Module9_Lab3.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Rushikasrithamaddula11/FMML-LABS-AND-MODULES/blob/main/Module9_Lab3.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Module 9: Convolutional Neural Networks\n",
        "## **Lab 3**\n",
        "### Module coordinator: Kushagra Agarwal"
      ],
      "metadata": {
        "id": "kCpbL40ggQf1"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## 1. Understanding Convolutions"
      ],
      "metadata": {
        "id": "0hAW8ptqVeyP"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "<img src=\"https://miro.medium.com/max/464/0*e-SMFTzO8r7skkpc\" width=650px/>"
      ],
      "metadata": {
        "id": "q6wfvhccKxWx"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "id": "yZD5S7IQgHbU"
      },
      "outputs": [],
      "source": [
        "import cv2\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Importing some pytorch packages\n",
        "import torch\n",
        "from torch.nn import Conv2d"
      ],
      "metadata": {
        "id": "BDE4WBHalreb"
      },
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Central to CNNs, a convolution operation is a linear element-wise multiplication operation between a small filter/kernel and same-sized patch from the image. We move this filter over the image like a sliding window from top left to bottom right. For each point on the image, a value is calculated based on the filter using a convolution operation. These filters can do simplest task like checking if there is a vertical line in the image or complicated task like detecting a human eye in the image.\n",
        "\n",
        "Let's look at the convolution formula:\n",
        "\n",
        "Convolution between image\n",
        "$f(x, y)$ and kernel $k(x, y)$ is\n",
        "$$f(x,y) * k(x,y) = \\sum \\limits _{i=0} ^{W-1} \\sum \\limits _{j=0} ^{H-1} f(i, j) k(x − i, y − j)$$\n",
        "\n",
        "where $W$ and $H$ are the the width and height of the image.\n",
        "\n",
        "The code demonstrates the convolution operation of a 2D matrix (image) with various filters"
      ],
      "metadata": {
        "id": "hbpRXyTpVv7u"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "<img src=\"https://www.researchgate.net/profile/Chaim-Baskin/publication/318849314/figure/fig1/AS:614287726870532@1523469015098/Image-convolution-with-an-input-image-of-size-7-7-and-a-filter-kernel-of-size-3-3.png\" alt=\"Convolution\" width=650px height=280px/>"
      ],
      "metadata": {
        "id": "amI6DTS0Ksvo"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# 2D 3x3 binary image with vertical edge\n",
        "image1 = np.array([[1,1,0], [1,1,0], [1,1,0]])\n",
        "\n",
        "# 2D 3x3 binary image with horizontal edge\n",
        "image2 = np.array([[0,0,0], [0,0,0], [1,1,1]])\n",
        "\n",
        "# On plotting the images\n",
        "plt.imshow(image1, cmap='gray', extent=[0, 3, 3, 0])\n",
        "plt.show()\n",
        "plt.imshow(image2, cmap='gray', extent=[0, 3, 3, 0])\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 853
        },
        "id": "IalqupPPkDil",
        "outputId": "c1b9476f-dd26-4719-d915-1c968033357d"
      },
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Vertical Line filter\n",
        "filter = np.array([[1,0,-1],\n",
        "                   [1,0,-1],\n",
        "                   [1,0,-1]])\n",
        "\n",
        "# Applying filter to first image\n",
        "output = np.sum(np.multiply(image1, filter))\n",
        "print('Output from first image: ', output)\n",
        "\n",
        "# Applying filter to second image\n",
        "output = np.sum(np.multiply(image2, filter))\n",
        "print('Output from second image: ', output)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "g42INjCaketK",
        "outputId": "c1820a84-cc7f-4edc-d542-0760fc299fe8"
      },
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Output from first image:  3\n",
            "Output from second image:  0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Horizontal edge filter\n",
        "filter = np.array([[-1,-1,-1],\n",
        "                   [ 0, 0, 0],\n",
        "                   [ 1, 1, 1]])\n",
        "\n",
        "output = np.sum(np.multiply(image1, filter))\n",
        "print('Output from first image: ', output)\n",
        "\n",
        "output = np.sum(np.multiply(image2, filter))\n",
        "print('Output from second image: ', output)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Tba3ySYUk2df",
        "outputId": "2cf34950-666e-417e-f13f-fd59541c2823"
      },
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Output from first image:  0\n",
            "Output from second image:  3\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Non-zero output suggests that there is a vertical edge present in the first image and not present in the second image. Similarly, horizontal edge is detected in second."
      ],
      "metadata": {
        "id": "BmYcPhDgk_in"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Let's define a function to use convolution layer from Pytorch and use our own kernel to detect edges in image"
      ],
      "metadata": {
        "id": "UNdrDtAKqyj2"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def apply_conv(image, kernel, padding=0, stride=1):\n",
        "\n",
        "  #--------IMAGE PREPROCESSING-------\n",
        "  image = torch.from_numpy(image)\n",
        "  # Pytorch requires input to convolution in (N,C,H,W), where N = batch size and C=#channels in input\n",
        "  input = image.view((1,1,image.shape[0], image.shape[1]))\n",
        "\n",
        "  # --------------KERNEL-------------\n",
        "  kernel = torch.from_numpy(kernel.astype(np.float32))\n",
        "\n",
        "  # Pytorch requires kernel of shape (N,C,H,W), where N = batch size and C=#channels in input\n",
        "  kernel = kernel.view((1,1,kernel.shape[0], kernel.shape[1]))\n",
        "\n",
        "  # ---------CONVOLUTION LAYER from Pytorch--------\n",
        "  conv = Conv2d(in_channels=1, out_channels=1, kernel_size=kernel.shape, padding=padding, stride=stride)\n",
        "\n",
        "  # Set the kernel weights in the convolution layer\n",
        "  conv.weight = torch.nn.Parameter(kernel)\n",
        "\n",
        "  # ---------APPLY CONVOLUTION--------\n",
        "  output = conv(input.float())\n",
        "  output_img = output.data.numpy()  # Tensor to back in numpy\n",
        "  output_img = output_img.reshape((-1, output_img.shape[-1])) # Reshape to 2D image\n",
        "\n",
        "  return output_img"
      ],
      "metadata": {
        "id": "G5fRJziBk3YB"
      },
      "execution_count": 25,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Our original lotus image\n",
        "image = cv2.imread('/content/grid1.jpg', 0)\n",
        "\n",
        "filter = np.array([[-1,-1,-1],\n",
        "                   [ 0, 0, 0],\n",
        "                   [ 1, 1, 1]])\n",
        "\n",
        "out1 = apply_conv(image, filter, padding=0, stride=1)\n",
        "\n",
        "filter = np.array([[1,0,-1],\n",
        "                   [1,0,-1],\n",
        "                   [1,0,-1]])\n",
        "\n",
        "out2 = apply_conv(image, filter, padding=0, stride=1)"
      ],
      "metadata": {
        "id": "1HPV6fFZloyc"
      },
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "fig = plt.figure(figsize=(10,6))\n",
        "ax = fig.add_subplot(1,3,1)\n",
        "ax.imshow(image, cmap='gray')\n",
        "ax.set_title('Original Image')\n",
        "ax = fig.add_subplot(1,3,2)\n",
        "ax.set_title('Horizontal edge')\n",
        "ax.imshow(out1, cmap='gray')\n",
        "ax = fig.add_subplot(1,3,3)\n",
        "ax.imshow(out2, cmap='gray')\n",
        "ax.set_title('Vertical edge')\n",
        "plt.show()\n"
      ],
      "metadata": {
        "id": "xgwXwbUKnmEr",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        },
        "outputId": "66e1b93d-4840-4bf0-c741-a08903ceeef6"
      },
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1000x600 with 3 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Pooling Layers\n",
        "\n",
        "Pooling layers are used to reduce the dimensions of the feature maps. Thus, it reduces the number of parameters to learn and the amount of computation performed in the network. The pooling layer summarises the features present in a region of the feature map generated by a convolution layer.\n",
        "\n",
        "1) Max Pooling:\n",
        "\n",
        "<img src='https://media.geeksforgeeks.org/wp-content/uploads/20190721025744/Screenshot-2019-07-21-at-2.57.13-AM.png' height=150px/>\n",
        "\n",
        "2) Average Pooling:\n",
        "\n",
        "<img src='https://media.geeksforgeeks.org/wp-content/uploads/20190721030705/Screenshot-2019-07-21-at-3.05.56-AM.png' height=150px/>"
      ],
      "metadata": {
        "id": "FpA0yEk1BgRb"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Softmax layer/activation\n",
        "Recall that logistic regression produces a decimal between 0 and 1.0. For example, a logistic regression output of 0.8 from an email classifier suggests an 80% chance of an email being spam and a 20% chance of it being not spam. Clearly, the sum of the probabilities of an email being either spam or not spam is 1.0.\n",
        "\n",
        "Softmax extends this idea into a multi-class world. That is, Softmax assigns decimal probabilities to each class in a multi-class problem. Those decimal probabilities must add up to 1.0. This additional constraint helps training converge more quickly than it otherwise would.\n",
        "Softmax is implemented through a neural network layer just before the output layer. The Softmax layer must have the same number of nodes as the output layer.\n",
        "\n",
        "<img src='https://miro.medium.com/max/1400/1*ReYpdIZ3ZSAPb2W8cJpkBg.jpeg' height=170px />"
      ],
      "metadata": {
        "id": "eu3QIU7AEO_x"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Learning to train a CNN network"
      ],
      "metadata": {
        "id": "P6grxC0TKKSF"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Import packages\n",
        "import torch\n",
        "import torchvision\n",
        "import torchvision.transforms as transforms\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import torch.optim as optim\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np"
      ],
      "metadata": {
        "id": "qlO-uZUHnn_-"
      },
      "execution_count": 28,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#  Images returned from torchvision dataset classes is in range [0,1]\n",
        "# We transform them to tensors and normalize them to range [-1,1] using 'Normalize' transform\n",
        "\n",
        "transform = transforms.Compose(\n",
        "    [transforms.ToTensor(),\n",
        "     transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])\n",
        "\n",
        "batch_size = 128\n",
        "\n",
        "trainset = torchvision.datasets.CIFAR10(root='./data', train=True,\n",
        "                                        download=True, transform=transform)\n",
        "trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size,\n",
        "                                          shuffle=True, num_workers=2)\n",
        "\n",
        "testset = torchvision.datasets.CIFAR10(root='./data', train=False,\n",
        "                                       download=True, transform=transform)\n",
        "testloader = torch.utils.data.DataLoader(testset, batch_size=batch_size,\n",
        "                                         shuffle=False, num_workers=2)\n",
        "\n",
        "# Classes in CIFAR10\n",
        "classes = ('plane', 'car', 'bird', 'cat',\n",
        "           'deer', 'dog', 'frog', 'horse', 'ship', 'truck')"
      ],
      "metadata": {
        "id": "NnezCUbwGqzd",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "3cb262b8-2966-423e-ea2d-ae2d4c053f57"
      },
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data/cifar-10-python.tar.gz\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "100%|██████████| 170498071/170498071 [00:02<00:00, 67554737.84it/s]\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Extracting ./data/cifar-10-python.tar.gz to ./data\n",
            "Files already downloaded and verified\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print('Training data shape : ', trainset.data.shape, len(trainset.targets))\n",
        "print('Testing data shape : ', testset.data.shape, len(testset.targets))\n",
        "\n",
        "# Find the unique numbers from the train labels\n",
        "nClasses = len(classes)\n",
        "print('Total number of outputs : ', nClasses)\n",
        "print('Output classes : ', classes)"
      ],
      "metadata": {
        "id": "e2M57DhHGupn",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a804b615-1693-461d-8aca-3e3aedf768e1"
      },
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Training data shape :  (50000, 32, 32, 3) 50000\n",
            "Testing data shape :  (10000, 32, 32, 3) 10000\n",
            "Total number of outputs :  10\n",
            "Output classes :  ('plane', 'car', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck')\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def train(num_epochs, model, train_loader, loss_func, optimizer):\n",
        "\n",
        "  # Training mode\n",
        "  model.train()\n",
        "\n",
        "  train_losses = []\n",
        "  train_acc = []\n",
        "\n",
        "  # Train the model\n",
        "  for epoch in range(num_epochs):\n",
        "    running_loss = 0\n",
        "    running_acc = 0\n",
        "    for i, (images, labels) in enumerate(train_loader):\n",
        "\n",
        "      # clear gradients for this training step\n",
        "      optimizer.zero_grad()\n",
        "\n",
        "      # Forward pass\n",
        "      output = model(images)\n",
        "\n",
        "      # Calculate loss\n",
        "      loss = loss_func(output, labels)\n",
        "\n",
        "      # Backpropagation, compute gradients\n",
        "      loss.backward()\n",
        "\n",
        "      # Apply gradients\n",
        "      optimizer.step()\n",
        "\n",
        "      # Running loss\n",
        "      running_loss += loss.item()\n",
        "\n",
        "      # indices of max probabilities\n",
        "      _, preds = torch.max(output, dim=1)\n",
        "\n",
        "      # Calculate number of correct predictions\n",
        "      correct = (preds.float() == labels).sum()\n",
        "      running_acc += correct\n",
        "\n",
        "      # Average loss and acc values\n",
        "      epoch_loss = running_loss / len(train_loader.dataset)\n",
        "      epoch_acc = running_acc / len(train_loader.dataset)\n",
        "\n",
        "    train_losses.append(epoch_loss)\n",
        "    train_acc.append(epoch_acc)\n",
        "    print ('Epoch {}/{}, Loss: {:.4f}, Accuracy: {:.4f}'.format(epoch + 1, num_epochs, epoch_loss, epoch_acc*100))\n",
        "\n",
        "  return train_losses, train_acc"
      ],
      "metadata": {
        "id": "_haw697lHCZs"
      },
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def test_model(model, testloader):\n",
        "  correct = 0\n",
        "  total = 0\n",
        "  # Deactivate autograd engine (don't compute grads since we're not training)\n",
        "  with torch.no_grad():\n",
        "    for data in testloader:\n",
        "        images, labels = data\n",
        "        # Calculate outputs by running images through the network\n",
        "        outputs = model(images)\n",
        "        # The class with the highest value is what we choose as prediction\n",
        "        _, predicted = torch.max(outputs.data, 1)\n",
        "        total += labels.size(0)\n",
        "        correct += (predicted == labels).sum().item()\n",
        "\n",
        "  print('Accuracy of the network: %d %%' % (\n",
        "      100 * correct / total))"
      ],
      "metadata": {
        "id": "x1Wi6vW7IHcR"
      },
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# CNN with 2 CONV layers and 3 FC layers\n",
        "class Net(nn.Module):\n",
        "    def __init__(self):\n",
        "        super().__init__()\n",
        "        self.conv1 = nn.Conv2d(3, 16, 3)\n",
        "        self.pool = nn.MaxPool2d(2, 2)\n",
        "        self.conv2 = nn.Conv2d(16, 32, 5)\n",
        "        self.fc1 = nn.Linear(32 * 5 * 5, 512)\n",
        "        self.fc2 = nn.Linear(512, 128)\n",
        "        # output layer 10 classes\n",
        "        self.fc3 = nn.Linear(128, 10)\n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.pool(F.relu(self.conv1(x)))\n",
        "        x = self.pool(F.relu(self.conv2(x)))\n",
        "        # flatten all dimensions except batch\n",
        "        x = torch.flatten(x, 1)\n",
        "        x = F.relu(self.fc1(x))\n",
        "        x = F.relu(self.fc2(x))\n",
        "        x = self.fc3(x)\n",
        "        return x"
      ],
      "metadata": {
        "id": "RgxbRadcHIms"
      },
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model = Net()\n",
        "print(model)"
      ],
      "metadata": {
        "id": "02meBxVOHLNL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4ab699f0-5695-4b1b-ff4e-4bda1fc75425"
      },
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Net(\n",
            "  (conv1): Conv2d(3, 16, kernel_size=(3, 3), stride=(1, 1))\n",
            "  (pool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "  (conv2): Conv2d(16, 32, kernel_size=(5, 5), stride=(1, 1))\n",
            "  (fc1): Linear(in_features=800, out_features=512, bias=True)\n",
            "  (fc2): Linear(in_features=512, out_features=128, bias=True)\n",
            "  (fc3): Linear(in_features=128, out_features=10, bias=True)\n",
            ")\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Cross Entropy loss for multi-class classification\n",
        "criterion = nn.CrossEntropyLoss()"
      ],
      "metadata": {
        "id": "lfKHypeYHNHO"
      },
      "execution_count": 35,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# SGD optimizer with momentum\n",
        "optimizer = optim.SGD(model.parameters(), lr=0.05, momentum=0.9)"
      ],
      "metadata": {
        "id": "MuDnJL28HPKP"
      },
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "num_epochs = 5  # iterations\n",
        "train_losses, train_acc = train(num_epochs, model, trainloader, criterion, optimizer)"
      ],
      "metadata": {
        "id": "AgKhwMrtHRCn",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "67ff71df-13a6-4e6e-ad09-d38bbaf0caa6"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/lib/python3.10/multiprocessing/popen_fork.py:66: RuntimeWarning: os.fork() was called. os.fork() is incompatible with multithreaded code, and JAX is multithreaded, so this will likely lead to a deadlock.\n",
            "  self.pid = os.fork()\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5, Loss: 0.0129, Accuracy: 39.4040\n",
            "Epoch 2/5, Loss: 0.0092, Accuracy: 57.8560\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "fig = plt.figure(figsize=(10,4))\n",
        "ax = fig.add_subplot(1,2, 1)\n",
        "ax.plot(np.arange(1,len(train_losses)+1),train_losses)\n",
        "plt.xlabel('Training loss')\n",
        "plt.ylabel('Epochs')\n",
        "ax.set_title('Loss vs Epochs')\n",
        "ax = fig.add_subplot(1,2, 2)\n",
        "ax.plot(np.arange(1,len(train_acc)+1),train_acc)\n",
        "plt.xlabel('Training accuracy')\n",
        "plt.ylabel('Epochs')\n",
        "ax.set_title('Accuracy vs Epochs')\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "tM2wHKGuHToB"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Accuracy on test data after training\n",
        "test_model(model, testloader)"
      ],
      "metadata": {
        "id": "3sHK9hhmI-VY"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Questions\n",
        "1) List some reasons why we should prefer CNN over ANN for image classification?\n",
        "\n",
        "2) Try improving the CNN performance further by tuning the hyperparameters(epochs, optimizer, LR etc). Report the improved test accuracy.\n",
        "\n",
        "3) What happens if you reduce the number of convolution layers to only 1?\n",
        "\n",
        "4) Why didn't we use the Softmax activation in the last layer of CNN?\n"
      ],
      "metadata": {
        "id": "RBQeCEB6REnH"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "1) CNNs (Convolutional Neural Networks) are preferred over ANNs (Artificial Neural Networks) for image classification due to several reasons:\n",
        "\n",
        "   a) **Feature Extraction**: CNNs are designed to automatically learn and extract relevant features from images through convolutional layers. This ability helps in capturing hierarchical patterns present in images, making them more suitable for image-related tasks compared to ANNs, which treat images as flat vectors and may struggle with capturing spatial information.\n",
        "   \n",
        "   b) **Parameter Sharing**: CNNs use parameter sharing in convolutional layers, where the same set of weights is applied across different parts of the input image. This reduces the number of parameters, making CNNs more efficient and effective for image data, which typically has high dimensionality.\n",
        "   \n",
        "   c) **Translation Invariance**: CNNs exhibit translation invariance, meaning they can recognize patterns regardless of their position in the image. This property is essential for tasks like object detection and recognition in images, where the position of the object may vary.\n",
        "   \n",
        "   d) **Sparse Connectivity**: CNNs have sparse connectivity, where each neuron is connected only to a small region of the input volume. This reduces the computational cost and makes CNNs more scalable to larger images compared to ANNs.\n",
        "   \n",
        "   e) **Pooling Layers**: CNNs often incorporate pooling layers, which further reduce the spatial dimensions of the input while retaining the most important features. Pooling helps in achieving translational invariance and controlling overfitting.\n",
        "\n",
        "2) To improve CNN performance, you can experiment with various hyperparameters such as:\n",
        "\n",
        "   a) **Number of Epochs**: Increasing the number of epochs allows the model to train for a longer duration, potentially improving performance. However, be cautious of overfitting.\n",
        "   \n",
        "   b) **Optimizer**: Try different optimizers such as Adam, RMSprop, or SGD with momentum. Each optimizer has its advantages and may perform differently depending on the dataset and model architecture.\n",
        "   \n",
        "   c) **Learning Rate (LR)**: Adjusting the learning rate can significantly impact the training process. Try different learning rates or incorporate learning rate schedules (e.g., learning rate decay) to find an optimal value.\n",
        "   \n",
        "   d) **Batch Size**: Experiment with different batch sizes to see how it affects training dynamics and convergence speed.\n",
        "   \n",
        "   e) **Data Augmentation**: Augmenting training data with techniques like rotation, flipping, and scaling can help improve generalization and reduce overfitting.\n",
        "   \n",
        "   After tuning these hyperparameters, evaluate the model on a separate test dataset to report the improved test accuracy.\n",
        "\n",
        "3) If you reduce the number of convolution layers to only 1 in a CNN, the model's ability to learn hierarchical features from the input images would be severely limited. Deep CNN architectures typically consist of multiple convolutional layers followed by pooling layers, allowing the model to learn complex patterns at different levels of abstraction. With only one convolutional layer, the model may struggle to capture meaningful features from the input images, leading to poorer performance in tasks such as image classification or object detection.\n",
        "\n",
        "4) The Softmax activation function is typically not used in the last layer of a CNN for image classification tasks because it is primarily used for multi-class classification problems where each input belongs to exactly one class. In image classification, however, the problem is often treated as multi-label classification, where an image may belong to multiple classes simultaneously (e.g., an image containing both a cat and a dog). Therefore, the last layer of a CNN for image classification typically uses a different activation function such as Sigmoid or ReLU to accommodate multi-label classification scenarios."
      ],
      "metadata": {
        "id": "CrDvYo4BwOfu"
      }
    }
  ]
}