07- Convolutional Neural Networks/Lab/Lab07 - Convolutional Neural Network.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "7e941aa0",
   "metadata": {},
   "source": [
    "# Convolutional Neural Network (ConvNet/CNN)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b189a7f",
   "metadata": {},
   "source": [
    "Convolutional Neural Networks (ConvNet/CNN) are Deep Learning models that can take an input image, assign relevance (learnable weights and biases) to various aspects/objects in the image, and distinguish between them. <br />\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ae1707f8",
   "metadata": {},
   "source": [
    "<center><img src=\"./_dependents/rgb.png\" width=\"500\" /></center>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7735f8a3",
   "metadata": {},
   "source": [
    "You often see images divided into three colour planes — Red, Green, and Blue — RGB. <br />\n",
    "But they can be stored in a variety of colour spaces, including Grayscale, RGB, HSV, CMYK, and others. <br />\n",
    "\n",
    "\n",
    "With the basic machine learning techniques, it is required to hand-engineering features and analyse the figures dimension by dimension. \n",
    "Imagine now taking a photo which exceeds 8K (7680x4320) dimensions and how computationally intensive it will be.\n",
    "The ConvNet's job is to compress the images into a format that is easier to process while preserving elements that are important for obtaining a decent prediction. This is critical for designing an architecture that is capable of learning features while also being scalable to large datasets. <br />\n",
    "So, while basic machine learning techniques require hand-engineering of features, ConvNets can learn these filters/characteristics through training.\n",
    "\n",
    "\n",
    "Below an example of a CNN architecture for classification. Starting from a picture, it analyses through its layers and output the correct class of object present in the image (a car in this case)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bd36dbc4",
   "metadata": {},
   "source": [
    "<center><img src=\"./_dependents/basic_architecture.png\" width=\"800\" /></center>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "57a47498",
   "metadata": {},
   "source": [
    "## Using PyTorch for creating a CNN architecture"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d0782bb3",
   "metadata": {},
   "source": [
    "In this experiment we will use the ```PyTorch``` library for creating our own CNN model.\n",
    "\n",
    "First of all, let us import the necessary libraries with the following commands:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "9c895e95",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "d:\\Anaconda3\\envs\\imperial\\lib\\site-packages\\tqdm\\auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "import torch # PyTorch is an open source machine learning framework.\n",
    "import torchvision # The torchvision package consists of popular datasets, model architectures, and common image transformations for computer vision.\n",
    "from torchvision import transforms\n",
    "from PIL import Image # PIL is a library to process images\n",
    "import numpy as np # package to work with numbers\n",
    "from sklearn.metrics import recall_score, precision_score, f1_score, accuracy_score, confusion_matrix\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "09ced771",
   "metadata": {},
   "source": [
    "We can create a ```device``` variable to check if a GPU is available.\n",
    "If the output of device variable is 'cuda', then it means that PyTorch will use a GPU for the analysis. Otherwise, CPU is used instead."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "d56bb9df",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GPU available: True\n",
      "NVIDIA GeForce RTX 3070\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "device(type='cuda')"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Device configuration\n",
    "print('GPU available: {}' .format(torch.cuda.is_available())) \n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "train_on_gpu = torch.cuda.is_available()\n",
    "\n",
    "device\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "737e41c8",
   "metadata": {},
   "source": [
    "## Load CIFAR10 Dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "99aced5d",
   "metadata": {},
   "source": [
    "We will now download the ```CIFAR10``` dataset that we will use for the laboratory and save it in a root folder path called 'data'. ```CIFAR10``` is a dataset used for image recognition that is commonly used for training image processing systems. The dataset is made up of 60000 32x32 colour images in 10 classes, with 6000 images per class. There are 50000 training images and 10000 test images.\n",
    "\n",
    "\n",
    "Additional information on the dataset can be found [here](https://www.cs.toronto.edu/~kriz/cifar.html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "55273fe8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n",
      "Dataset CIFAR10\n",
      "    Number of datapoints: 50000\n",
      "    Root location: dataset\n",
      "    Split: Train\n",
      "    StandardTransform\n",
      "Transform: Compose(\n",
      "               ToTensor()\n",
      "               Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5))\n",
      "           )\n",
      "------\n",
      "Dataset CIFAR10\n",
      "    Number of datapoints: 10000\n",
      "    Root location: dataset\n",
      "    Split: Test\n",
      "    StandardTransform\n",
      "Transform: ToTensor()\n"
     ]
    }
   ],
   "source": [
    "from torchvision import datasets\n",
    "from torchvision.transforms import ToTensor\n",
    "from torch.utils.data.sampler import SubsetRandomSampler\n",
    "\n",
    "# specify the image classes\n",
    "classes = ['airplane', 'automobile', 'bird', 'cat', 'deer',\n",
    "           'dog', 'frog', 'horse', 'ship', 'truck']\n",
    "\n",
    "# convert data to a normalized torch.FloatTensor\n",
    "transform = transforms.Compose([\n",
    "    transforms.ToTensor(),\n",
    "    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))\n",
    "    ])\n",
    "\n",
    "train_data = datasets.CIFAR10(\n",
    "    root = 'dataset', # path where the data will be downloaded\n",
    "    train = True,  # true if you want to download the training set, false for the testing set                        \n",
    "    transform = transform, # transform the data into a tensor that will be used later for analysis\n",
    "    download = True,            \n",
    ")\n",
    "\n",
    "test_data = datasets.CIFAR10(\n",
    "    root = 'dataset', \n",
    "    train = False, \n",
    "    transform = ToTensor()\n",
    ")\n",
    "\n",
    "print(train_data)\n",
    "print('------')\n",
    "print(test_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "178b1f1e",
   "metadata": {},
   "source": [
    "Let's examine the training data with a random index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "4cd02efd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import random\n",
    "\n",
    "idx = random.randint(0, train_data.data.shape[0])\n",
    "\n",
    "plt.imshow(train_data.data[idx], cmap='gray')\n",
    "plt.title('%s' % classes[train_data.targets[idx]])\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "08ec0c44",
   "metadata": {},
   "source": [
    "## Preparing data with DataLoader"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e37e6b57",
   "metadata": {},
   "source": [
    "The Dataset retrieves our dataset’s features and labels one sample at a time. While training a model, we want to pass samples in “minibatches”, reshuffle the data at every epoch to reduce model overfitting.\n",
    "\n",
    "```DataLoader``` is an iterable variable that we can use in PyTorch to automatically perform these functions.\n",
    "\n",
    "For additional information on ```DataLoader```, you can refer to the [official documentation](https://pytorch.org/docs/stable/data.html)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab06ce14",
   "metadata": {},
   "source": [
    "First let's split our training set into training and validation. Validation can be used for testing the scores of our model during training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "1909528a",
   "metadata": {},
   "outputs": [],
   "source": [
    "validation_size = 0.2\n",
    "\n",
    "# obtain training indices that will be used for validation\n",
    "num_train = len(train_data)\n",
    "indices = list(range(num_train))\n",
    "np.random.shuffle(indices)\n",
    "split = int(np.floor(validation_size * num_train))\n",
    "train_idx, valid_idx = indices[split:], indices[:split]\n",
    "\n",
    "# define samplers for obtaining training and validation batches\n",
    "train_sampler = SubsetRandomSampler(train_idx)\n",
    "valid_sampler = SubsetRandomSampler(valid_idx)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "96ffa426",
   "metadata": {},
   "source": [
    "Below we will load the dataset into DataLoader of batch_size=64 features and labels respectively. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "45585139",
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.utils.data import DataLoader\n",
    "\n",
    "batch_size=64\n",
    "shuffle_data =  True\n",
    "\n",
    "train_dataloader = DataLoader(train_data, \n",
    "                              batch_size=batch_size, \n",
    "                              sampler=train_sampler)\n",
    "\n",
    "val_dataloader = DataLoader(train_data, \n",
    "                              batch_size=batch_size, \n",
    "                              sampler=valid_sampler)\n",
    "\n",
    "test_dataloader = DataLoader(test_data, \n",
    "                             batch_size=batch_size,\n",
    "                             shuffle=shuffle_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0790c652",
   "metadata": {},
   "source": [
    "We can iterate over the DataLoader variable and extracting the features of the image and the corresponding label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "80197598",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1800x288 with 20 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# helper function to un-normalize and display an image\n",
    "def imshow(img):\n",
    "    img = img / 2 + 0.5  # unnormalize\n",
    "    plt.imshow(np.transpose(img, (1, 2, 0)))  # convert from Tensor image\n",
    "\n",
    "# obtain one batch of training images\n",
    "dataiter = iter(train_dataloader)\n",
    "images, labels = dataiter.next()  # if this line gives you error, change it into: images, labels = next(dataiter)\n",
    "images = images.numpy() # convert images to numpy for display\n",
    "# plot the images in the batch, along with the corresponding labels\n",
    "fig = plt.figure(figsize=(25, 4))\n",
    "# display 20 images\n",
    "for idx in np.arange(20):\n",
    "    ax = fig.add_subplot(2, 10, idx+1, xticks=[], yticks=[])\n",
    "    imshow(images[idx])\n",
    "    ax.set_title(classes[labels[idx]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "edb64153",
   "metadata": {},
   "outputs": [],
   "source": [
    "images.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0d84492d",
   "metadata": {},
   "source": [
    "## Create CNN Architecture"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ce1a41c4",
   "metadata": {},
   "source": [
    "We will now create the architecture for the CNN. In this example we will use two convolutional layers followed by three fully-connected layers. \n",
    "\n",
    "When creating the network, we also need to specify the activation function of each layer. We will use the Rectified Linear Units (ReLU) in this case.\n",
    "\n",
    "If necessary, we can also introduce a regularisation layer to reduce possible overfitting. A simple way to do this can be to use dropout layers.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9a333a13",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as F\n",
    "# define the CNN architecture\n",
    "\n",
    "class CNN(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(CNN, self).__init__()\n",
    "        self.conv1 = nn.Conv2d(3, 6, 5) #n_channels = 3, rgb image. These parameters can be modified and optimised based on the task and/or dataset\n",
    "        self.pool = nn.MaxPool2d(2, 2) # Applies 2D max-pooling to reduce the spatial dimensions of the input volume\n",
    "        self.conv2 = nn.Conv2d(6, 16, 5)\n",
    "        #self.conv2_drop = nn.Dropout2d()\n",
    "        self.fc1 = nn.Linear(16 * 5 * 5, 120) # Fully connected layers\n",
    "        self.fc2 = nn.Linear(120, 84) \n",
    "        self.fc3 = nn.Linear(84, 10)\n",
    "    \n",
    "    \n",
    "    # The forward pass defines the way we compute the output using the given layers and functions.\n",
    "    def forward(self, x):\n",
    "        x = self.pool(F.relu(self.conv1(x))) # activation function\n",
    "        x = self.pool(F.relu(self.conv2(x)))\n",
    "        #x = self.conv2_drop(x)\n",
    "        x = x.view(-1, 16 * 5 * 5)  # Flattens the output of a multi-dimensional volume \n",
    "                                    # such that we can apply fully connected layers to it\n",
    "        x = F.relu(self.fc1(x))\n",
    "        x = F.relu(self.fc2(x))\n",
    "        x = self.fc3(x)\n",
    "        return x\n",
    "\n",
    "# create the model CNN\n",
    "model = CNN()\n",
    "\n",
    "# move tensors to GPU if CUDA is available\n",
    "if train_on_gpu:\n",
    "    model.cuda()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "35352a16",
   "metadata": {},
   "source": [
    "We can print the architecture of the CNN network to investigate its layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ec17275c",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(model)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a339bda5",
   "metadata": {},
   "source": [
    "## Define Loss and Optimser "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a63fbd4f",
   "metadata": {},
   "source": [
    "**Loss**\n",
    "\n",
    "Deep neural network are trained via an algorithm called stochastic gradient descent optimization.\n",
    "It is necessary to repeatedly estimate the error for the model's current state as part of the optimiaation method. In order to update the weights and lower the loss on the subsequent evaluation, it is necessary to select an error function, also known as a loss function, that may be used to estimate the loss of the model.\n",
    "\n",
    "There exist multiple and different losses based on the scope of the analysis.\n",
    "In this experiment, we are interested in classifying the number present in the image. We are then performing a multi-class classification. Our goal is to predict the probability of the sample belonging to each known class (from number 0 to 9).\n",
    "\n",
    "We can then use a loss called ```Cross-entropy```. It will calculate a score to summarise the average difference between the actual and predicted probability distributions for all classes. To achieve the best results we have to minimise this loss\n",
    "\n",
    "\n",
    "\n",
    "**Optimiser**\n",
    "\n",
    "When training the model, we need to minimise the loss function and update the weights of the model at each epoch. \n",
    "In other words, we need to optimise the neural network.\n",
    "An optimiser is a function or an algorithm that modifies the attributes of the neural network, such as weights and learning rate. It works for reducing the loss and improve the overall accuracy.\n",
    "\n",
    "There are various types of optimiser. In our case, we will use ```Adam``` which is one of the most popular gradient descent optimisation algorithms.\n",
    "\n",
    "When setting up the optimiser, we also need to decide at which speed the model will learn. This is done via the learning rate.\n",
    "It will tell the model how fast or slow we will move towards the optimal weights\n",
    "<center><img src=\"./_dependents/learning_rate.png\" width=\"300\" /></center>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fc22f38d",
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = torch.nn.CrossEntropyLoss().to(device)  # cross-entropy loss for classification\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=0.001) \n",
    "\n",
    "optimizer"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "06996a06",
   "metadata": {},
   "source": [
    "## Train the model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b356d95b",
   "metadata": {},
   "source": [
    "We have finally set up all the parameters and we can move on to train the model on our dataset.\n",
    "We need to set the number of iterations that we want to train our model. This number needs to be carefully decided: if it is too low, then we risk of underfitting our model, if it is too high, we risk of overfitting it.\n",
    "<center><img src=\"./_dependents/overfitting_model.png\" width=\"300\" /></center>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "9553f769",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[INFO] TRAINING EPOCH 1\n",
      "[INFO] Epoch 1, Training Loss: 1.441293\n",
      "[INFO] VALIDATION EPOCH 1\n",
      "[INFO] Epoch 1, Validation Loss: 0.305656\n",
      "Validation loss decreased (inf --> 0.305656).  Saving model ...\n",
      "-----------------------------------------------------------\n",
      "[INFO] TRAINING EPOCH 2\n",
      "[INFO] Epoch 2, Training Loss: 1.208449\n",
      "[INFO] VALIDATION EPOCH 2\n",
      "[INFO] Epoch 2, Validation Loss: 0.274612\n",
      "Validation loss decreased (0.305656 --> 0.274612).  Saving model ...\n",
      "-----------------------------------------------------------\n",
      "[INFO] TRAINING EPOCH 3\n",
      "[INFO] Epoch 3, Training Loss: 1.118767\n",
      "[INFO] VALIDATION EPOCH 3\n",
      "[INFO] Epoch 3, Validation Loss: 0.255395\n",
      "Validation loss decreased (0.274612 --> 0.255395).  Saving model ...\n",
      "-----------------------------------------------------------\n",
      "[INFO] TRAINING EPOCH 4\n",
      "[INFO] Epoch 4, Training Loss: 1.047516\n",
      "[INFO] VALIDATION EPOCH 4\n",
      "[INFO] Epoch 4, Validation Loss: 0.234214\n",
      "Validation loss decreased (0.255395 --> 0.234214).  Saving model ...\n",
      "-----------------------------------------------------------\n",
      "[INFO] TRAINING EPOCH 5\n"
     ]
    }
   ],
   "source": [
    "# create the model CNN\n",
    "model = CNN()\n",
    "\n",
    "# move tensors to GPU if CUDA is available\n",
    "if train_on_gpu:\n",
    "    model.cuda()\n",
    "\n",
    "    \n",
    "criterion = torch.nn.CrossEntropyLoss().to(device)  # cross-entropy loss for classification\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=0.001) \n",
    "# number of epochs to train the model\n",
    "n_epochs = 5\n",
    "\n",
    "#List to store loss to visualize\n",
    "train_losslist = []\n",
    "valid_loss_min = np.Inf # track change in validation loss\n",
    "\n",
    "for epoch in range(1, n_epochs+1):\n",
    "\n",
    "    # keep track of training and validation loss\n",
    "    train_loss = 0.0\n",
    "    valid_loss = 0.0\n",
    "    \n",
    "    ###################\n",
    "    # train the model #\n",
    "    ###################\n",
    "    print('[INFO] TRAINING EPOCH {}'.format(epoch))\n",
    "    model.train()\n",
    "    for data, target in train_dataloader:\n",
    "        # move tensors to GPU if CUDA is available\n",
    "        # if train_on_gpu:\n",
    "        #     data, target = data.cuda(), target.cuda()\n",
    "        # clear the gradients of all optimized variables\n",
    "        optimizer.zero_grad()\n",
    "        # forward pass: compute predicted outputs by passing inputs to the model\n",
    "        output = model(data)\n",
    "        # calculate the batch loss\n",
    "        loss = criterion(output, target)\n",
    "        # backward pass: compute gradient of the loss with respect to model parameters\n",
    "        loss.backward()\n",
    "        # perform a single optimization step (parameter update)\n",
    "        optimizer.step()\n",
    "        # update training loss\n",
    "        train_loss += loss.item()*data.size(0)\n",
    "    \n",
    "    train_loss = train_loss/len(train_dataloader.dataset)\n",
    "    train_losslist.append(train_loss)\n",
    "    print('[INFO] Epoch {}, Training Loss: {:.6f}'.format(epoch, train_loss))\n",
    "        \n",
    "    ######################    \n",
    "    # validate the model #\n",
    "    ######################\n",
    "    print('[INFO] VALIDATION EPOCH {}'.format(epoch))\n",
    "    model.eval()\n",
    "    for data, target in val_dataloader:\n",
    "        # move tensors to GPU if CUDA is available\n",
    "        # if train_on_gpu:\n",
    "        #     data, target = data.cuda(), target.cuda()\n",
    "        # forward pass: compute predicted outputs by passing inputs to the model\n",
    "        output = model(data)\n",
    "        # calculate the batch loss\n",
    "        loss = criterion(output, target)\n",
    "        # update average validation loss \n",
    "        valid_loss += loss.item()*data.size(0)\n",
    "    \n",
    "    # calculate average losses\n",
    "    valid_loss = valid_loss/len(val_dataloader.dataset)\n",
    "    print('[INFO] Epoch {}, Validation Loss: {:.6f}'.format(epoch, valid_loss))\n",
    "\n",
    "    # save model if validation loss has decreased\n",
    "    if valid_loss <= valid_loss_min:\n",
    "        print('Validation loss decreased ({:.6f} --> {:.6f}).  Saving model ...'.format(\n",
    "        valid_loss_min,\n",
    "        valid_loss))\n",
    "        torch.save(model.state_dict(), 'model_cifar.pt')\n",
    "        valid_loss_min = valid_loss\n",
    "    print('-----------------------------------------------------------')    \n",
    "\n",
    "plt.plot([*range(n_epochs)], train_losslist)\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.title(\"Performance of Model 1\")\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d6c6d07f",
   "metadata": {},
   "source": [
    "## Test the model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9b903483",
   "metadata": {},
   "source": [
    "We will now load the model which achieved the best results during training and we will test it on the testing set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "83a816ae",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.load_state_dict(torch.load('model_cifar.pt'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c371586d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# track test loss\n",
    "test_loss = 0.0\n",
    "class_correct = list(0. for i in range(10))\n",
    "class_total = list(0. for i in range(10))\n",
    "\n",
    "total_test_predicted_labels = []\n",
    "total_test_true_labels = []\n",
    "\n",
    "model.eval()\n",
    "# iterate over test data\n",
    "for data, target in test_dataloader:\n",
    "    # move tensors to GPU if CUDA is available\n",
    "    if train_on_gpu:\n",
    "        data, target = data.cuda(), target.cuda()\n",
    "    # forward pass: compute predicted outputs by passing inputs to the model\n",
    "    output = model(data)\n",
    "    # calculate the batch loss\n",
    "    loss = criterion(output, target)\n",
    "    # update test loss \n",
    "    test_loss += loss.item()*data.size(0)\n",
    "    # convert output probabilities to predicted class\n",
    "    _, pred = torch.max(output, 1)    \n",
    "    # compare predictions to true label\n",
    "    correct_tensor = pred.eq(target.data.view_as(pred))\n",
    "    correct = np.squeeze(correct_tensor.numpy()) if not train_on_gpu else np.squeeze(correct_tensor.cpu().numpy())\n",
    "    \n",
    "    total_test_true_labels.append(target.cpu().numpy())\n",
    "    total_test_predicted_labels.append(pred.cpu().numpy())\n",
    "\n",
    "    # calculate test accuracy for each object class\n",
    "    for i in range(target.shape[0]):\n",
    "        label = target.data[i]\n",
    "        class_correct[label] += correct[i].item()\n",
    "        class_total[label] += 1\n",
    "        \n",
    "        \n",
    "y_true = np.asarray([item for sublist in total_test_true_labels for item\n",
    "                             in sublist])\n",
    "y_pred = np.asarray([item for sublist in total_test_predicted_labels for\n",
    "                               item in sublist])        \n",
    "\n",
    "# average test loss\n",
    "test_loss = test_loss/len(test_dataloader.dataset)\n",
    "print('Test Loss: {:.6f}\\n'.format(test_loss))\n",
    "\n",
    "for i in range(10):\n",
    "    if class_total[i] > 0:\n",
    "        print('Test Accuracy of %5s: %2d%% (%2d/%2d)' % (\n",
    "            classes[i], 100 * class_correct[i] / class_total[i],\n",
    "            np.sum(class_correct[i]), np.sum(class_total[i])))\n",
    "    else:\n",
    "        print('Test Accuracy of %5s: N/A (no training examples)' % (classes[i]))\n",
    "\n",
    "print('\\nTest Accuracy (Overall): %2d%% (%2d/%2d)' % (\n",
    "    100. * np.sum(class_correct) / np.sum(class_total),\n",
    "    np.sum(class_correct), np.sum(class_total)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c15bc843",
   "metadata": {},
   "source": [
    "## Investigate results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "71ac51b1",
   "metadata": {},
   "outputs": [],
   "source": [
    "accuracy = accuracy_score(y_true, y_pred)\n",
    "recall = recall_score(y_true, y_pred, average='weighted')\n",
    "precision = precision_score(y_true, y_pred, average='weighted')\n",
    "f1 = f1_score(y_true, y_pred, average='weighted')\n",
    "cf = confusion_matrix(y_true, y_pred)\n",
    "# cf_norm = confusion_matrix(y_true, y_pred, normalize='true') * 100\n",
    "\n",
    "print('ACCURACY: {:.2f}\\n RECALL: {:.2f}\\n PRECISION: {:.2f}\\n F1: {:.2f}\\n  '.format(accuracy, \n",
    "                                                                                      recall,\n",
    "                                                                                      precision, \n",
    "                                                                                      f1))\n",
    "print(cf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6da03e98",
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns\n",
    "fig, ax = plt.subplots(figsize=(15,15))\n",
    "sns.heatmap(cf, annot=True, ax = ax, cmap='Blues', fmt='g')\n",
    "ax.set_xticklabels(classes)\n",
    "ax.set_yticklabels(classes)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "288d01e6",
   "metadata": {},
   "source": [
    "# Useful Resources"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7669b971",
   "metadata": {},
   "source": [
    "- Documentation for the Convolutional Layer - https://pytorch.org/docs/stable/generated/torch.nn.Conv2d.html \n",
    "- Guide and discussion on optimisers - https://medium.com/mlearning-ai/optimizers-in-deep-learning-7bf81fed78a0\n",
    "- CIFAR10 Tutorial - https://pytorch.org/tutorials/beginner/blitz/cifar10_tutorial.html\n",
    "- CNN implementation on CIFAR10 - https://shonit2096.medium.com/cnn-on-cifar10-data-set-using-pytorch-34be87e09844"
   ]
  }
 ],
 "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": "3.9.6 (default, Oct 18 2022, 12:41:40) \n[Clang 14.0.0 (clang-1400.0.29.202)]"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "165px"
   },
   "toc_section_display": true,
   "toc_window_display": true
  },
  "vscode": {
   "interpreter": {
    "hash": "31f2aee4e71d21fbe5cf8b01ff0e069b9275f58929596ceb00d14d90e3e16cd6"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}