CheXpert_DenseNet121_FL.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Prerequisites"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import time\n",
    "import csv\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "from PIL import Image\n",
    "from barbar import Bar\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.backends.cudnn as cudnn\n",
    "import torchvision\n",
    "import torchvision.transforms as transforms\n",
    "from torch.utils.data import Dataset\n",
    "from torch.utils.data import DataLoader\n",
    "from torch.utils.data.dataset import random_split\n",
    "\n",
    "import sklearn.metrics as metrics\n",
    "from sklearn.metrics import roc_auc_score\n",
    "\n",
    "use_gpu = torch.cuda.is_available()\n",
    "print(use_gpu)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pre-define Value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train data length: 190527\n",
      "Valid data length: 202\n",
      "Test data length: 500\n"
     ]
    }
   ],
   "source": [
    "# Paths to the files with training, and validation sets.\n",
    "# Each file contains pairs (path to image, output vector)\n",
    "Traindata = pd.read_csv('./CheXpert-v1.0-small/train.csv')\n",
    "Traindata = Traindata[Traindata['Path'].str.contains(\"frontal\")] # use only frontal images\n",
    "Traindata = Traindata[500:]\n",
    "Traindata.to_csv('./CheXpert-v1.0-small/train_mod.csv', index = False)\n",
    "print(\"Train data length:\", len(Traindata))\n",
    "\n",
    "Validdata = pd.read_csv('./CheXpert-v1.0-small/valid.csv')\n",
    "Validdata = Validdata[Validdata['Path'].str.contains(\"frontal\")] # use only frontal images\n",
    "Validdata.to_csv('./CheXpert-v1.0-small/valid_mod.csv', index = False)\n",
    "print(\"Valid data length:\", len(Validdata))\n",
    "\n",
    "Testdata = Traindata.head(500) # use first 500 training data as test data (obs ratio is almost same!)\n",
    "Testdata.to_csv('./CheXpert-v1.0-small/test_mod.csv', index = False)\n",
    "print(\"Test data length:\", len(Testdata))\n",
    "\n",
    "pathFileTrain = './CheXpert-v1.0-small/train_mod.csv'\n",
    "pathFileValid = './CheXpert-v1.0-small/valid_mod.csv'\n",
    "pathFileTest = './CheXpert-v1.0-small/test_mod.csv'\n",
    "\n",
    "# Neural network parameters:\n",
    "nnIsTrained = False     # pre-trained using ImageNet\n",
    "nnClassCount = 14       # dimension of the output\n",
    "\n",
    "# Training settings: batch size, maximum number of epochs\n",
    "trBatchSize = 16\n",
    "trMaxEpoch = 3\n",
    "\n",
    "# Parameters related to image transforms: size of the down-scaled image, cropped image\n",
    "imgtransResize = (320, 320)\n",
    "imgtransCrop = 224\n",
    "\n",
    "# Class names\n",
    "class_names = ['No Finding', 'Enlarged Cardiomediastinum', 'Cardiomegaly', 'Lung Opacity', \n",
    "               'Lung Lesion', 'Edema', 'Consolidation', 'Pneumonia', 'Atelectasis', 'Pneumothorax', \n",
    "               'Pleural Effusion', 'Pleural Other', 'Fracture', 'Support Devices']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create a Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "class CheXpertDataSet(Dataset):\n",
    "    def __init__(self, data_PATH, transform = None, policy = \"ones\"):\n",
    "        \"\"\"\n",
    "        data_PATH: path to the file containing images with corresponding labels.\n",
    "        transform: optional transform to be applied on a sample.\n",
    "        Upolicy: name the policy with regard to the uncertain labels.\n",
    "        \"\"\"\n",
    "        image_names = []\n",
    "        labels = []\n",
    "\n",
    "        with open(data_PATH, \"r\") as f:\n",
    "            csvReader = csv.reader(f)\n",
    "            next(csvReader, None) # skip the header\n",
    "            for line in csvReader:\n",
    "                image_name = line[0]\n",
    "                label = line[5:]\n",
    "                \n",
    "                for i in range(14):\n",
    "                    if label[i]:\n",
    "                        a = float(label[i])\n",
    "                        if a == 1:\n",
    "                            label[i] = 1\n",
    "                        elif a == -1:\n",
    "                            if policy == \"ones\":\n",
    "                                label[i] = 1\n",
    "                            elif policy == \"zeroes\":\n",
    "                                label[i] = 0\n",
    "                            else:\n",
    "                                label[i] = 0\n",
    "                        else:\n",
    "                            label[i] = 0\n",
    "                    else:\n",
    "                        label[i] = 0\n",
    "                        \n",
    "                image_names.append('./' + image_name)\n",
    "                labels.append(label)\n",
    "\n",
    "        self.image_names = image_names\n",
    "        self.labels = labels\n",
    "        self.transform = transform\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        \"\"\"Take the index of item and returns the image and its labels\"\"\"\n",
    "        image_name = self.image_names[index]\n",
    "        image = Image.open(image_name).convert('RGB')\n",
    "        label = self.labels[index]\n",
    "        if self.transform is not None:\n",
    "            image = self.transform(image)\n",
    "        return image, torch.FloatTensor(label)\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.image_names)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load Datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train data length: 190527\n",
      "Valid data length: 202\n",
      "Test data length: 500\n"
     ]
    }
   ],
   "source": [
    "IMAGENET_MEAN = [0.485, 0.456, 0.406]  # mean of ImageNet dataset(for normalization)\n",
    "IMAGENET_STD = [0.229, 0.224, 0.225]   # std of ImageNet dataset(for normalization)\n",
    "\n",
    "# Tranform data\n",
    "normalize = transforms.Normalize(IMAGENET_MEAN, IMAGENET_STD)\n",
    "transformList = []\n",
    "\n",
    "transformList.append(transforms.Resize((imgtransCrop, imgtransCrop))) # 224\n",
    "# transformList.append(transforms.RandomResizedCrop(imgtransCrop))\n",
    "# transformList.append(transforms.RandomHorizontalFlip())\n",
    "transformList.append(transforms.ToTensor())\n",
    "# transformList.append(normalize)\n",
    "transformSequence = transforms.Compose(transformList)\n",
    "\n",
    "# Load dataset\n",
    "datasetTrain = CheXpertDataSet(pathFileTrain, transformSequence, policy = \"ones\")\n",
    "print(\"Train data length:\", len(datasetTrain))\n",
    "\n",
    "datasetValid = CheXpertDataSet(pathFileValid, transformSequence)\n",
    "print(\"Valid data length:\", len(datasetValid))\n",
    "\n",
    "datasetTest = CheXpertDataSet(pathFileTest, transformSequence, policy = \"ones\")\n",
    "print(\"Test data length:\", len(datasetTest))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create train and test Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class CheXpertTrainer():\n",
    "\n",
    "    def train(model, dataLoaderTrain, dataLoaderVal, nnClassCount, trMaxEpoch, checkpoint):\n",
    "        optimizer = optim.Adam(model.parameters(), lr = 0.0001, # setting optimizer & scheduler\n",
    "                               betas = (0.9, 0.999), eps = 1e-08, weight_decay = 0) \n",
    "        loss = torch.nn.BCELoss() # setting loss function\n",
    "        \n",
    "        if checkpoint != None and use_gpu: # loading checkpoint\n",
    "            modelCheckpoint = torch.load(checkpoint)\n",
    "            model.load_state_dict(modelCheckpoint['state_dict'])\n",
    "            optimizer.load_state_dict(modelCheckpoint['optimizer'])\n",
    "            \n",
    "        # Train the network\n",
    "        lossMIN = 100000\n",
    "        train_start = []\n",
    "        train_end = []\n",
    "        for epochID in range(0, trMaxEpoch):\n",
    "            train_start.append(time.time()) # training starts\n",
    "            losst = CheXpertTrainer.epochTrain(model, dataLoaderTrain, optimizer, trMaxEpoch, nnClassCount, loss)\n",
    "            train_end.append(time.time()) # training ends\n",
    "            lossv = CheXpertTrainer.epochVal(model, dataLoaderVal, optimizer, trMaxEpoch, nnClassCount, loss)\n",
    "            print(\"Training loss: {:.3f},\".format(losst), \"Valid loss: {:.3f}\".format(lossv))\n",
    "            \n",
    "            if lossv < lossMIN:\n",
    "                lossMIN = lossv\n",
    "                torch.save({'epoch': epochID + 1, 'state_dict': model.state_dict(), \n",
    "                            'best_loss': lossMIN, 'optimizer' : optimizer.state_dict()}, \n",
    "                           'm-epoch_FL' + str(epochID + 1) + '.pth.tar')\n",
    "                print('Epoch ' + str(epochID + 1) + ' [save] loss = ' + str(lossv))\n",
    "            else:\n",
    "                print('Epoch ' + str(epochID + 1) + ' [----] loss = ' + str(lossv))\n",
    "\n",
    "        train_time = np.array(train_end) - np.array(train_start)\n",
    "        print(\"Training time for each epoch: {} seconds\".format(train_time.round(0)))\n",
    "        params = model.state_dict()\n",
    "        return params\n",
    "       \n",
    "        \n",
    "    def epochTrain(model, dataLoaderTrain, optimizer, epochMax, classCount, loss):\n",
    "        losstrain = 0\n",
    "        model.train()\n",
    "\n",
    "        for batchID, (varInput, target) in enumerate(Bar(dataLoaderTrain)):\n",
    "            \n",
    "            varTarget = target.cuda(non_blocking = True)\n",
    "            varOutput = model(varInput)\n",
    "            lossvalue = loss(varOutput, varTarget)\n",
    "                       \n",
    "            optimizer.zero_grad()\n",
    "            lossvalue.backward()\n",
    "            optimizer.step()\n",
    "            \n",
    "            losstrain += lossvalue.item()\n",
    "            \n",
    "        return losstrain / len(dataLoaderTrain)\n",
    "    \n",
    "    \n",
    "    def epochVal(model, dataLoaderVal, optimizer, epochMax, classCount, loss):\n",
    "        model.eval()\n",
    "        lossVal = 0\n",
    "\n",
    "        with torch.no_grad():\n",
    "            for i, (varInput, target) in enumerate(dataLoaderVal):\n",
    "                \n",
    "                target = target.cuda(non_blocking = True)\n",
    "                varOutput = model(varInput)\n",
    "                \n",
    "                lossVal += loss(varOutput, target)\n",
    "                \n",
    "        return lossVal / len(dataLoaderVal)\n",
    "\n",
    "    \n",
    "    def computeAUROC(dataGT, dataPRED, classCount):\n",
    "        # Computes area under ROC curve \n",
    "        # dataGT: ground truth data\n",
    "        # dataPRED: predicted data\n",
    "        outAUROC = []\n",
    "        datanpGT = dataGT.cpu().numpy()\n",
    "        datanpPRED = dataPRED.cpu().numpy()\n",
    "        \n",
    "        for i in range(classCount):\n",
    "            try:\n",
    "                outAUROC.append(roc_auc_score(datanpGT[:, i], datanpPRED[:, i]))\n",
    "            except ValueError:\n",
    "                pass\n",
    "        return outAUROC\n",
    "    \n",
    "    \n",
    "    def test(model, dataLoaderTest, nnClassCount, checkpoint, class_names):\n",
    "        cudnn.benchmark = True\n",
    "        \n",
    "        if checkpoint != None and use_gpu:\n",
    "            modelCheckpoint = torch.load(checkpoint)\n",
    "            model.load_state_dict(modelCheckpoint['state_dict'])\n",
    "\n",
    "        if use_gpu:\n",
    "            outGT = torch.FloatTensor().cuda()\n",
    "            outPRED = torch.FloatTensor().cuda()\n",
    "        else:\n",
    "            outGT = torch.FloatTensor()\n",
    "            outPRED = torch.FloatTensor()\n",
    "       \n",
    "        model.eval()\n",
    "        \n",
    "        with torch.no_grad():\n",
    "            for i, (input, target) in enumerate(dataLoaderTest):\n",
    "\n",
    "                target = target.cuda()\n",
    "                outGT = torch.cat((outGT, target), 0).cuda()\n",
    "\n",
    "                bs, c, h, w = input.size()\n",
    "                varInput = input.view(-1, c, h, w)\n",
    "            \n",
    "                out = model(varInput)\n",
    "                outPRED = torch.cat((outPRED, out), 0)\n",
    "        aurocIndividual = CheXpertTrainer.computeAUROC(outGT, outPRED, nnClassCount)\n",
    "        aurocMean = np.array(aurocIndividual).mean()\n",
    "        print('AUROC mean ', aurocMean)\n",
    "        \n",
    "        for i in range (0, len(aurocIndividual)):\n",
    "            print(class_names[i], ' ', aurocIndividual[i])\n",
    "        \n",
    "        return outGT, outPRED"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define the Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "class DenseNet121(nn.Module):\n",
    "    \"\"\"Model modified.\n",
    "    The architecture of our model is the same as standard DenseNet121\n",
    "    except the classifier layer which has an additional sigmoid function.\n",
    "    \"\"\"\n",
    "    def __init__(self, out_size):\n",
    "        super(DenseNet121, self).__init__()\n",
    "        self.densenet121 = torchvision.models.densenet121(pretrained = False)\n",
    "        num_ftrs = self.densenet121.classifier.in_features\n",
    "        self.densenet121.classifier = nn.Sequential(\n",
    "            nn.Linear(num_ftrs, out_size),\n",
    "            nn.Sigmoid()\n",
    "        )\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.densenet121(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train the model with Federated Learning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# FOR MULTIPLE COMMUNICATION ROUNDS\n",
    "com_round = 3\n",
    "fraction = 1.0\n",
    "epoch = 3\n",
    "batch = 48\n",
    "num_clients = 5\n",
    "\n",
    "'''\n",
    "# Divide datasetTrain_ex\n",
    "datasetTrain_1, datasetTrain_2, datasetTrain_3, datasetTrain_4, datasetTrain_5, dataleft = random_split(datasetTrain, \n",
    "                                                                                                        [100, 100, 100, 100, 100,\n",
    "                                                                                                         len(datasetTrain) - 500])\n",
    "'''\n",
    "# Divide datasetTrain_real\n",
    "datasetTrain_1, datasetTrain_2, datasetTrain_3, datasetTrain_4, datasetTrain_5 = random_split(datasetTrain, \n",
    "                                                                                              [38106, 38106, 38105, 38105, 38105])\n",
    "\n",
    "\n",
    "# Define 5 DataLoaders\n",
    "dataLoaderTrain_1 = DataLoader(dataset = datasetTrain_1, batch_size = trBatchSize,\n",
    "                               shuffle = True, num_workers = 2, pin_memory = True)\n",
    "dataLoaderTrain_2 = DataLoader(dataset = datasetTrain_2, batch_size = trBatchSize,\n",
    "                               shuffle = True, num_workers = 2, pin_memory = True)\n",
    "dataLoaderTrain_3 = DataLoader(dataset = datasetTrain_3, batch_size = trBatchSize,\n",
    "                               shuffle = True, num_workers = 2, pin_memory = True)\n",
    "dataLoaderTrain_4 = DataLoader(dataset = datasetTrain_4, batch_size = trBatchSize,\n",
    "                               shuffle = True, num_workers = 2, pin_memory = True)\n",
    "dataLoaderTrain_5 = DataLoader(dataset = datasetTrain_5, batch_size = trBatchSize,\n",
    "                               shuffle = True, num_workers = 2, pin_memory = True)\n",
    "\n",
    "# Define Valid and Test DataLoaders\n",
    "dataLoaderVal = DataLoader(dataset = datasetValid, batch_size = trBatchSize, \n",
    "                           shuffle = False, num_workers = 2, pin_memory = True)\n",
    "dataLoaderTest = DataLoader(dataset = datasetTest, num_workers = 2, pin_memory = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of clients:  5\n",
      "38106/38106: [===============================>] - ETA 2.1ssss\n",
      "Training loss: 0.403, Valid loss: 0.405\n",
      "Epoch 1 [save] loss = tensor(0.4050, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.379, Valid loss: 0.418\n",
      "Epoch 2 [----] loss = tensor(0.4181, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.371, Valid loss: 0.416\n",
      "Epoch 3 [----] loss = tensor(0.4156, device='cuda:0')\n",
      "Training time for each epoch: [622. 224. 189.] seconds\n",
      "Client 1 training ends with 1039 seconds in round 1\n",
      "38106/38106: [===============================>] - ETA 2.1ssss\n",
      "Training loss: 0.374, Valid loss: 0.407\n",
      "Epoch 1 [save] loss = tensor(0.4069, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.366, Valid loss: 0.415\n",
      "Epoch 2 [----] loss = tensor(0.4151, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.358, Valid loss: 0.424\n",
      "Epoch 3 [----] loss = tensor(0.4245, device='cuda:0')\n",
      "Training time for each epoch: [730. 234. 191.] seconds\n",
      "Client 2 training ends with 1158 seconds in round 1\n",
      "38105/38105: [===============================>] - ETA 2.0ssss\n",
      "Training loss: 0.366, Valid loss: 0.456\n",
      "Epoch 1 [save] loss = tensor(0.4555, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.358, Valid loss: 0.420\n",
      "Epoch 2 [save] loss = tensor(0.4199, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.350, Valid loss: 0.430\n",
      "Epoch 3 [----] loss = tensor(0.4301, device='cuda:0')\n",
      "Training time for each epoch: [733. 242. 192.] seconds\n",
      "Client 3 training ends with 1171 seconds in round 1\n",
      "38105/38105: [===============================>] - ETA 2.1ssss\n",
      "Training loss: 0.361, Valid loss: 0.374\n",
      "Epoch 1 [save] loss = tensor(0.3739, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.9sss\n",
      "Training loss: 0.352, Valid loss: 0.430\n",
      "Epoch 2 [----] loss = tensor(0.4299, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.343, Valid loss: 0.450\n",
      "Epoch 3 [----] loss = tensor(0.4501, device='cuda:0')\n",
      "Training time for each epoch: [730. 248. 190.] seconds\n",
      "Client 4 training ends with 1171 seconds in round 1\n",
      "38105/38105: [===============================>] - ETA 2.1ssss\n",
      "Training loss: 0.359, Valid loss: 0.399\n",
      "Epoch 1 [save] loss = tensor(0.3992, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.350, Valid loss: 0.396\n",
      "Epoch 2 [save] loss = tensor(0.3963, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.340, Valid loss: 0.415\n",
      "Epoch 3 [----] loss = tensor(0.4151, device='cuda:0')\n",
      "Training time for each epoch: [737. 264. 189.] seconds\n",
      "Client 5 training ends with 1193 seconds in round 1\n",
      "[[ Round 1 Ends]]\n",
      "38106/38106: [===============================>] - ETA 2.3ssss\n",
      "Training loss: 0.356, Valid loss: 0.405\n",
      "Epoch 1 [save] loss = tensor(0.4046, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.346, Valid loss: 0.461\n",
      "Epoch 2 [----] loss = tensor(0.4615, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.334, Valid loss: 0.395\n",
      "Epoch 3 [save] loss = tensor(0.3947, device='cuda:0')\n",
      "Training time for each epoch: [742. 274. 190.] seconds\n",
      "Client 1 training ends with 1209 seconds in round 2\n",
      "38106/38106: [===============================>] - ETA 2.2ssss\n",
      "Training loss: 0.354, Valid loss: 0.377\n",
      "Epoch 1 [save] loss = tensor(0.3767, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.342, Valid loss: 0.394\n",
      "Epoch 2 [----] loss = tensor(0.3944, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.330, Valid loss: 0.382\n",
      "Epoch 3 [----] loss = tensor(0.3823, device='cuda:0')\n",
      "Training time for each epoch: [740. 286. 189.] seconds\n",
      "Client 2 training ends with 1219 seconds in round 2\n",
      "38105/38105: [===============================>] - ETA 1.9ssss\n",
      "Training loss: 0.350, Valid loss: 0.395\n",
      "Epoch 1 [save] loss = tensor(0.3951, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 1.0sss\n",
      "Training loss: 0.336, Valid loss: 0.377\n",
      "Epoch 2 [save] loss = tensor(0.3773, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.321, Valid loss: 0.423\n",
      "Epoch 3 [----] loss = tensor(0.4230, device='cuda:0')\n",
      "Training time for each epoch: [726. 321. 246.] seconds\n",
      "Client 3 training ends with 1296 seconds in round 2\n",
      "38105/38105: [===============================>] - ETA 2.0ssss\n",
      "Training loss: 0.347, Valid loss: 0.375\n",
      "Epoch 1 [save] loss = tensor(0.3749, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.9sss\n",
      "Training loss: 0.330, Valid loss: 0.405\n",
      "Epoch 2 [----] loss = tensor(0.4052, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.311, Valid loss: 0.412\n",
      "Epoch 3 [----] loss = tensor(0.4116, device='cuda:0')\n",
      "Training time for each epoch: [733. 252. 199.] seconds\n",
      "Client 4 training ends with 1190 seconds in round 2\n",
      "38105/38105: [===============================>] - ETA 2.3ssss\n",
      "Training loss: 0.345, Valid loss: 0.380\n",
      "Epoch 1 [save] loss = tensor(0.3804, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 1.1sss\n",
      "Training loss: 0.327, Valid loss: 0.443\n",
      "Epoch 2 [----] loss = tensor(0.4432, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.304, Valid loss: 0.422\n",
      "Epoch 3 [----] loss = tensor(0.4216, device='cuda:0')\n",
      "Training time for each epoch: [717. 317. 200.] seconds\n",
      "Client 5 training ends with 1239 seconds in round 2\n",
      "[[ Round 2 Ends]]\n",
      "38106/38106: [===============================>] - ETA 2.0ssss\n",
      "Training loss: 0.343, Valid loss: 0.409\n",
      "Epoch 1 [save] loss = tensor(0.4087, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 1.0sss\n",
      "Training loss: 0.320, Valid loss: 0.493\n",
      "Epoch 2 [----] loss = tensor(0.4933, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.7sss\n",
      "Training loss: 0.293, Valid loss: 0.397\n",
      "Epoch 3 [save] loss = tensor(0.3971, device='cuda:0')\n",
      "Training time for each epoch: [685. 236. 195.] seconds\n",
      "Client 1 training ends with 1121 seconds in round 3\n",
      "38106/38106: [===============================>] - ETA 1.8ssss\n",
      "Training loss: 0.340, Valid loss: 0.404\n",
      "Epoch 1 [save] loss = tensor(0.4040, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.7sss\n",
      "Training loss: 0.314, Valid loss: 0.464\n",
      "Epoch 2 [----] loss = tensor(0.4644, device='cuda:0')\n",
      "38106/38106: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.284, Valid loss: 0.499\n",
      "Epoch 3 [----] loss = tensor(0.4991, device='cuda:0')\n",
      "Training time for each epoch: [707. 265. 193.] seconds\n",
      "Client 2 training ends with 1168 seconds in round 3\n",
      "38105/38105: [===============================>] - ETA 2.2ssss\n",
      "Training loss: 0.335, Valid loss: 0.438\n",
      "Epoch 1 [save] loss = tensor(0.4380, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.8sss\n",
      "Training loss: 0.304, Valid loss: 0.439\n",
      "Epoch 2 [----] loss = tensor(0.4395, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.269, Valid loss: 0.450\n",
      "Epoch 3 [----] loss = tensor(0.4504, device='cuda:0')\n",
      "Training time for each epoch: [743. 261. 192.] seconds\n",
      "Client 3 training ends with 1200 seconds in round 3\n",
      "38105/38105: [===============================>] - ETA 2.3ssss\n",
      "Training loss: 0.330, Valid loss: 0.403\n",
      "Epoch 1 [save] loss = tensor(0.4034, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.9sss\n",
      "Training loss: 0.294, Valid loss: 0.500\n",
      "Epoch 2 [----] loss = tensor(0.5001, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.252, Valid loss: 0.497\n",
      "Epoch 3 [----] loss = tensor(0.4973, device='cuda:0')\n",
      "Training time for each epoch: [750. 269. 194.] seconds\n",
      "Client 4 training ends with 1216 seconds in round 3\n",
      "38105/38105: [===============================>] - ETA 2.4ssss\n",
      "Training loss: 0.327, Valid loss: 0.397\n",
      "Epoch 1 [save] loss = tensor(0.3966, device='cuda:0')\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "38105/38105: [===============================>] - ETA 0.9sss\n",
      "Training loss: 0.285, Valid loss: 0.404\n",
      "Epoch 2 [----] loss = tensor(0.4040, device='cuda:0')\n",
      "38105/38105: [===============================>] - ETA 0.6sss\n",
      "Training loss: 0.241, Valid loss: 0.514\n",
      "Epoch 3 [----] loss = tensor(0.5143, device='cuda:0')\n",
      "Training time for each epoch: [747. 270. 193.] seconds\n",
      "Client 5 training ends with 1214 seconds in round 3\n",
      "[[ Round 3 Ends]]\n",
      "Model trained\n"
     ]
    }
   ],
   "source": [
    "dT = [datasetTrain_1, datasetTrain_2, datasetTrain_3, datasetTrain_4, datasetTrain_5]\n",
    "dLT = [dataLoaderTrain_1, dataLoaderTrain_2, dataLoaderTrain_3, dataLoaderTrain_4, dataLoaderTrain_5]\n",
    "\n",
    "model = DenseNet121(nnClassCount).cuda() # Step 0: Initialize global model and load the model\n",
    "model = torch.nn.DataParallel(model).cuda()\n",
    "\n",
    "for i in range(com_round):\n",
    "    print(\"[[[ Round {} Start ]]]\".format(i + 1))\n",
    "    params = [None] * num_clients\n",
    "    sel_clients = sorted(random.sample(range(num_clients), \n",
    "                                       round(num_clients*fraction))) # Step 1: select random fraction of clients\n",
    "    print(\"The number of clients:\", len(sel_clients))\n",
    "    for j in sel_clients: # Step 2: send weights to clients\n",
    "        print(\"<< Client {} Training Start >>\".format(j + 1))\n",
    "        train_valid_start = time.time()\n",
    "        params[j] = CheXpertTrainer.train(model, dLT[j], dataLoaderVal, # Step 3: Perform local computations\n",
    "                                          nnClassCount, trMaxEpoch = epoch, checkpoint = None)\n",
    "        train_valid_end = time.time()\n",
    "        client_time = round(train_valid_end - train_valid_start)\n",
    "        print(\"<< Client {} Training End: {} seconds elapsed >>\".format(j + 1, client_time))\n",
    "        \n",
    "    fidx = [idx for idx in range(len(params)) if params[idx] != None][0]\n",
    "    lidx = [idx for idx in range(len(params)) if params[idx] != None][-1]\n",
    "    for key in params[fidx]: # Step 4: return updates to server\n",
    "        weights, weightn = [], []\n",
    "        for k in sel_clients:\n",
    "            weights.append(params[k][key]*len(dT[k]))\n",
    "            weightn.append(len(dT[k]))\n",
    "        params[lidx][key] = sum(weights) / sum(weightn) # weighted averaging model weights\n",
    "\n",
    "    model = DenseNet121(nnClassCount).cuda()\n",
    "    model = torch.nn.DataParallel(model).cuda()\n",
    "    model.load_state_dict(params[lidx]) # Step 5: server updates global state\n",
    "    print(\"[[[ Round {} End ]]]\".format(i + 1))\n",
    "    \n",
    "print(\"Global model trained\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Test and ROC Curves"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "outGT, outPRED = CheXpertTrainer.test(model, dataLoaderTest, nnClassCount, None, class_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUROC mean  0.8348945604671515\n",
      "No Finding   0.8968205337690632\n",
      "Enlarged Cardiomediastinum   0.8036729134376027\n",
      "Cardiomegaly   0.8920946782178217\n",
      "Lung Opacity   0.7752718815513626\n",
      "Lung Lesion   0.8722817267121065\n",
      "Edema   0.8744698897370653\n",
      "Consolidation   0.7207095709570958\n",
      "Pneumonia   0.7781929911974755\n",
      "Atelectasis   0.7993981546065088\n",
      "Pneumothorax   0.9068198040360782\n",
      "Pleural Effusion   0.9133\n",
      "Pleural Other   0.8096667193176433\n",
      "Fracture   0.7679689830002981\n",
      "Support Devices   0.877856\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2160x720 with 14 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(nnClassCount):\n",
    "    fpr, tpr, threshold = metrics.roc_curve(outGT.cpu()[:,i], outPRED.cpu()[:,i])\n",
    "    roc_auc = metrics.auc(fpr, tpr)\n",
    "    f = plt.subplot(2, 7, i+1)\n",
    "\n",
    "    plt.title('ROC for: ' + class_names[i])\n",
    "    plt.plot(fpr, tpr, label = 'U-ones: AUC = %0.2f' % roc_auc)\n",
    "\n",
    "    plt.legend(loc = 'lower right')\n",
    "    plt.plot([0, 1], [0, 1],'r--')\n",
    "    plt.xlim([0, 1])\n",
    "    plt.ylim([0, 1])\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.xlabel('False Positive Rate')\n",
    "\n",
    "fig_size = plt.rcParams[\"figure.figsize\"]\n",
    "plt.rcParams[\"figure.figsize\"] = (30, 10)\n",
    "\n",
    "plt.savefig(\"ROC.png\", dpi = 1000)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Computational Stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Took about 20 min per client(3 epoch) for a single round.\n",
    "# Took about 60 min per clinet(3 epoch) for whole training."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}