02. ConvolutionalNeuralNetworks/Convolutional Layers II.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Convolutional Layers II.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "**INITIALIZATION:**"
      ],
      "metadata": {
        "id": "Qm_mbO5veN3H"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@ GETTING LIBRARIES AND DEPENDENCIES: UNCOMMENT BELOW: \n",
        "# !python /content/simplepreprocessor.py\n",
        "# !unzip /content/drive/MyDrive/Data/animals.zip"
      ],
      "metadata": {
        "id": "fQq2yTvVxHSQ"
      },
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "id": "TZWcLw4laYpn"
      },
      "outputs": [],
      "source": [
        "#@ DOWNLOADING NECESSARY LIBRARIES AND DEPENDENCIES: \n",
        "import os\n",
        "from imutils import paths\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "from sklearn.preprocessing import LabelBinarizer\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import classification_report\n",
        "\n",
        "from keras.models import Sequential\n",
        "from keras.layers.convolutional import Conv2D\n",
        "from keras.layers.core import Activation\n",
        "from keras.layers.core import Flatten\n",
        "from keras.layers.core import Dense\n",
        "from keras import backend as K\n",
        "from tensorflow.keras.optimizers import SGD\n",
        "from keras.preprocessing.image import img_to_array\n",
        "from keras.datasets import cifar10\n",
        "\n",
        "from simplepreprocessor import SimplePreprocessor\n",
        "from simplepreprocessor import SimpleDatasetLoader"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**SHALLOW NEURAL NETWORK:**"
      ],
      "metadata": {
        "id": "Io5UhABgh3ae"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@ DEFINING IMAGE TO ARRAY PREPROCESSOR:\n",
        "class ImageToArrayPreprocessor:                             # Defining Image To Preprocessor Class. \n",
        "    def __init__(self, dataFormat=None):                    # Initializing Constructor Function. \n",
        "        self.dataFormat = dataFormat                        # Initialization. \n",
        "    \n",
        "    def preprocess(self, image):                            # Initializing Preprocess Method. \n",
        "        return img_to_array(image,\n",
        "                            data_format=self.dataFormat)    # Converting Image2Array. "
      ],
      "metadata": {
        "id": "vOurqlOqsphK"
      },
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@ DEFINING SHALLOWNET ARCHITECTURE:\n",
        "class ShallowNet:                                           # Defining Shallow Network. \n",
        "    @staticmethod\n",
        "    def build(width, height, depth, classes):               # Initializing Build Method. \n",
        "        model = Sequential()                                # Initializing Sequential Model. \n",
        "        inputShape = (height, width, depth)                 # Initializing Input Shape.\n",
        "        if K.image_data_format() == \"channels_first\":\n",
        "            inputShape = (depth, height, width)             # Initializing Input Shape Format. \n",
        "        model.add(Conv2D(32, (3, 3), padding=\"same\", \n",
        "                         input_shape = inputShape))         # Initializing Convolutional Layer. \n",
        "        model.add(Flatten())                                # Converting into 1D List. \n",
        "        model.add(Dense(classes))                           # Adding Dense Output Layer. \n",
        "        model.add(Activation(\"softmax\"))                    # Initializing Softmax Activation Function. \n",
        "        return model"
      ],
      "metadata": {
        "id": "WVviatxcx9Jk"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "**SHALLOWNET ON ANIMALS:**"
      ],
      "metadata": {
        "id": "unansEf4iAe9"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@ GETTING THE DATASET: \n",
        "sp = SimplePreprocessor(32, 32)                             # Resizing Images to Fixed Size. \n",
        "iap = ImageToArrayPreprocessor()                            # Initializing Channel Ordering. \n",
        "imagePaths = list(paths.list_images('/content/animals'))    # Initializing Image Dataset.\n",
        "sdl = SimpleDatasetLoader(preprocessors=[sp, iap])          # Initializing Dataset Loader. \n",
        "(data, labels) = sdl.load(imagePaths, verbose=500)          # Initializing Dataset. \n",
        "data = data.astype(\"float\") / 255.0                         # Normalizing Images. \n",
        "\n",
        "#@ PREPARING THE DATASET:\n",
        "(trainX, testX, trainY, testY) = train_test_split(\n",
        "    data, labels, test_size=0.25, random_state=42)          # Partitioning Training Data and Test Data. \n",
        "trainY = LabelBinarizer().fit_transform(trainY)             # Converting Labels into Vectors. \n",
        "testY = LabelBinarizer().fit_transform(testY)               # Converting Labels into Vectors. "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kKyVFZ7_hxNg",
        "outputId": "86ab8241-a56d-4179-efaa-87be35db37f2"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[INFO] processed 500/3000\n",
            "[INFO] processed 1000/3000\n",
            "[INFO] processed 1500/3000\n",
            "[INFO] processed 2000/3000\n",
            "[INFO] processed 2500/3000\n",
            "[INFO] processed 3000/3000\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@ INITIALIZING OPTIMIZER AND MODEL: \n",
        "opt = SGD(0.005)                                                  # Initializing SGD Optimizer. \n",
        "model = ShallowNet.build(width=32,height=32,depth=3, classes=3)   # Initializing Shallow Neural Network. \n",
        "model.compile(loss=\"categorical_crossentropy\",optimizer=opt,\n",
        "              metrics=[\"accuracy\"])                               # Compiling ShallowNet Model. \n",
        "H = model.fit(trainX,trainY, validation_data=(testX,testY), \n",
        "              batch_size=32, epochs=100, verbose=1)               # Training ShallowNet Model. "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_5caxdTt5uIK",
        "outputId": "8fed262d-4cd0-478f-9262-ee9fe6df2858"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "71/71 [==============================] - 14s 11ms/step - loss: 1.0239 - accuracy: 0.4836 - val_loss: 1.2221 - val_accuracy: 0.4987\n",
            "Epoch 2/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.9223 - accuracy: 0.5307 - val_loss: 0.8796 - val_accuracy: 0.5587\n",
            "Epoch 3/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.8668 - accuracy: 0.5720 - val_loss: 0.8817 - val_accuracy: 0.5307\n",
            "Epoch 4/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.8464 - accuracy: 0.5889 - val_loss: 0.9042 - val_accuracy: 0.5080\n",
            "Epoch 5/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.8243 - accuracy: 0.5920 - val_loss: 0.9071 - val_accuracy: 0.5240\n",
            "Epoch 6/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.8048 - accuracy: 0.6173 - val_loss: 0.8283 - val_accuracy: 0.5947\n",
            "Epoch 7/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.7911 - accuracy: 0.6236 - val_loss: 0.8387 - val_accuracy: 0.5653\n",
            "Epoch 8/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.7924 - accuracy: 0.6156 - val_loss: 0.9241 - val_accuracy: 0.5440\n",
            "Epoch 9/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.7772 - accuracy: 0.6378 - val_loss: 0.8369 - val_accuracy: 0.5827\n",
            "Epoch 10/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.7766 - accuracy: 0.6324 - val_loss: 0.8910 - val_accuracy: 0.5453\n",
            "Epoch 11/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.7680 - accuracy: 0.6467 - val_loss: 0.8182 - val_accuracy: 0.5973\n",
            "Epoch 12/100\n",
            "71/71 [==============================] - 0s 6ms/step - loss: 0.7576 - accuracy: 0.6449 - val_loss: 0.8223 - val_accuracy: 0.5947\n",
            "Epoch 13/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.7566 - accuracy: 0.6556 - val_loss: 0.8097 - val_accuracy: 0.5973\n",
            "Epoch 14/100\n",
            "71/71 [==============================] - 0s 6ms/step - loss: 0.7487 - accuracy: 0.6538 - val_loss: 0.8751 - val_accuracy: 0.5800\n",
            "Epoch 15/100\n",
            "71/71 [==============================] - 1s 9ms/step - loss: 0.7386 - accuracy: 0.6636 - val_loss: 0.8173 - val_accuracy: 0.6000\n",
            "Epoch 16/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.7359 - accuracy: 0.6698 - val_loss: 0.8593 - val_accuracy: 0.5600\n",
            "Epoch 17/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.7373 - accuracy: 0.6680 - val_loss: 0.8862 - val_accuracy: 0.5733\n",
            "Epoch 18/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.7259 - accuracy: 0.6724 - val_loss: 0.8084 - val_accuracy: 0.6013\n",
            "Epoch 19/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.7219 - accuracy: 0.6693 - val_loss: 0.8303 - val_accuracy: 0.5987\n",
            "Epoch 20/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.7168 - accuracy: 0.6764 - val_loss: 0.8223 - val_accuracy: 0.6053\n",
            "Epoch 21/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.7122 - accuracy: 0.6764 - val_loss: 0.8286 - val_accuracy: 0.5947\n",
            "Epoch 22/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.7061 - accuracy: 0.6916 - val_loss: 0.8479 - val_accuracy: 0.5640\n",
            "Epoch 23/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6991 - accuracy: 0.6907 - val_loss: 0.8241 - val_accuracy: 0.6040\n",
            "Epoch 24/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6995 - accuracy: 0.6898 - val_loss: 0.8328 - val_accuracy: 0.6000\n",
            "Epoch 25/100\n",
            "71/71 [==============================] - 0s 6ms/step - loss: 0.6928 - accuracy: 0.6933 - val_loss: 0.8374 - val_accuracy: 0.5880\n",
            "Epoch 26/100\n",
            "71/71 [==============================] - 1s 9ms/step - loss: 0.6925 - accuracy: 0.6951 - val_loss: 0.8733 - val_accuracy: 0.5787\n",
            "Epoch 27/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6904 - accuracy: 0.7018 - val_loss: 0.8256 - val_accuracy: 0.5987\n",
            "Epoch 28/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6880 - accuracy: 0.7031 - val_loss: 0.8506 - val_accuracy: 0.5760\n",
            "Epoch 29/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.6861 - accuracy: 0.7067 - val_loss: 0.8232 - val_accuracy: 0.5933\n",
            "Epoch 30/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6746 - accuracy: 0.7040 - val_loss: 0.8345 - val_accuracy: 0.5840\n",
            "Epoch 31/100\n",
            "71/71 [==============================] - 0s 6ms/step - loss: 0.6686 - accuracy: 0.7129 - val_loss: 0.8493 - val_accuracy: 0.5947\n",
            "Epoch 32/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6692 - accuracy: 0.7076 - val_loss: 0.8380 - val_accuracy: 0.5893\n",
            "Epoch 33/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6640 - accuracy: 0.7182 - val_loss: 0.8399 - val_accuracy: 0.5840\n",
            "Epoch 34/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6600 - accuracy: 0.7151 - val_loss: 0.8380 - val_accuracy: 0.5880\n",
            "Epoch 35/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6573 - accuracy: 0.7124 - val_loss: 0.8292 - val_accuracy: 0.5973\n",
            "Epoch 36/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6563 - accuracy: 0.7156 - val_loss: 0.9098 - val_accuracy: 0.5773\n",
            "Epoch 37/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6548 - accuracy: 0.7124 - val_loss: 0.8326 - val_accuracy: 0.5880\n",
            "Epoch 38/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6475 - accuracy: 0.7231 - val_loss: 0.8377 - val_accuracy: 0.5907\n",
            "Epoch 39/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6463 - accuracy: 0.7302 - val_loss: 0.8572 - val_accuracy: 0.5680\n",
            "Epoch 40/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.6428 - accuracy: 0.7222 - val_loss: 0.8464 - val_accuracy: 0.5960\n",
            "Epoch 41/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6393 - accuracy: 0.7271 - val_loss: 0.8594 - val_accuracy: 0.5960\n",
            "Epoch 42/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6379 - accuracy: 0.7298 - val_loss: 0.8879 - val_accuracy: 0.5573\n",
            "Epoch 43/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6335 - accuracy: 0.7329 - val_loss: 0.8518 - val_accuracy: 0.5880\n",
            "Epoch 44/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.6337 - accuracy: 0.7302 - val_loss: 0.8470 - val_accuracy: 0.5867\n",
            "Epoch 45/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6267 - accuracy: 0.7284 - val_loss: 0.9210 - val_accuracy: 0.5720\n",
            "Epoch 46/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6243 - accuracy: 0.7289 - val_loss: 0.9157 - val_accuracy: 0.5747\n",
            "Epoch 47/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6209 - accuracy: 0.7382 - val_loss: 0.8630 - val_accuracy: 0.5667\n",
            "Epoch 48/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6204 - accuracy: 0.7396 - val_loss: 0.8507 - val_accuracy: 0.5907\n",
            "Epoch 49/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6138 - accuracy: 0.7418 - val_loss: 0.8801 - val_accuracy: 0.5613\n",
            "Epoch 50/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.6117 - accuracy: 0.7413 - val_loss: 0.8537 - val_accuracy: 0.5973\n",
            "Epoch 51/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6063 - accuracy: 0.7453 - val_loss: 0.9375 - val_accuracy: 0.5560\n",
            "Epoch 52/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.6066 - accuracy: 0.7533 - val_loss: 0.9024 - val_accuracy: 0.5880\n",
            "Epoch 53/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6087 - accuracy: 0.7364 - val_loss: 0.8814 - val_accuracy: 0.5693\n",
            "Epoch 54/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.6026 - accuracy: 0.7533 - val_loss: 0.9289 - val_accuracy: 0.5747\n",
            "Epoch 55/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5990 - accuracy: 0.7444 - val_loss: 0.8703 - val_accuracy: 0.5933\n",
            "Epoch 56/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5973 - accuracy: 0.7502 - val_loss: 0.8697 - val_accuracy: 0.5680\n",
            "Epoch 57/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5933 - accuracy: 0.7533 - val_loss: 0.8942 - val_accuracy: 0.5573\n",
            "Epoch 58/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5879 - accuracy: 0.7573 - val_loss: 0.8869 - val_accuracy: 0.5667\n",
            "Epoch 59/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5867 - accuracy: 0.7520 - val_loss: 0.8753 - val_accuracy: 0.5733\n",
            "Epoch 60/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5818 - accuracy: 0.7600 - val_loss: 0.8779 - val_accuracy: 0.5640\n",
            "Epoch 61/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5835 - accuracy: 0.7582 - val_loss: 0.8781 - val_accuracy: 0.5693\n",
            "Epoch 62/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5804 - accuracy: 0.7560 - val_loss: 0.8855 - val_accuracy: 0.5827\n",
            "Epoch 63/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5792 - accuracy: 0.7609 - val_loss: 0.8888 - val_accuracy: 0.5667\n",
            "Epoch 64/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5771 - accuracy: 0.7618 - val_loss: 0.9107 - val_accuracy: 0.5587\n",
            "Epoch 65/100\n",
            "71/71 [==============================] - 0s 6ms/step - loss: 0.5720 - accuracy: 0.7604 - val_loss: 0.8875 - val_accuracy: 0.5640\n",
            "Epoch 66/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.5699 - accuracy: 0.7653 - val_loss: 0.8953 - val_accuracy: 0.5613\n",
            "Epoch 67/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5687 - accuracy: 0.7676 - val_loss: 0.8932 - val_accuracy: 0.5587\n",
            "Epoch 68/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5651 - accuracy: 0.7649 - val_loss: 0.8923 - val_accuracy: 0.5613\n",
            "Epoch 69/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5625 - accuracy: 0.7711 - val_loss: 0.9104 - val_accuracy: 0.5760\n",
            "Epoch 70/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5603 - accuracy: 0.7711 - val_loss: 0.9340 - val_accuracy: 0.5520\n",
            "Epoch 71/100\n",
            "71/71 [==============================] - 0s 6ms/step - loss: 0.5567 - accuracy: 0.7751 - val_loss: 0.9065 - val_accuracy: 0.5707\n",
            "Epoch 72/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5565 - accuracy: 0.7764 - val_loss: 0.8991 - val_accuracy: 0.5693\n",
            "Epoch 73/100\n",
            "71/71 [==============================] - 1s 9ms/step - loss: 0.5525 - accuracy: 0.7684 - val_loss: 0.9645 - val_accuracy: 0.5760\n",
            "Epoch 74/100\n",
            "71/71 [==============================] - 1s 9ms/step - loss: 0.5526 - accuracy: 0.7733 - val_loss: 0.9296 - val_accuracy: 0.5520\n",
            "Epoch 75/100\n",
            "71/71 [==============================] - 0s 6ms/step - loss: 0.5502 - accuracy: 0.7769 - val_loss: 0.9105 - val_accuracy: 0.5600\n",
            "Epoch 76/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5469 - accuracy: 0.7720 - val_loss: 0.9200 - val_accuracy: 0.5640\n",
            "Epoch 77/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5474 - accuracy: 0.7813 - val_loss: 0.9186 - val_accuracy: 0.5640\n",
            "Epoch 78/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5429 - accuracy: 0.7840 - val_loss: 0.9404 - val_accuracy: 0.5520\n",
            "Epoch 79/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5421 - accuracy: 0.7804 - val_loss: 0.9229 - val_accuracy: 0.5653\n",
            "Epoch 80/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.5408 - accuracy: 0.7916 - val_loss: 0.9176 - val_accuracy: 0.5653\n",
            "Epoch 81/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5363 - accuracy: 0.7840 - val_loss: 0.9699 - val_accuracy: 0.5467\n",
            "Epoch 82/100\n",
            "71/71 [==============================] - 0s 6ms/step - loss: 0.5338 - accuracy: 0.7867 - val_loss: 0.9244 - val_accuracy: 0.5640\n",
            "Epoch 83/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5304 - accuracy: 0.7862 - val_loss: 0.9315 - val_accuracy: 0.5680\n",
            "Epoch 84/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5300 - accuracy: 0.7938 - val_loss: 0.9277 - val_accuracy: 0.5693\n",
            "Epoch 85/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.5274 - accuracy: 0.7907 - val_loss: 0.9418 - val_accuracy: 0.5507\n",
            "Epoch 86/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5258 - accuracy: 0.7889 - val_loss: 0.9289 - val_accuracy: 0.5707\n",
            "Epoch 87/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5254 - accuracy: 0.7898 - val_loss: 0.9316 - val_accuracy: 0.5693\n",
            "Epoch 88/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5222 - accuracy: 0.7960 - val_loss: 0.9354 - val_accuracy: 0.5627\n",
            "Epoch 89/100\n",
            "71/71 [==============================] - 1s 9ms/step - loss: 0.5165 - accuracy: 0.7978 - val_loss: 0.9999 - val_accuracy: 0.5587\n",
            "Epoch 90/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.5164 - accuracy: 0.8049 - val_loss: 0.9522 - val_accuracy: 0.5600\n",
            "Epoch 91/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5150 - accuracy: 0.8076 - val_loss: 0.9555 - val_accuracy: 0.5640\n",
            "Epoch 92/100\n",
            "71/71 [==============================] - 1s 7ms/step - loss: 0.5135 - accuracy: 0.8031 - val_loss: 0.9533 - val_accuracy: 0.5653\n",
            "Epoch 93/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5123 - accuracy: 0.8053 - val_loss: 0.9534 - val_accuracy: 0.5653\n",
            "Epoch 94/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5075 - accuracy: 0.8000 - val_loss: 0.9573 - val_accuracy: 0.5653\n",
            "Epoch 95/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5064 - accuracy: 0.8031 - val_loss: 0.9625 - val_accuracy: 0.5653\n",
            "Epoch 96/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5081 - accuracy: 0.8022 - val_loss: 0.9558 - val_accuracy: 0.5600\n",
            "Epoch 97/100\n",
            "71/71 [==============================] - 1s 8ms/step - loss: 0.5040 - accuracy: 0.8053 - val_loss: 1.0013 - val_accuracy: 0.5507\n",
            "Epoch 98/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.5023 - accuracy: 0.8084 - val_loss: 0.9642 - val_accuracy: 0.5613\n",
            "Epoch 99/100\n",
            "71/71 [==============================] - 1s 9ms/step - loss: 0.5003 - accuracy: 0.8000 - val_loss: 0.9646 - val_accuracy: 0.5600\n",
            "Epoch 100/100\n",
            "71/71 [==============================] - 0s 7ms/step - loss: 0.4983 - accuracy: 0.8107 - val_loss: 0.9699 - val_accuracy: 0.5613\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**MODEL EVALUATION:**"
      ],
      "metadata": {
        "id": "NmiSDX9wAOrK"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@ INITIALIZING MODEL EVALUATION:\n",
        "predictions = model.predict(testX, batch_size=32)                   # Getting Model Predictions. \n",
        "print(classification_report(testY.argmax(axis=1),\n",
        "                            predictions.argmax(axis=1), \n",
        "                            target_names=[\"cat\", \"dog\", \"panda\"]))  # Inspecting Classification Report. "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "acUFOfHfAUl6",
        "outputId": "66bcbde0-2e40-4e0c-8fba-3ab4557214c9"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         cat       0.49      0.56      0.52       239\n",
            "         dog       0.48      0.46      0.47       262\n",
            "       panda       0.75      0.67      0.71       249\n",
            "\n",
            "    accuracy                           0.56       750\n",
            "   macro avg       0.57      0.56      0.57       750\n",
            "weighted avg       0.57      0.56      0.56       750\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@ INSPECTING TRAINING LOSS AND ACCURACY:\n",
        "plt.style.use(\"ggplot\")\n",
        "plt.figure()\n",
        "plt.plot(np.arange(0, 100), H.history[\"loss\"], label=\"train_loss\")\n",
        "plt.plot(np.arange(0, 100), H.history[\"val_loss\"], label=\"val_loss\")\n",
        "plt.plot(np.arange(0, 100), H.history[\"accuracy\"], label=\"train_acc\")\n",
        "plt.plot(np.arange(0, 100), H.history[\"val_accuracy\"], label=\"val_acc\")\n",
        "plt.title(\"Training Loss and Accuracy\")\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"Loss/Accuracy\")\n",
        "plt.legend()\n",
        "plt.show();"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 299
        },
        "id": "zScJ-oFZBCQs",
        "outputId": "8f53bc09-18bd-4042-97c9-af2532e40291"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**SHALLOWNET ON CIFAR-10:**"
      ],
      "metadata": {
        "id": "VH9Mktk89E9h"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@ GETTING THE DATASET:\n",
        "((trainX, trainY), (testX, testY)) = cifar10.load_data()                # Loading the Dataset. \n",
        "trainX = trainX.astype('float') / 255.0                                 # Normalizing the Dataset. \n",
        "testX = testX.astype('float') / 255.0                                   # Normalizing the Dataset. \n",
        "\n",
        "#@ PROCESSING THE DATASET:\n",
        "lb = LabelBinarizer()                                                   # Initializing Label Binarizer. \n",
        "trainY = lb.fit_transform(trainY)                                       # Converting Labels into Vectors. \n",
        "testY = lb.transform(testY)                                             # Converting Labels into Vectors. \n",
        "labelNames = ['airplane', 'automobile', 'bird', 'cat', 'deer', \n",
        "              'dog', 'frog', 'horse', 'ship', 'truck']                  # Initializing Label Names.       "
      ],
      "metadata": {
        "id": "CliaE7ZMCfjs"
      },
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@ INITIALIZING OPTIMIZER AND MODEL: \n",
        "opt = SGD(0.01)                                                   # Initializing SGD Optimizer. \n",
        "model = ShallowNet.build(width=32,height=32,depth=3, classes=10)  # Initializing Shallow Neural Network. \n",
        "model.compile(loss=\"categorical_crossentropy\",optimizer=opt,\n",
        "              metrics=[\"accuracy\"])                               # Compiling ShallowNet Model. \n",
        "H = model.fit(trainX,trainY, validation_data=(testX,testY), \n",
        "              batch_size=32, epochs=40, verbose=1)                # Training ShallowNet Model. "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PZLP_RmK_X9C",
        "outputId": "48adffce-7b67-42e6-9cff-590ac9afbb0d"
      },
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/40\n",
            "1563/1563 [==============================] - 10s 6ms/step - loss: 1.8265 - accuracy: 0.3619 - val_loss: 1.7312 - val_accuracy: 0.4042\n",
            "Epoch 2/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.7196 - accuracy: 0.4130 - val_loss: 1.7295 - val_accuracy: 0.4007\n",
            "Epoch 3/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6985 - accuracy: 0.4204 - val_loss: 1.7096 - val_accuracy: 0.4057\n",
            "Epoch 4/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6869 - accuracy: 0.4264 - val_loss: 1.7156 - val_accuracy: 0.4086\n",
            "Epoch 5/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6784 - accuracy: 0.4291 - val_loss: 1.7293 - val_accuracy: 0.3999\n",
            "Epoch 6/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6728 - accuracy: 0.4308 - val_loss: 1.7241 - val_accuracy: 0.4021\n",
            "Epoch 7/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6683 - accuracy: 0.4339 - val_loss: 1.7158 - val_accuracy: 0.4080\n",
            "Epoch 8/40\n",
            "1563/1563 [==============================] - 8s 5ms/step - loss: 1.6610 - accuracy: 0.4374 - val_loss: 1.7177 - val_accuracy: 0.4079\n",
            "Epoch 9/40\n",
            "1563/1563 [==============================] - 8s 5ms/step - loss: 1.6564 - accuracy: 0.4372 - val_loss: 1.7337 - val_accuracy: 0.4016\n",
            "Epoch 10/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6518 - accuracy: 0.4385 - val_loss: 1.7597 - val_accuracy: 0.3884\n",
            "Epoch 11/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6478 - accuracy: 0.4391 - val_loss: 1.7413 - val_accuracy: 0.4002\n",
            "Epoch 12/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6434 - accuracy: 0.4431 - val_loss: 1.7271 - val_accuracy: 0.4088\n",
            "Epoch 13/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6401 - accuracy: 0.4440 - val_loss: 1.7503 - val_accuracy: 0.3934\n",
            "Epoch 14/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6351 - accuracy: 0.4467 - val_loss: 1.7408 - val_accuracy: 0.4001\n",
            "Epoch 15/40\n",
            "1563/1563 [==============================] - 13s 8ms/step - loss: 1.6332 - accuracy: 0.4462 - val_loss: 1.7298 - val_accuracy: 0.4043\n",
            "Epoch 16/40\n",
            "1563/1563 [==============================] - 10s 6ms/step - loss: 1.6297 - accuracy: 0.4466 - val_loss: 1.7479 - val_accuracy: 0.3956\n",
            "Epoch 17/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6261 - accuracy: 0.4503 - val_loss: 1.7322 - val_accuracy: 0.4014\n",
            "Epoch 18/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6232 - accuracy: 0.4492 - val_loss: 1.7369 - val_accuracy: 0.4020\n",
            "Epoch 19/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6210 - accuracy: 0.4502 - val_loss: 1.7439 - val_accuracy: 0.3950\n",
            "Epoch 20/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6173 - accuracy: 0.4505 - val_loss: 1.7379 - val_accuracy: 0.4007\n",
            "Epoch 21/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6148 - accuracy: 0.4525 - val_loss: 1.7435 - val_accuracy: 0.3998\n",
            "Epoch 22/40\n",
            "1563/1563 [==============================] - 9s 5ms/step - loss: 1.6132 - accuracy: 0.4524 - val_loss: 1.7441 - val_accuracy: 0.4034\n",
            "Epoch 23/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6096 - accuracy: 0.4536 - val_loss: 1.7448 - val_accuracy: 0.4036\n",
            "Epoch 24/40\n",
            "1563/1563 [==============================] - 11s 7ms/step - loss: 1.6077 - accuracy: 0.4545 - val_loss: 1.7499 - val_accuracy: 0.4004\n",
            "Epoch 25/40\n",
            "1563/1563 [==============================] - 9s 5ms/step - loss: 1.6052 - accuracy: 0.4561 - val_loss: 1.7495 - val_accuracy: 0.3968\n",
            "Epoch 26/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6030 - accuracy: 0.4556 - val_loss: 1.7578 - val_accuracy: 0.3960\n",
            "Epoch 27/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.6006 - accuracy: 0.4571 - val_loss: 1.7807 - val_accuracy: 0.3900\n",
            "Epoch 28/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.5990 - accuracy: 0.4572 - val_loss: 1.7594 - val_accuracy: 0.3959\n",
            "Epoch 29/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.5956 - accuracy: 0.4598 - val_loss: 1.7687 - val_accuracy: 0.3883\n",
            "Epoch 30/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.5941 - accuracy: 0.4602 - val_loss: 1.7621 - val_accuracy: 0.3944\n",
            "Epoch 31/40\n",
            "1563/1563 [==============================] - 8s 5ms/step - loss: 1.5923 - accuracy: 0.4613 - val_loss: 1.7679 - val_accuracy: 0.3929\n",
            "Epoch 32/40\n",
            "1563/1563 [==============================] - 9s 5ms/step - loss: 1.5906 - accuracy: 0.4594 - val_loss: 1.7737 - val_accuracy: 0.3907\n",
            "Epoch 33/40\n",
            "1563/1563 [==============================] - 9s 5ms/step - loss: 1.5878 - accuracy: 0.4611 - val_loss: 1.7712 - val_accuracy: 0.3898\n",
            "Epoch 34/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.5865 - accuracy: 0.4617 - val_loss: 1.7686 - val_accuracy: 0.3953\n",
            "Epoch 35/40\n",
            "1563/1563 [==============================] - 9s 5ms/step - loss: 1.5849 - accuracy: 0.4632 - val_loss: 1.7902 - val_accuracy: 0.3840\n",
            "Epoch 36/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.5829 - accuracy: 0.4637 - val_loss: 1.7738 - val_accuracy: 0.3971\n",
            "Epoch 37/40\n",
            "1563/1563 [==============================] - 9s 5ms/step - loss: 1.5816 - accuracy: 0.4637 - val_loss: 1.7711 - val_accuracy: 0.3887\n",
            "Epoch 38/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.5795 - accuracy: 0.4640 - val_loss: 1.7855 - val_accuracy: 0.3854\n",
            "Epoch 39/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.5779 - accuracy: 0.4650 - val_loss: 1.7818 - val_accuracy: 0.3925\n",
            "Epoch 40/40\n",
            "1563/1563 [==============================] - 9s 6ms/step - loss: 1.5763 - accuracy: 0.4640 - val_loss: 1.7793 - val_accuracy: 0.3862\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@ INITIALIZING MODEL EVALUATION:\n",
        "predictions = model.predict(testX, batch_size=32)                   # Getting Model Predictions. \n",
        "print(classification_report(testY.argmax(axis=1),\n",
        "                            predictions.argmax(axis=1), \n",
        "                            target_names=labelNames))               # Inspecting Classification Report. "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mPkqXCvuAHk6",
        "outputId": "2dabcedd-4bd2-4678-8629-6f336d645c85"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "    airplane       0.47      0.39      0.42      1000\n",
            "  automobile       0.50      0.43      0.46      1000\n",
            "        bird       0.28      0.29      0.28      1000\n",
            "         cat       0.26      0.25      0.25      1000\n",
            "        deer       0.31      0.39      0.35      1000\n",
            "         dog       0.31      0.31      0.31      1000\n",
            "        frog       0.41      0.41      0.41      1000\n",
            "       horse       0.46      0.40      0.43      1000\n",
            "        ship       0.46      0.55      0.50      1000\n",
            "       truck       0.43      0.45      0.44      1000\n",
            "\n",
            "    accuracy                           0.39     10000\n",
            "   macro avg       0.39      0.39      0.39     10000\n",
            "weighted avg       0.39      0.39      0.39     10000\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@ INSPECTING TRAINING LOSS AND ACCURACY:\n",
        "plt.style.use(\"ggplot\")\n",
        "plt.figure()\n",
        "plt.plot(np.arange(0, 40), H.history[\"loss\"], label=\"train_loss\")\n",
        "plt.plot(np.arange(0, 40), H.history[\"val_loss\"], label=\"val_loss\")\n",
        "plt.plot(np.arange(0, 40), H.history[\"accuracy\"], label=\"train_acc\")\n",
        "plt.plot(np.arange(0, 40), H.history[\"val_accuracy\"], label=\"val_acc\")\n",
        "plt.title(\"Training Loss and Accuracy\")\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"Loss/Accuracy\")\n",
        "plt.legend()\n",
        "plt.show();"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 299
        },
        "id": "ne6QOzlzA62z",
        "outputId": "4c9f49d2-d84d-4383-f74f-b5fc51714819"
      },
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    }
  ]
}