )\n",
      "      )\n",
      "      (1): Bottleneck (\n",
      "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "      (2): Bottleneck (\n",
      "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "      (3): Bottleneck (\n",
      "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "      (4): Bottleneck (\n",
      "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "      (5): Bottleneck (\n",
      "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True)\n",
      "        (relu): ReLU (inplace)\n",
      "      )\n",
      "    )\n",
      "    (7): AvgPool2d (size=7, stride=7, padding=0, ceil_mode=False, count_include_pad=True)\n",
      "  )\n",
      "  (fc): Linear (4096 -> 4)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "from torch.optim import lr_scheduler\n",
    "import torch.nn as nn\n",
    "\n",
    "class MyModel(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(MyModel, self).__init__()\n",
    "        resnet50 = torchvision.models.resnet50(pretrained=True)\n",
    "        for param in resnet50.parameters():\n",
    "            param.requires_grad = False\n",
    "        num_ftrs = resnet50.fc.in_features\n",
    "        front = list(resnet50.children())[:7]\n",
    "        back = list(resnet50.children())[-2:]\n",
    "        self.model = nn.Sequential(*(front + [back[0]]))\n",
    "        self.fc = nn.Linear(4096, 4)\n",
    "        \n",
    "    def forward(self, image):\n",
    "        x = self.model(image)\n",
    "        x = x.view(-1, 4096)\n",
    "#         print(x.size())\n",
    "        x = self.fc(x)\n",
    "        return x\n",
    "\n",
    "resnet50 = MyModel()\n",
    "print(resnet50)\n",
    "resnet50.cuda()\n",
    "\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.SGD(resnet50.fc.parameters(), lr=0.001, momentum=0.9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([16, 4])\n"
     ]
    }
   ],
   "source": [
    "for sample_batched in train_loader:\n",
    "    # get the inputs\n",
    "    #print(sample_batched)\n",
    "    inputs = sample_batched['x']\n",
    "    labels = sample_batched['y']\n",
    "\n",
    "    # wrap them in Variable\n",
    "    if use_gpu:\n",
    "        inputs = Variable(inputs).type(torch.FloatTensor).cuda()\n",
    "        labels = Variable(labels).type(torch.LongTensor).cuda()\n",
    "    else:\n",
    "        inputs, labels = Variable(inputs).type(torch.FloatTensor), Variable(labels).type(torch.LongTensor)\n",
    "\n",
    "    # zero the parameter gradients\n",
    "    optimizer.zero_grad()\n",
    "\n",
    "    # forward\n",
    "    outputs = resnet50(inputs)\n",
    "    print(outputs.size())\n",
    "#     _, preds = torch.max(outputs.data, 1)\n",
    "#     loss = criterion(outputs, labels)\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training complete in 7m 39s\n",
      "Best val Acc: 0.468750\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_M = train_model(resnet50, criterion, optimizer,\n",
    "                       num_epochs=100, trainVal=['train','val'],verbose=False) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "macro AUC for the model trained on full data is :  0.585491703728\n",
      "Confusion matrix, without normalization\n",
      "[[ 635 1051    0    0]\n",
      " [ 500 1439    0    0]\n",
      " [ 124  314    0    0]\n",
      " [  63  155    0    0]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_s, y_t=inference(model_M,test_loader)\n",
    "r_AUC_wholeSet = get_AUC(y_s, y_t,True)\n",
    "print('macro AUC for the model trained on full data is : ', r_AUC_wholeSet['macro'])\n",
    "\n",
    "# Compute confusion matrix\n",
    "cnf_matrix = confusion_matrix(y_t.dot([0,1,2,3]), np.argmax(y_s,axis=1))\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "# Plot non-normalized confusion matrix\n",
    "plt.figure()\n",
    "class_names = ['0: None Class','1: Infiltration','2: Pneumothorax','3: Cardiomegaly']\n",
    "plot_confusion_matrix(cnf_matrix, classes=class_names,\n",
    "                      title='Confusion matrix')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "RESNET 1 sequential Layer\n",
    "+ Train Time: 7m:24s\n",
    "+ parameters in final layer: 65,540\n",
    "+ Best validation Accuracy: 48%\n",
    "+ Macro AUC: 59%\n",
    "\n",
    "```\n",
    "[[ 240 1446    0    0]\n",
    " [ 121 1818    0    0]\n",
    " [  31  407    0    0]\n",
    " [  23  195    0    0]]\n",
    "```\n",
    "\n",
    "Comments: Only predicted infiltration or no class, far more infiltration.\n",
    "\n",
    "RESNET 2 sequential Layer\n",
    "+ Train Time: 7m:35s\n",
    "+ parameters in final layer: 32,772\n",
    "+ Best validation Accuracy: 47%\n",
    "+ Macro AUC: 58%\n",
    "\n",
    "```\n",
    "[[ 333 1353    0    0]\n",
    " [ 169 1770    0    0]\n",
    " [  43  395    0    0]\n",
    " [  26  192    0    0]]\n",
    "```\n",
    "\n",
    "Comments: Predicted most as infiltration and some as No class, pneumothorax and cardiomegaly did not get any predictions.\n",
    "Macro AUC was 2% better than FT.\n",
    "\n",
    "RESNET 3 sequential Layer\n",
    "+ Train Time: 7m:39s\n",
    "+ parameters in final layer: 16,388\n",
    "+ Best validation Accuracy: 47%%\n",
    "+ Macro AUC: 59% \n",
    "\n",
    "```\n",
    "[[ 635 1051    0    0]\n",
    " [ 500 1439    0    0]\n",
    " [ 124  314    0    0]\n",
    " [  63  155    0    0]]\n",
    "```\n",
    "\n",
    "Comments: Once again the number of no class predicted went up. The model still does not predict pneumothorax or cardiomegaly. \n",
    "\n",
    "Findings:\n",
    "\n",
    "Given fewer layers the model will be less capable of abstraction, the reason that infiltration is identified more frequently by a architecture with fewer sequential layers could be due to infiltration being the most common class in the data set. As you increase the number of sequential layers in resnet the model predicts more images as being type None. These changes resulted in macro AUC scores that were far better than FE using the entire mode and slightly better scores than FT. While the scores improved this is still not a very effective architecture for this problem. "
   ]
  }
 ],
 "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}