ChallengeA.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "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": "3.8.3"
    },
    "colab": {
      "provenance": [],
      "collapsed_sections": []
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oU2z8w0-hkYE"
      },
      "source": [
        "# Challenge A by Salvatore Cavallaro (University of Catania)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uO0qGVADLxEf"
      },
      "source": [
        "Let's import the libraries which will be used to solve the Challenge Task."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-Uql0UpUhkYv"
      },
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import os\n",
        "import cv2\n",
        "from tqdm import tqdm\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "from imutils import face_utils\n",
        "import dlib"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CITFZ95EMN7l"
      },
      "source": [
        "The dataset and the related information are stored into a personal folder in Google Drive:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0shMCwqs8hFP",
        "outputId": "239a72ce-0207-4776-8d33-6ad268345c8e"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Dgt02wGhhkY0"
      },
      "source": [
        "Here we define some useful data:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PHWmtV1xhkY4"
      },
      "source": [
        "training_dir = \"/content/drive/MyDrive/Machine Learning Competition/challengeA_data/challengeA_data/images_train\"\n",
        "test_dir = \"/content/drive/MyDrive/Machine Learning Competition/challengeA_data/challengeA_data/images_test\"\n",
        "training_csv_dir = \"/content/drive/MyDrive/Machine Learning Competition/challengeA_data/challengeA_data/challengeA_train.csv\"\n",
        "test_csv_dir = \"/content/drive/MyDrive/Machine Learning Competition/challengeA_data/challengeA_data/challengeA_test.csv\"\n",
        "img_h, img_w, img_c = 48, 48, 3"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WMpvgIcFZOiZ"
      },
      "source": [
        "We define also two functions to plot diagrams related to *loss* and *accuracy* of trained models:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "J7a47jyaZbKW"
      },
      "source": [
        "def plotAccuracy(history):\n",
        "  plt.plot(history.history['accuracy'])\n",
        "  plt.plot(history.history['val_accuracy'])\n",
        "  plt.title('Model Accuracy')\n",
        "  plt.ylabel('accuracy')\n",
        "  plt.xlabel('epoch')\n",
        "  plt.legend(['train', 'validation'], loc='upper left')\n",
        "  plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DTAx2JLVZpp9"
      },
      "source": [
        "def plotLoss(history):\n",
        "  plt.plot(history.history['loss'])\n",
        "  plt.plot(history.history['val_loss'])\n",
        "  plt.title('Model Loss')\n",
        "  plt.ylabel('loss')\n",
        "  plt.xlabel('epoch')\n",
        "  plt.legend(['train', 'validation'], loc='upper left')\n",
        "  plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QFGOlbB5RK-X"
      },
      "source": [
        "#Explanatory analysis of the dataset"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-sUWwUHksK0a"
      },
      "source": [
        "Before beginning the explanatory analysis we read the images and labels of our datasets."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sdmUM6L_hkY_"
      },
      "source": [
        "### Reading of the dataset from the directory to a Numpy file"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LAR0TVnMhkZB"
      },
      "source": [
        "Here we define a function to read the dataset from the directory into a Tensor: it is a time consuming operation!"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kARY94KAhkZD"
      },
      "source": [
        "def readToTensor(directory):    \n",
        "    X_content = tf.zeros([0, img_h, img_w, img_c], dtype='uint8')    \n",
        "    for img in tqdm(os.listdir(directory)): \n",
        "        img_content = cv2.imread(os.path.join(directory, img)) \n",
        "        tmp_content = tf.convert_to_tensor(img_content) \n",
        "        X_content = tf.concat([X_content, [tmp_content]], 0)\n",
        "    \n",
        "    return X_content\n",
        "   "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GX1Ol7k5hkZE"
      },
      "source": [
        "Then we actually fill our tensors with training and test images:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8Hj325imhkZH"
      },
      "source": [
        "X_train = readToTensor(training_dir)\n",
        "X_test_label = readToTensor(test_dir)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "feRle-tMhkZM"
      },
      "source": [
        "The datasets are then saved as a binary file into the current directory"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "20YF77BUhkZN"
      },
      "source": [
        "X_train_save = X_train.numpy()\n",
        "np.save(\"datasetTrain\", X_train_save)\n",
        "\n",
        "X_test_save = X_test.numpy()\n",
        "np.save(\"datasetTest\", X_test_save)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uIUh01j-hkZO"
      },
      "source": [
        "### Reading of the dataset from the binary files"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nIFGIfnChkZP"
      },
      "source": [
        "A faster way to charge our images dataset is using the binary files .npy previously saved"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZeqrTLYKhkZ3"
      },
      "source": [
        "X_train = np.load(\"/content/drive/MyDrive/Machine Learning Competition/datasetTrain.npy\")\n",
        "X_test_label = np.load(\"/content/drive/MyDrive/Machine Learning Competition/datasetTest.npy\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fnOpROy2hkZ6"
      },
      "source": [
        "### Reading of the *labels* from the $.csv$ file"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gJ1DTBSVhkZ6"
      },
      "source": [
        "Now it's time to load information from .csv file provided for the challenge containing labeling information."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wrmrLsyChkZ7"
      },
      "source": [
        "def readCSVLabels(directory):\n",
        "    df = pd.read_csv(directory, sep=',',header=0)  #.csv content is put into a dataframe to be processed correctly as a table\n",
        "    df_sorted = df.sort_values(\"image_id\")    #read content is order to use an ascending approach according to the label id (same ordering of TRAINING directory -> X_train and labels match)   \n",
        "    df_sorted = df_sorted[\"emotion\"].values  #'emotion' column is what we are interested in -> we just keep it as a Numpy array\n",
        "    return df_sorted"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8N7vqLjzhkZ8"
      },
      "source": [
        "Y_train = readCSVLabels(training_csv_dir)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZIaAE3skwSo2"
      },
      "source": [
        "##Dataset statistics"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0ZvN0A0IwgJB"
      },
      "source": [
        "Let's try to discover how many images contain the *training dataset* and the *test dataset*."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "E8hceHR1wqjD",
        "outputId": "f1e034cf-0936-4325-8302-aa6b6ab2ed31"
      },
      "source": [
        "print(\"There are \"+str(X_train.shape[0])+\" images into the training dataset!\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "There are 30503 images into the training dataset!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "577q1TC_xAW3",
        "outputId": "00350b95-4274-412d-9a86-88c8d3285cff"
      },
      "source": [
        "print(\"There are \"+str(X_test_label.shape[0])+\" images into the test dataset to be labeled by our model!\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "There are 4038 images into the test dataset to be labeled by our model!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6TYDYrUoxfZE"
      },
      "source": [
        "Now we want to know something about the distribution of the 7 classes among the elements of the *training dataset*."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ivXJfZlxxw9T",
        "outputId": "d1d0cba5-4f4d-46bf-d7b9-177d63d54881"
      },
      "source": [
        "classes = []\n",
        "df = pd.read_csv(training_csv_dir, sep=',',header=0)\n",
        "for i in range(7):\n",
        "  classes.append(df[df.emotion == i].shape[0])\n",
        "  print(\"Class \"+str(i)+\": \"+str(classes[i]))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Class 0: 4198\n",
            "Class 1: 466\n",
            "Class 2: 4320\n",
            "Class 3: 7682\n",
            "Class 4: 5177\n",
            "Class 5: 3401\n",
            "Class 6: 5259\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3TXwgZVI0vjl"
      },
      "source": [
        "We can see that the unbalancing is mainly focused on class 1 (Disgust), so it will be necessary to consider if this aspect could lead to bad performances of our models in the following steps."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "j_REHquA1kzx"
      },
      "source": [
        "##Image properties"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "83UhQzB93OBH"
      },
      "source": [
        "Firstly, let's try to visualize an example image from our dataset (and the label):"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "id": "XWKiknpk3V_L",
        "outputId": "d96f68e2-049d-4961-8650-6a34f4f2fd79"
      },
      "source": [
        "plt.imshow(X_train[0])\n",
        "print(\"The label for this image is: \"+str(Y_train[0]))\n",
        "plt.show()  "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "The label for this image is: 6\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wKMGXxte3jNo"
      },
      "source": [
        "As we can see above the image is represented as a grayscale picture. Let's try to see the shape of the related Numpy array item from the dataset."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XE-sO9xA3-zQ",
        "outputId": "503e15e9-d06b-43ba-d346-0c8d7cdb6da2"
      },
      "source": [
        "print(\"The shape is: \"+str(X_train[0].shape))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "The shape is: (48, 48, 3)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "anPaCmIF4MRG"
      },
      "source": [
        "Therefore, the image is $48x48$ pixels with 3 channels. This information will be kept in consideration to define the architecture of the Models in the following steps."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "037WIOVz-y5_"
      },
      "source": [
        "##Patterns visualization"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sOj9dLUoTLAP"
      },
      "source": [
        "First, we split our original dataset into 7 different subsets through a Dictionary, one for each class."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ljvrz_raPUPM"
      },
      "source": [
        "dataset = {\n",
        "    \"class_0\" : X_train[Y_train==0],\n",
        "    \"class_1\" : X_train[Y_train==1],\n",
        "    \"class_2\" : X_train[Y_train==2],\n",
        "    \"class_3\" : X_train[Y_train==3],\n",
        "    \"class_4\" : X_train[Y_train==4],\n",
        "    \"class_5\" : X_train[Y_train==5],\n",
        "    \"class_6\" : X_train[Y_train==6]\n",
        "}"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6opOJKtPO0kJ"
      },
      "source": [
        "A first intresting information we could get from our dataset is **Average Image**. We will see how the different emotions are highly influenced by face treats (mouth, eyes, eyebrows, ...)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "bowa3vjFO_cQ",
        "outputId": "2f6ff147-040d-4e53-9c48-d6dac6387fc2"
      },
      "source": [
        "def mean(datasetClass, title, size = (img_h, img_w, img_c)):\n",
        "    mean_img = np.mean(datasetClass, axis = 0)\n",
        "    plt.imshow(mean_img.astype(np.uint8))\n",
        "    plt.title(f'Average {title}')\n",
        "    plt.axis('off')\n",
        "    plt.show()\n",
        "    return mean_img\n",
        "\n",
        "angry_mean = mean(dataset[\"class_0\"], 'angry')\n",
        "disgust_mean = mean(dataset[\"class_1\"], 'disgust')\n",
        "fear_mean = mean(dataset[\"class_2\"], 'fear')\n",
        "happy_mean = mean(dataset[\"class_3\"], 'happy')\n",
        "sad_mean = mean(dataset[\"class_4\"], 'sad')\n",
        "surprise_mean = mean(dataset[\"class_5\"], 'surprise')\n",
        "neutral_mean = mean(dataset[\"class_6\"], 'neutral')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XnQHdyOKbbUJ"
      },
      "source": [
        "Another information could be the **Contrast between images**. We might consider all the possible combination of classes, but we just show some of them to illustrate this method. We will observe that opposite sentiments shows high difference, while similar sentiments (e.g. negative) show low difference."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 758
        },
        "id": "ZmILo-LIcctn",
        "outputId": "1d83677b-587d-41b4-933d-8a52c687b007"
      },
      "source": [
        "contrast_mean = happy_mean - sad_mean\n",
        "plt.imshow(contrast_mean.astype(np.uint8))\n",
        "plt.title(f'Difference Between Happy and Sad Average')\n",
        "plt.axis('off')\n",
        "plt.show()\n",
        "\n",
        "contrast_mean = surprise_mean - neutral_mean\n",
        "plt.imshow(contrast_mean.astype(np.uint8))\n",
        "plt.title(f'Difference Between Surprise and Neutral Average')\n",
        "plt.axis('off')\n",
        "plt.show()\n",
        "\n",
        "contrast_mean = angry_mean - fear_mean\n",
        "plt.imshow(contrast_mean.astype(np.uint8))\n",
        "plt.title(f'Difference Between Angry and Fear Average')\n",
        "plt.axis('off')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAATsAAAD3CAYAAABigfO8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAX9klEQVR4nO3de7CkRXnH8e+zZw/LbUFANIK6lteIq0ZLRbxEqiRqUCLmijdWU0lEDdGKpogiRrkoJiok0QgxGiJeojEaNZKKsaJEJRaJEWMiIRpAFmEVVkDW5bLnbOeP7iPv9r5n5tmm35lzpn+fKoqdd963334v85yep3v6tRACIiKzbs20KyAiMgkKdiLSBAU7EWmCgp2INEHBTkSaoGAnIk2oEuzM7DwzO63z+uVm9n0z22Zmh5jZk83s2+n18TX2KbMnv49WGjM72syunXY97g4zu9rMjpl2PaZhbLBLJ+c2M7vVzG42s0vM7CQz+8m2IYSTQghnpPXngXcCzwgh7B9C2AqcDrwrvf67oQ5mSJ3zsM3MbjKzz5rZ/ZzbrvgPiZntZWbvMLNr0zFebWbnTrIO3ftoNTKzYGbf7H42zOxMM7ugQtkPSOWvvbtlOfb1prSvI4fe1yR5W3bHhRDWAxuAs4FTgPcts+69gb2B/+4s25C9dpvExd0Dx4UQ9gfuA3wf+NMp16em1wGPA54ArAeOBv6jpKCSa2ZmcyX7WoEOA06Yxo5rfFbMzIATgR+m/1c3tWsdQhj5H3A1cEy27AnATmBjen0BcCbwUODHQAC2Af8M/F9a97a0bB1wIDFYXg98L207l8p6CfAV4Bxga3pvHfB24BpikDkP2CetfzRwLfAa4AepzJd26roP8A7gu8AtwJc72z4RuAS4GfgGcLT3PADHAv/bed1bR2C/dOw70/FvI34gbgPumbY9FVgADkivzwDOHVVuZ7/PAS5Lx3AJ8Kiszq8F/jMd+0eBvZc5vr8HXj3i+APw4M7rC4Azs2twCrAFuLCz7PXAjakuL8y2fw9wEfGeOSYr856pTjcTP3hfAtak9w4D/ha4AbgK+J0R9X428HXgR8Bm4E2d9x6QjmtTOr83Aqdm984FwE3At4DfA64dc45OAb4NrE3LzgQu6Kyz7D3H7vfYm4APpn9fw12fq23AUfR/Vh5E/NxtTcfzIeAeoz7P2TH8LPHefGEqY6+0/B+A387W/Qbwi+nfPw38U7pWVwC/OuZaL3td0jYnEj+zW4HTuvUmNtJ+nxhbtgIfAw4eG8tKgl3n5L+858ZfuoHWjriInwTOJwaCewGXAi/rBLsF4GRgLfGGOwf4NHAwsdXxGeCtnQ/aAvGr8jwxCG0HDkrvvxv4InA4MAc8iRhADk8n6th08n4uvT503HkA9gX+CvhA5/1xdbw2K+9fgF9K//5cunA/33nveY5yH0MM8EemY9uU6rmuU+dLicHhYOBy4KRlju8N6Zq+AngkYHsY7BaAt6Vzu09n2TvTsqcRb/SHdba/BXhyOv97Z2W+lRjY59N/TwUsrfs14I3AXsADgSuBZy5zXEen41kDPIr4B+P47F59b6rzo4E7gIen988mBtmDgfsB/5Vfx55z9JBUv9/Igx1j7jlGB7ulunY/Vy9h98/Kg1O564BDiffSuXsQ7N5HDB7zqW5L9+iJwFc66x1BDNjriJ/jzcBLUz0eQwy0R4y41qOuyxHEgP6UdI3fDuzgrs/fq4CvAvdN+z8f+MiQwe6rpL+C7EGwI37NvYNdWyfPB77QuYDXdN4z4ofkQZ1lRwFXdW7m27L9/YD4F3RNeu/RPfU/BbgwW/aPwKYR52FbusA7gOuAR+5BHfNgdwbwJ+nm2JIu4NnpRrgNOMRR7nuAM7JyrwCe1qnzizrv/SFw3jLHNwe8kthSuCMd36bO++OC3Z10Wo3cFez26yz7GHBaZ/sPZHXolnk68KnuPtPyI7v3R1r2OuAvx93Lad1zgXOye/W+nfcvBU5I/74SeFbnvd/Kr2NWdiAGm2OJrZK92DXYjbznKAt214w53uOBr4/7PKf39iW2tJaCzvnAp9K/16d7cUN6fRbw/vTvXwO+lJV1PvAHy13rMdfljXSCV6rXndwVQy4Hnt55/z7Ez+TaUfu4O9/xDyc2WffUBuJfjetjegCIQWlzZ53uvw8lHuzXOusb8cO5ZGsIYaHzejuwP/Gr0N7EVlNfPX7FzI7rLJsHvjCi7seHED6fcg7PBS42syOIX1HH1TF3MbHV81jgm8SvAO8jBunvhBC2mtm9xpS7AdhkZid3yt2L2JJbsqXz7+3Zez8RQlgktoLfbWb7AL8OvN/MLg0hXD7iOJbcEEK4PVt2Uwjhx53X3832v5nl/RHxw/65dOx/HkI4m3jMh5nZzZ1154gtsN2kJPvZwEbiuVkH/E22Wn6O9k//Piyr43dH1PcnQggXpQ6pl2Vvldxz4+xyDs3s3sAfE1vC64mfrZucZT2P+AfqovT6Q8DnzezQEMINZvZZYj7ybcQGym+m9TYAR2bXZC0xnbFcPUddl13Oewhhu5lt7Wy+Afikme3sLFskNqS+t9zBFQ09MbPHE4Pdlws230xsOdwzhHCP9N8BIYRHdNYJnX/fSGzpPKKz/oEhdhSMcyNwOzGP0VePCztl3iOEsF/6QI0UQlgMIXyCeIKf4qhj6CnmEuBhxBvs4hDCt4D7E1sFFzuPfTNwVnYM+4YQPjL+1Iw8vttCCO8mfkiOSIu3EwPvkp/KN+sp6iAz26/z+v7EFuOobZbqcGsI4TUhhAcCvwD8rpk9nXjMV2XHvD6EcOwyRX2YmAa4XwjhQOJXY1tm3dz1xK+v3fp7nUrMV3bP2bh77scsf46XO1f58rekZY8MIRwAvAj/8W4iBvprzGwLMfjMAy9I738EeL6ZHUVsRCwF6c3Ee7h7XPuHEF4+op6jrsv1xK+oAKQ/vod0tt1MTPl097d3CGHZQAd7GOzM7AAzew7w18Tm9Tf3ZHuAEML1xBzVO1J5a8zsQWb2tGXW30nMqZyTWjqY2eFm9kzHvnYC7wfeaWaHmdmcmR1lZuuADwLHmdkz0/K90xCR+44uNfZYmdlzgYOAyx11/D5wiJkd2KnbdmJu55XcFdwuAU5aeu0o973ASWZ2ZKrTfmb2bDNbP+4Yeo7p1en49zGztWa2idgy+Hpa5TLgBelcPYuYg/N4cxrW8lRiZ0reqlquPs8xswen3sFbiH9YdhK/Zt5qZqekus6Z2cb0B7jPeuCHIYTbzewJ3PXB9fgY8DozOyjdFyeP22BJCOGLxBzfps7icffcZcAJZjZvZo8Dfrmz7Q3E43/gmF2vJ6ZbbjGzw4mdKmOldZ9OvEY/k/57NLEVt9QrexGxVXU68NF0f0LsSHqomb041X3ezB5vZg8fU8/lrsvHiefpSWa2F7GF3w3Y5wFnmdmGVPdD0+dxJG+w+4yZ3UqMqKcSv3691LltnxOJTddvEVsPHyd+717OKcB3gK+a2Y+AzxNbRR6vJX5N/Dfi1+63EXv1NhO/ir6eeCNtJt4Yo87JZ8xsGzGvcRYx17I0pGbZOoYQ/of4V/FKi2MVl77KXUz8y3lp5/V6YlJ57LGHEP6d+FXiXcTz+B1iHqfEdmKv9RZii/KVxOT0len9VwHHEXOWLwQ84yW3pHpdR/xKdFI6Fx4PIR7rNuBfgT8LIXwhfd1e+kBeler6F8Qe/j6vAE5P9+8biQHM683Er65XEf9AXzh69d28gdi5AYDjnjuN+C3kprTvD3e23U68576S7qEnjqjzY4l/ID4LfMJZ1xcDl4UQPhdC2LL0HzGv/Cgz2xhCuCOVd0xWt1uBZxC/4l5HvO5LnVXLWfa6pM/UycRG1fXEe+AHxG+EEL+mf5qY4riV2H8wdkygpQSfSFVmdjSx9T+2pSwyipntT/wj+5AQwlWl5ei3sSKy4pjZcWa2b8r5vp347ezqu1Omgp2IrETPJX4lvo6Y0jgh3M2vofoaKyJNUMtORJqwkn5kX2zjxo27NU/zFuvOnTvzVXZbtri4uNs6CwsLI1/3ldO3L08LOl+nM5B4ZDme7Up46lxrX3378xxrrX2VruPRd47WrFkz8jXA3NxclXXyZfk2fXXsq/MVV1xR72JPgVp2ItIEBTsRaYKCnYg0QcFORJowEx0UfUoS0KUJcU85Hp5kvyfZXWtffYYcqlTrPNa6Zp5zNO3zUbL/0mNd7dSyE5EmKNiJSBMU7ESkCTORs/PkMkoHFdcaMFxLX37OMyDUoyRHNeTA39Kc2SSPo1Zesa+cvnst57n2+bK+cmvdQyuZWnYi0gQFOxFpgoKdiDRBwU5EmjATHRQengSwJ0ldK5HtGQhcOoB4yA6KIeX7rzXriqcTo7STq9Y56ys7v9alHQt5Od7ZdGaNWnYi0gQFOxFpgoKdiDRhJnJ2tXJttdbxDOz05Ho8A4j7rLTcW59p54hK9j/p4/KUlefx+uroWceTn1zt1LITkSYo2IlIExTsRKQJCnYi0gR1UHR4ZjQpTUB7BnbWmnlikrPn1txmqFlGpt0Z4jHJ2VP6Or5q3ecrmVp2ItIEBTsRaYKCnYg0oZmcnWem4tLJAnK1BvUOOUB1kj8Gr/mUsEnmlkr2VXMShlp555IJBZSzExFZpRTsRKQJCnYi0gQFOxFpgjoo9rCcSSZuayb2a61TkoCflQ6K0k6dIQeHlwwqHnJA+0qmlp2INEHBTkSaoGAnIk2YiZzdkPk4zzpDPWGq5g/oPUpm4p32ZAVD5jWHnBW6ZP+lg4Fr3eernVp2ItIEBTsRaYKCnYg0QcFORJowEx0UtRKutcrxzDzh2c5Tn1K1ZucorU+tWT48SsuZZCfGkHX0zFRcUu5qo5adiDRBwU5EmqBgJyJNmNmc3TQHFZfmoxYXF8eu06ckv9JXR88T0Ep4f0A/1EDjWte+ND/nyZF51MohegbhayIAEZFVSsFORJqgYCciTVCwE5EmzEQHRZ9pzgbh6cToSxKXDqqt1UGR60us1xpEWzrQ1rOOZ5bqWtc15+2MqDXj81CdMRpULCKySinYiUgTFOxEpAkzm7MbKpdRqz41yynJt3ieMFUzRzWUvuMYakIFzzkbMmfXx5OfLPnhv3J2IiKrlIKdiDRBwU5EmqBgJyJNmIkOitJZT0p4ZrXoSwh79j83Nzd2m76yayXka52jvI6Tns241ky8ns6HSc8MM06ta68OChGRVUrBTkSaoGAnIk1QsBORJsxEB4XHkKPoPR0UnnJyno6XvmWlx1qSkPf8qsCzTl8dPcdf+gsOzzTkJcdR8xclno6eko44zXoiIjLDFOxEpAkKdiLShJnI2dXKx9V6VF3pYwrz/Etffi5/3CLUe9zjJHN2pYbKGfada085nrxaaR1rPd5wqM/CaqOWnYg0QcFORJqgYCciTVCwE5EmNN1BUSux7xlU7EkuLyws7PJ6x44du61Tq4Oir475rCt9dfasU6uDojRJ7pmqPD+P3o6FcfvqUzrIvHRw9jhDPn5gJVPLTkSaoGAnIk1QsBORJsxEzq5PyWMBS8rtW+ZZpy/3li/ryzXV4vkhfulAW49aj7YcanZlKMvZec5Z3/5KB6sP9UjGWczhqWUnIk1QsBORJijYiUgTFOxEpAkz20FRYsgkccnA37VrfZcnT4r3dX7k8sHBfUo7CEpn6/DM+uLpEMh5Hj/ZxzOAulYnUumMO6X3bI36rDZq2YlIExTsRKQJCnYi0oSZyNnVyncMlf/wluOZ9dazLJ9QoK/sPvk6nnxUzR+Vlz4ZqwbPOevLoeZ5vdIJBUqVDJ4fcnDySqaWnYg0QcFORJqgYCciTVCwE5EmzEQHRZ+SRPZq6KDoS6R7BvF6Etn5sloDqL1KBl6XljNuG/ANcvbMXjLkPTPUvmaRWnYi0gQFOxFpgoKdiDRhJnJ2Qw7krPUj6lr5ltKngk1yAHXJjMPLLRtXp77z4ZkIIS9nfn5+7DoeNXN2Q10jDSoWEZlhCnYi0gQFOxFpgoKdiDRhJjooPCY5y2uf0qS9pz55B0Vf0r607JJyhlxnqI4nzyBvz8wxkzbU+dBMxSIiq5SCnYg0QcFORJowEzm7vrxFrR+1e3I7Q/043VMf7/49POes1o/sPet5cnZ9ebRpDiif9L40U7GfWnYi0gQFOxFpgoKdiDRBwU5EmjATHRQeQ3ZQ1BqAWTqod6jkcq3HJNac9aTW/qfdiVFrQHtJh8SQ9/BKppadiDRBwU5EmqBgJyJNULATkSY000HhMdQvEWoa8lGGJTyPICz9BYVH6RT0nl+LlJj0LxFKflEyyfOxkqhlJyJNULATkSYo2IlIE5rO2Q2V/6o1aLPmYNyc59jz3FvfMk99vPXr2984npmbJ5l/muSg7z7TnpF7JVPLTkSaoGAnIk1QsBORJijYiUgT1EExgW36lM7wMeSA4ZLHPXo6FUrr7Nmub/+egba19j9u30Mb6hGIs9hhoZadiDRBwU5EmqBgJyJNaDpn51GSu5jkLLze7Wo97nHI46hVTskjGUsHA08yRzbkZAWaqVhEZEYo2IlIExTsRKQJCnYi0oSZ6KDwzPRQK4nvUSuJP8kBxDXLyZf1DeotvR6eAcOlg4hzk7xnps0zm/Fqp5adiDRBwU5EmqBgJyJNmImc3WrIbdVSa3IAzyDaWrk27+y5JYOBS/O1K+26DmmaE16sJGrZiUgTFOxEpAkKdiLSBAU7EWnCTHRQeAw5YLhkxoqaM4oMNUC5tBOjdICqZ9bjlTb4NT9HNetTUvYk11lt1LITkSYo2IlIExTsRKQJzeTshpwF2DMYdqh99/Hkvkr3N2SOKs//TfKa9ak1m7F3UHWNdWrVUTk7EZFVSsFORJqgYCciTVCwE5EmNNNB0WeogbbTflRd3/5LOy3GmfQjImsl7Uv27+lomHRif6gOI3VQiIisUgp2ItIEBTsRacLM5uxW+ky0k87rDTURQV8usDSPV5KP80xE4Cmn1JC5rVpPaSsxVI53mtSyE5EmKNiJSBMU7ESkCQp2ItKEme2gKEkcT7tTw1Pnktk6vOV4Oh/yZZ4OCm9nSEnHwtzc3G7r5Mv6EvuefZc8trFmh0WtwdEeQ83cs5KoZSciTVCwE5EmKNiJSBNmNmc3lGn/QNqz/1ozFS8uLu62Tr5syJxdrXVqzSZcM2dXK9c37YkIVhO17ESkCQp2ItIEBTsRaYKCnYg0YSY6KFbDAMiSJLm3o6HW4x7z/S0sLOy2zo4dO0Zu491Xn1qDij0825XMsOJ9bGJJR0KtTozSGW9WO7XsRKQJCnYi0gQFOxFpgoKdiDShmQ6K0qRsrZlISvQlxPs6BDxJcs9sJXnnw5133jl2ndtvv323dTwdFLV+1ZDXB2B+fn6X132dEaUzoYxbp7SDovQemuTMKKudWnYi0gQFOxFpgoKdiDRhJnJ2fWrNvJpvV5p/GfLRfSVll85CnOfx+mZG8QyGLs2X5rm20vzkULzXZ6jc2mp4ROS0qGUnIk1QsBORJijYiUgTFOxEpAkz0UEx5EDKkoGlpcn30u1qTc1d6zyWDnStNROIZ7aSkn3Xmrq9b1npYPGhzOLgZLXsRKQJCnYi0gQFOxFpwkzk7PqUDCqeZE6k5r5K8kZ9eSzPgN18nb4f2dca0F26/3xZ6UDsWufMo7SOJedWMxWLiMwwBTsRaYKCnYg0QcFORJowsx0UJTyzldTqWKg5M0rJI/885fRtkz9e0TPrSGny29Mh0NdBsW7durHr5Mtqddh4B4vnyzwDn2t1LHjK0awnIiKrlIKdiDRBwU5EmtBMzm6Ss8X2yfNYfXmknDdv4sm1eXJda9fuejv0/Tjds06+r9KJADzHkdenb5nn6WKedTy50Jq5rlqDsyVSy05EmqBgJyJNULATkSYo2IlIE2aig6J0sGVJMrnWYGBPYr+0g6IvkZ4n4PuOI0/se85hXweBZ7vSAdzz8/MjX/fVqXRQcclsMp5B1n1Kz9lQs0vPIrXsRKQJCnYi0gQFOxFpwkzk7DyG/AH/JJU+qcqTs/PklvJ9LS4u7rbOkDMV58fhGVTct45nwHBJDrXmBA8l16NvG8+P/JWzExGZEQp2ItIEBTsRaYKCnYg0YSY6KIbsNJhmh4Q32e0ZVJyX5XkEoidp7dlXqdKZij2Diktmd56VJP60O9mmRS07EWmCgp2INEHBTkSaMLM5u9IBoEOV41Fr8K3nx+ieAbueffUNKs6VXh9Pzq5vnTxnV/oD/iFzdJ6B156ntM3KYPlJUMtORJqgYCciTVCwE5EmKNiJSBOshcSkiIhadiLSBAU7EWmCgp2INEHBTkSaoGAnIk1QsBORJvw/tT/OQ3sttzcAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAARoAAAD3CAYAAADCHptSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAUSklEQVR4nO3dfaxl1VnH8d9z78AMDMNLeWeAscFITIEqaku1KZi0Rd4CWKuloUxRGyGiJLYVlRhFWqWmlL5QpWlqpVCKLW0RrK9EHUOniG2KVqEq0PI6UN6Z6VDocJd/rHVls+8+56y7WM/e58x8P8lN7jln77XXfnvO2s9Zey8LIQgAPM0NXQEA2z8CDQB3BBoA7gg0ANwRaAC4I9AAcFcl0JjZFWb2u43X55rZw2a2xcz2NrOfMrP/Ta9Pq7FMoJSZ/b6ZXT10PXYkEwONmX3bzJ4xs81m9qSZbTSzc8zs/+cNIZwTQrg4Tb+TpA9IemMIYbcQwmOS/kDS5en19V4r46mxHbaY2RNm9iUzOyRz3uPM7H7vOtaQ6hrM7IKh6zKEtP4LaT8v/t3Y8/K3u+2f26I5JYSwRtI6SZdIukDSJ0ZMu7+kVZL+q/HeutbrbGa2omQ+J6eEEHaTdKCkhyV9ZOD6eFgv6XFJZ3ktYMr2aZcH05fi4t8pNQufsP6u23+wbR9CGPsn6duSXt9671WSFiQdkV7/uaT3SPohSd+VFCRtkfSPku5K0z6T3lspaQ/FQLVJ0gNp3vlU1tslfVnSZZIeS5+tlPR+SfcqnuBXSNolTX+cpPslvVPSd1KZZzfquoukSyXdI+kpSTc35j1G0kZJT0r6d0nH5W4HSSdK+p/G6846Slqd1n0hrf8WSQel9/ZJ814oaZuk3dPriyV9cFy5jeWeLOm2tA4bJR3VqvO7JP1HWve/kLRqzDqulrRZ0lskPSfpxxuf/UDar+tTXR6VdGFrO18p6QlJd0j6TUn3t+pyQarLs5LeLenzreV/WNKHRtTttxSPpc2Sbpd0euOzt6f9+v60/G9JOqHx+cslbUjz/oOkyyVdPWI5xzXr3fps5PEi6ey03psl3S3pV9plpvV/SNJVy9n+ab7rWtN+SNKH0//LPZ8OUzw3H0v78dOS9myUfbSkr6e6fC4dN+/JOeZGHlslgSa9f6+kc5uBpnVArhhzkn5R0sfSht1P0q2LOyZtmG2Sfk3SCsUD+DJJN0h6maQ1km6U9EeNnbhN8fJsJ8UAsFXSXunzj0r6Z0lrJc1L+knFk3dt2tAnKrbs3pBe7ztpO0jaVfGk+lTj80l1vL9V3r9IelP6/+8VT6ITGp+dnlHujyoG11endVuf6rmyUedbFQPbyxRPhHPG7Ou3KR6s82k5H+kINB9P++SVigHjh9PnlyiezHtJOlgxoLQDzW2SDknzH6j4pbRn+nxFWpcfG1G3N6f1mJP0C2neAxvHzPclvSPV/VxJD0qy9PlXFC/nV0p6neIJtKxAM+l4kXSS4glsko5VPAaPbh2j70t12GU521/ximCrpDXp9Xya7pjC8+kHU/1XStpX8Xhb/GLbWfFL+XzF8+lnFYPe4vk99pjzCDS3KH2jaRmBRvHS6lm9+Fv5DEn/1Ngw9zY+s3RQHdZ47zWSvtXYic+0lvcdxW+fufTZKzvqf4Fa3yyS/k7S+jHbYYtiFP++4oF85DLq2A40Fyt+g69Q/JY7X/FkXZXqvHdGuX8q6eJWuf8t6dhGnc9sfPbHkq4Ys69vahxwZ0h6RNJOrf16cGP6WyW9Jf1/t6TjG5/9spYGml9sLe9vJL0j/X+ypNsnHY+NeW+TdGrjmLmz8dmuqa4HSDpU8URb3fj8Go0PNAtpPy/+/XzB8XK9pPMbZT6nMa3JjO1/s6Sz0v9vkHRXyfk0YrmnSfp6+v91iq0ia3x+s144v8cec6P+XsqvTmsVryWXa51ipNyUkstPKkbj/RrT3Nf4f1/FA+drjen/Nr2/6LEQwrbG662SdpO0j+KJe9eIerx5scxU7msVv2lHOS2EsGcq8zxJG8zsgMw6tm1QPACPlvQNxSb9sYoB8s4Qk+iTyl0n6Z2tdThE8Zt/0UMd22WJlNj+acVmtCT9ZVrPk1qTjirvIL14vzX/H/XelZLOTP+fKemqrrql+p1lZrc11vMIxf27pF4hhK3p391SvZ4IIXy3Me09o5aTPBhC2LPx91lNOF7M7AQzu8XMHk+fndiq3yMhhO+NWb9J2/8axQAiSW9Nr6Xln08ys/3N7Foze8DMnpZ0daOuB0l6IKQI0jF/zjG3RFGgMbOfUAw0NxfMfp9iBN6nsSN3DyG8ojFNcyUfVfyGf0Vj+j1CTMpO8qik7yk2abvqcVXrgFodQrhkUqEhhOdDCF+Q9LziwTapjqGjmI2SDpd0uqQNIYTbFb99T1QMQjnrfp+k97bWYdcQwmcmb5ol3qZ4PNxoZg8ptlBWKTaNc2xSvGRa1PWLXHs7XC/pKDM7QrFF8+mls0hmtk7xku08SXunYP+fii2+nHrtZWarG+8dmjFf28jjxcxWSvq8Yo5o/1S/v27Vr+sYaJq0/T8n6TgzO1jxmFkMNMs9nyTpD9N7R4YQdlcM8ot13SRprZk1697cl0XH3LICjZntbmYnS7pWsen5jeXML0khhE2KOYlLU3lzZnaYmR07YvoFxYPsMjPbL9VjrZkdn7GsBUl/JukDZnaQmc2b2WvSgXG1pFPM7Pj0/qr00+LB40uVLDpVMR9xR0YdH5a0t5nt0ajbVklfk/SreiGwbJR0zuLrjHI/LukcM3t1qtNqMzvJzNZMWocO6yVdJOlHGn9vknSime2dMf9nJf22me1lZmsVg8JY6Rv+OsWT5tYQwr0jJl2teGI8IklmdrZii2aiEMI9kr4q6SIz29nMXiup5FekccfLzor5jkckbTOzEyS9cZnlj93+IYRHFHONn1S8dL4jrd+yzqdkjWIa4Km0r97d+Owril+g55nZinScv6rxedExlxtobjSzzYrR7ELFxNrZmfN2OUtx59yu+CvBdRp/yXKBpDsl3ZKaejcptgZyvEvx0uTfFC/13idpLoRwn6RTJf2O4gFyn+IGH7dNbjSzLZKelvRexevzxZ/tR9YxhPBNSZ+RdHdqbi42MzcoNntvbbxeo5icm7juIYSvKiZAL1fcjncqXpMvi5kdo9gk/mgI4aHG3w2pzDPGlyApJuPvV/zF5ybFffpsxnxXSjpSYy6bUmvvUsWT4OE0/Zczyl70VsXk5eOSfk/Sp5Yx72IdRh4vIYTNkn5dMdg+kZZ3Q27Zy9j+10h6vV5ozSxa7vl0keIl+1OSviTpC431fE4xAfxLivmpMyX9ldK+LD3mFrPyQFVmdq5ionjcN6vM7FBJ35R0QAjh6V4qh2Uxs39V/BHhk6VlcK8TqjCzAy3eajJnZocr9mv64oR55iT9hqRrCTLTw8yONbMD0qXTeklHKf4IUWzae2hiduys+GvHyxWb3NdK+pNRE6fk7MOKvwD9TB8VRLbDFS8DVysmpX8u5YKKcekEwB2XTgDcDX7ptGXLliVNqhf/hL/0tSTNz89PnKb93tzc0rias6yu99pyWoal07SXPwut0JxtVquc0mWVbtf2dF3z5UyzsLCw7GWVTrNq1ao6O6QQLRoA7gg0ANwRaAC4I9AAcDd4MjhHrcRiqZzE3qR5Xsp0JcvvUpLU7ponJ2FdS2lyPKc+7WRs6XbNSQaXllOj3GlAiwaAOwINAHcEGgDuBs/RlHaQKynHs+Pd0NfXXjmBXO18h2dHu5I8Uum+98wPDn3M9IkWDQB3BBoA7gg0ANwRaAC4GzwZnKNWwrhWYq1WB62XMl9fy6q5XiV3S9dK4OfI3a9ed6ZvL3fud6FFA8AdgQaAOwINAHeD52hqXe+WXkuX5AlqPSkvt+ySzmfTqNa2rpWfy+nQ6XnDZK0bamchj0OLBoA7Ag0AdwQaAO4INADcDZ4MzlHrLt8+l1UzYVtyF3qpPu8ozk2+lpSTY+ikaa0k7tDrkYMWDQB3BBoA7gg0ANzNRI6mFs/OTkMPAdtnx8McfY7U4FlOTtl9P5mvjRwNAIhAA6AHBBoA7gg0ANxNZTK4VrKtdHjX5ZabW5+hDZ1EzVErqT10gnRHGkolBy0aAO4INADcEWgAuBs8RzOLHZtq5To8l19ax5wnug2tzycObs9PPOwTLRoA7gg0ANwRaAC4I9AAcDd4MrhUn09imzaz2mGw9M70kmmGNgtDoPSJFg0AdwQaAO4INADcDZ6jGfpJ+CW6ljU3Vxazaw3v6jXfNI7m4PU0Q8/t6nmcz0IHQlo0ANwRaAC4I9AAcEegAeBu8GSwZ0emkmRfraRdl66EsVcy2POpc6XbsSQZnFPO0EP0ltax1jRt09g5kBYNAHcEGgDuCDQA3A2eo/FUa4SDErm5l5IcjWceKafcWh3tcnIbJeVKZZ3YpjG3UYIOewB2SAQaAO4INADcEWgAuJvKZHCfScO2vodSaS+vtFNfzjApJUOp9NmhctR7k/Td0W3Iu+Bn9emKtGgAuCPQAHBHoAHgjkADwN3gyeDShGBJz0+vR0B2KU2iLiwsLHmv5DGhXXXsKrtE6br1OQRJSXI8p5zcaaa9x3ffaNEAcEegAeCOQAPA3eA5Gs8OcrVyKZ51zCn7+eefr7L8Wp36ct7LyTd0lVOSj5rGnFHO9tiRhs2lRQPAHYEGgDsCDQB3BBoA7gZPBufo847Vvh/lmbP8Ph/TWSsZ3KXdYbAr8euVeK/ZGW7IpO2sJoxp0QBwR6AB4I5AA8Dd4Dma0qEyvIaS7ZLzxL+SjmZddfLMG5RM0/fNiG2lN8vmlN3eZ13zdN2IOmSepNbQMn2jRQPAHYEGgDsCDQB3BBoA7gZPBnep1SHLa3iV0iFRcnh22PMcgiRHTqK5z32fkwzO0WdyeOghckrRogHgjkADwB2BBoC7wXM0Odfp0ziU7aSyS/MPpU+m8xqm1fPJ/LXWIydnVvpUwNIbP71MY2e8HLRoALgj0ABwR6AB4I5AA8Dd4Mlgz+FlS5aVI/cu35Kypq3TWM3kY8lT+ErvlM9JBtcaIrg0YT5kkr9vtGgAuCPQAHBHoAHgbvAcTWnHtpxr+ZJOW11q3QzY55PpuuR0hPTsjFZrCNj2vt+2bduSadrv5eRjcnM2JfWethsv+0aLBoA7Ag0AdwQaAO4INADcDZ4MHlqtDlFena9K5yvttFUrYVuyrFztOuUkcWd1iOLtBS0aAO4INADcEWgAuNtucjRDX0uX3hxZazjTkicM1nq6Ye7y+1TrZtWcsmuVU+tmSG6qBLBDItAAcEegAeCOQAPA3VQmg6cxmdVUM7FYctd36R3e7SfT9fl0Q8+yaya1S5dXMk9JnWqV0zdaNADcEWgAuCPQAHBHoAHgbiqTwW2z8BjEPsejLh1nPKcn6tDbuqS3bN/jg+csv2Sa0uVPW6/sLrRoALgj0ABwR6AB4G7wHE2f1+C18h9953Fq5Wg81apTraFtaqk1tE0tOdtjGnM2tGgAuCPQAHBHoAHgjkADwN3gyeBSJeNqe3aaypmnNNGckyBt35ndpc/xwWvON6kcz4Rx33eG1zBt9ZFo0QDoAYEGgDsCDQB3M5ujKblO9xySxbMz3rR1TvR8Wlyfw/8OraTeXfNMe85IokUDoAcEGgDuCDQA3BFoALibymSw15jEOcvqMm13x9bqeJhzV3zpGOJdaiWMczonDn0MDTmUzdDHZxdaNADcEWgAuCPQAHA3eI6m7xviJi3Ls1NfqZwOWUM/dc5raN/SpyvW2h7Ttq2nsTNeDlo0ANwRaAC4I9AAcEegAeBu8GRwjqGHgPXssDd0EnfSNLnJ8T6Tlp5D0A5ZzsLCwsRpZvGJfxItGgA9INAAcEegAeBuKnM0Jdf7nk/vb6t1vV1ado5aTxz0zEfV6qw5dE5i6OXPAlo0ANwRaAC4I9AAcEegAeBu8GRwadKyVoK4pLNT6TTbyzAltTqNlXagLDH0E/Y871SvMY83WjQA3BFoALgj0ABwN3iOxlOfT+avUe6o+Wp1ovPM4/SZa8oZBaEkRzT00xWHfpqfJ1o0ANwRaAC4I9AAcEegAeBuKpPBtZ6M57Wsmonekvk8k4Y56+r5NMOcZc1CgrTWDwYl+4MOewB2SAQaAO4INADcDZ6j8bwhbcgOWZ55hPn5ebeyS20vww/nqJVr8nya4bShRQPAHYEGgDsCDQB3BBoA7gZPBncZssOeJ88OezXK7ZL7VMBaT4LzSpCWJmNLh/HpsxPdLCSRadEAcEegAeCOQAPAHYEGgLvBk8HTltQdOrHnOdxLyTS1hojJLbs9ZnlpMrYk0Vtzv/Y5fvws3M1OiwaAOwINAHcEGgDuBs/R5Kg1bG7pNXCOofMdXkP71pwvJ2/RHkqlnbMpNfSd4rPQqc4TLRoA7gg0ANwRaAC4I9AAcDcTyeAufXb08+p81SXnkaQ5SdRacsv1Go+8dFvX6ozn+eSAErPQOa8LLRoA7gg0ANwRaAC4m8ocTUlOxLODWsn1fq1l5S6/1rC9OWp1fMzpwNi1Hjk3XnreMFmSN6o1ZNCsokUDwB2BBoA7Ag0AdwQaAO4GTwZ73i1dUk7pXeA569HV+a0kaTr0MCVdai2/1vbwVCvRO4vjvpeiRQPAHYEGgDsCDQB3U5mjaecycjoy1cr19JlbKC07Z/QAz+v00uFla6k1lG2NeUbN57X9GQUBAEYg0ABwR6AB4I5AA8Dd4Mng0ifDeT1hz7NjlWdnwJy7nmvd4d41BIpXMnjoYVJyjoecY7jvDpTThhYNAHcEGgDuCDQA3A2eo6nVGa/WyAR9PoUvd/leQ/vWzC0M2WGvi9cxVFp2l1rbfxbyNrRoALgj0ABwR6AB4I5AA8DdVCaDS6aptazScmolCD2VrL9nh7mh7wKvdTzkPDmx77vppw0tGgDuCDQA3BFoALgbPEdTS+nNkLVGQehzSN5aOYGa+ZchO9GVrkfJzZFd89Uc/ritpOxpzNnQogHgjkADwB2BBoA7Ag0AdzaNHcsAbF9o0QBwR6AB4I5AA8AdgQaAOwINAHcEGgDu/g+VJDb5fSG9EgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PGEdtiINeZlj"
      },
      "source": [
        "One more analysis can be done on **Variability** analyzing *Standard Deviation*. In particular, we will see how much variability is present for each class through lighter areas."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "2651-HDnevrm",
        "outputId": "1e020b31-8cf4-4ad4-b680-12dc790878f1"
      },
      "source": [
        "def std(datasetClass, title, size = (img_h, img_w, img_c)):\n",
        "    std_img = np.std(datasetClass, axis = 0)\n",
        "    plt.imshow(std_img.astype(np.uint8))\n",
        "    plt.title(f'Variability {title}')\n",
        "    plt.axis('off')\n",
        "    plt.show()\n",
        "    return std_img\n",
        "\n",
        "angry_mean = std(dataset[\"class_0\"], 'angry')\n",
        "disgust_mean = std(dataset[\"class_1\"], 'disgust')\n",
        "fear_mean = std(dataset[\"class_2\"], 'fear')\n",
        "happy_mean = std(dataset[\"class_3\"], 'happy')\n",
        "sad_mean = std(dataset[\"class_4\"], 'sad')\n",
        "surprise_mean = std(dataset[\"class_5\"], 'surprise')\n",
        "neutral_mean = std(dataset[\"class_6\"], 'neutral')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uP3Naj-xfm6i"
      },
      "source": [
        "A last step of the exploratory Data Analysis is the usage of **Principal Component Analysis** to visualize the components that are more relevant for each class in terms of variability."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9XXGVRd7f-4G",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "7f81bfae-dbb6-43a4-e1b9-53226b50a134"
      },
      "source": [
        "from sklearn.decomposition import PCA\n",
        "from math import ceil\n",
        "\n",
        "def eigenimages(datasetClass, title, n_comp = 0.7, size = (img_h, img_w, img_c)):\n",
        "    pca = PCA(n_components = n_comp, whiten = True)\n",
        "    pca.fit(np.reshape(datasetClass, (-1, img_h*img_w*img_c)))\n",
        "    print('Number of PC: ', pca.n_components_)\n",
        "    return pca\n",
        "  \n",
        "def plot_pca(pca, size = (img_h, img_w, img_c)):\n",
        "    n = pca.n_components_\n",
        "    fig = plt.figure(figsize=(8, 8))\n",
        "    r = int(n**.5)\n",
        "    c = ceil(n/ r)\n",
        "    for i in range(n):\n",
        "        ax = fig.add_subplot(r, c, i + 1, xticks = [], yticks = [])\n",
        "        ax.imshow((pca.components_[i].reshape(size)*110).astype(np.uint8), \n",
        "                  cmap='Greys_r')\n",
        "    plt.axis('off')\n",
        "    plt.show()\n",
        "    \n",
        "plot_pca(eigenimages(dataset[\"class_0\"], 'angry'))\n",
        "plot_pca(eigenimages(dataset[\"class_1\"], 'disgust'))\n",
        "plot_pca(eigenimages(dataset[\"class_2\"], 'fear'))\n",
        "plot_pca(eigenimages(dataset[\"class_3\"], 'happy'))\n",
        "plot_pca(eigenimages(dataset[\"class_4\"], 'sad'))\n",
        "plot_pca(eigenimages(dataset[\"class_5\"], 'surprise'))\n",
        "plot_pca(eigenimages(dataset[\"class_6\"], 'neutral'))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Number of PC:  14\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 14 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Number of PC:  10\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Number of PC:  15\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 15 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Number of PC:  13\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 13 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Number of PC:  14\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 14 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Number of PC:  14\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 14 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Number of PC:  12\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 12 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xMDo2UawhkZ9"
      },
      "source": [
        "# Models "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VGl6oaT4W_iL"
      },
      "source": [
        "###First model: Support Vector Machine"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7xUzmhIoYkxI"
      },
      "source": [
        "The first model proposed to solve our challenge is a *Support Vector Machine* trained on our dataset to recognize one of the 7 different emotions given a specific image. We first import all the necessary libraries, that are different than the ones used for the remaining four NN Models. In particular, the model is implemented using *scikit-learn*."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3F10BHnJZSM7"
      },
      "source": [
        "import matplotlib as mpl\n",
        "from skimage.feature import hog\n",
        "from skimage.color import rgb2grey\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import accuracy_score\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.svm import LinearSVC"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jts_2a-jZdYH"
      },
      "source": [
        "Here we define a function *create_features* which takes the 3-channel image from our dataset, processes it in order to get a flatten version and the *Histogram of oriented gradients*, then combines them."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hbNkx6TRZhUI"
      },
      "source": [
        "def create_features(img):\n",
        "    features = img.flatten()\n",
        "    grey_image = rgb2grey(img)\n",
        "    hog_features = hog(grey_image, block_norm='L2-Hys', pixels_per_cell=(16, 16))\n",
        "    flat_features = np.hstack(features)\n",
        "    return flat_features"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CY-B08Lya6cn"
      },
      "source": [
        "Then, another function is defined to iterate our dataset and create an array of processed images using the function defined above."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "J4Kpwxa7bCfv"
      },
      "source": [
        "def create_feature_matrix(X_train):\n",
        "    features_list = []    \n",
        "    for img in tqdm(X_train):\n",
        "        image_features = create_features(img)\n",
        "        features_list.append(image_features)        \n",
        "    feature_matrix = np.array(features_list)\n",
        "    return feature_matrix"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1Q2Yb2jSbjru"
      },
      "source": [
        "Here we actually create our Numpy array from the given dataset."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oxXM8YkxbcEO",
        "outputId": "836233c7-8082-48fe-fda5-4221481ec9e1"
      },
      "source": [
        "feature_matrix = create_feature_matrix(X_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 30503/30503 [09:40<00:00, 52.55it/s]\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PhqjtLtWbU22"
      },
      "source": [
        "Next the dataset is standardized and processed using Principal Component Analysis in order to reduce the number of features, choosing the most releavant."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cvoaMLmebwfO"
      },
      "source": [
        "ss = StandardScaler()\n",
        "X_train_stand = ss.fit_transform(feature_matrix)\n",
        "pca = PCA(n_components=500)\n",
        "X_train_pca = ss.fit_transform(X_train_stand)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "E-OD77cJcL9v"
      },
      "source": [
        "Then the dataset is split into *train* and *test*."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mXVRdKExcMTt"
      },
      "source": [
        "X = pd.DataFrame(X_train_pca)\n",
        "y = pd.Series(Y_train)\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.3, random_state=1234123)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aQLfFpCpc9xd"
      },
      "source": [
        "Now it's time to train our SVM using our dataset."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l2uYPVc2c-PR"
      },
      "source": [
        "model1 = LinearSVC()\n",
        "model1.fit(X_train, y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZXSqjYl7dWad"
      },
      "source": [
        "Finally we check the *test* accuracy."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EZiKppWiddT0",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "297c9325-c1c3-4a6c-cea5-9a6c195cb6b9"
      },
      "source": [
        "y_pred = model1.predict(X_test)\n",
        "accuracy = accuracy_score(y_test, y_pred)\n",
        "print('Model accuracy is: ', accuracy)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model accuracy is:  0.2777838487596984\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EEO7geFOeGJK"
      },
      "source": [
        "##Neural networks approach"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5PYqjQlVhkZ9"
      },
      "source": [
        "Let's define some parameters useful for our NN models."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sd-cQxurhkZ-"
      },
      "source": [
        "EPOCHS = 50\n",
        "BATCH_SIZE = 64\n",
        "VERBOSE = 1\n",
        "NB_CLASSES = 7\n",
        "N_HIDDEN = 512\n",
        "VALIDATION_SPLIT = 0.1\n",
        "TEST_SPLIT = 0.1"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PaFyZDRPhkaF"
      },
      "source": [
        "### Second model: Multilayer Perceptron Neural Network"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ep-H2gaRTmge"
      },
      "source": [
        "A second model to execute our task of image classification is a MultiLayer Perceptron Neural Network. Let's define some functions to process our data in order to be suitable for feeding our NN."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iEva9JXlUJOt"
      },
      "source": [
        "def preProcessMLP(X_train, Y_train):\n",
        "  X_tensor = tf.convert_to_tensor(X_train, dtype=tf.uint8) \n",
        "  items_size = X_tensor.shape[0]\n",
        "  test_size = round(items_size * TEST_SPLIT)\n",
        "  train_size = items_size - test_size\n",
        "  X_tensor_test = X_tensor[train_size:train_size+test_size]\n",
        "  X_tensor_train = X_tensor[:train_size]\n",
        "  Y_tensor_test = Y_train[train_size:train_size+test_size]\n",
        "  Y_tensor_train = Y_train[:train_size]\n",
        "  X_tensor_train = tf.reshape(X_tensor_train, [train_size, -1])\n",
        "  X_tensor_test = tf.reshape(X_tensor_test, [test_size, -1])\n",
        "\n",
        "  return X_tensor_train, X_tensor_test, Y_tensor_train, Y_tensor_test\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MAtaWLlQhkaG"
      },
      "source": [
        "Now it's time to execute the preprocessing:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yr7uQHxKhkaG"
      },
      "source": [
        "X_train, X_test, Y_train, Y_test = preProcessMLP(X_train, Y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O01Wi2-ihkaH"
      },
      "source": [
        "And also normalizing the two datasets:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ikQgjaoehkaI"
      },
      "source": [
        "X_train /= 255\n",
        "X_test /= 255"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "v0hDleLvhkaK"
      },
      "source": [
        "An useful tool to elaborate the dataset is One hot encoding:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AHaWDgmqhkaL"
      },
      "source": [
        "Y_train = keras.utils.to_categorical(Y_train, NB_CLASSES)\n",
        "Y_test = keras.utils.to_categorical(Y_test, NB_CLASSES)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DYEHw2cShkaL"
      },
      "source": [
        "Now let's define the model!"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "28IFHI9fhkaL"
      },
      "source": [
        "model2 = tf.keras.models.Sequential()\n",
        "model2.add(keras.layers.Dense(N_HIDDEN*4, input_shape=(X_train.shape[1], ), activation='relu'))\n",
        "model2.add(keras.layers.Dense(N_HIDDEN*2, activation='relu'))\n",
        "model2.add(keras.layers.Dense(N_HIDDEN, activation='relu'))\n",
        "model2.add(keras.layers.Dense(NB_CLASSES, activation='softmax'))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gza1BtcwhkaM",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "aa0097a8-e9a5-4d2f-812d-c535da838b25"
      },
      "source": [
        "model2.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense (Dense)                (None, 2048)              14157824  \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 1024)              2098176   \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 512)               524800    \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 7)                 3591      \n",
            "=================================================================\n",
            "Total params: 16,784,391\n",
            "Trainable params: 16,784,391\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jAyHjkXTXMCD"
      },
      "source": [
        "Here we define the optimizer and also some tools to manage overfitting:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_fsLnJBChkaN"
      },
      "source": [
        "opt = tf.keras.optimizers.Adam(learning_rate=0.001)\n",
        "model2.compile(opt, loss='categorical_crossentropy', metrics=['accuracy'])\n",
        "lrd = tf.keras.callbacks.ReduceLROnPlateau(monitor = 'val_loss', patience = 20, verbose = 1, factor = 0.50, min_lr = 1e-10)\n",
        "es = tf.keras.callbacks.EarlyStopping(verbose=1, patience=20)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_cxAhn0bWb5j"
      },
      "source": [
        "Here we train the model defined above, splitting the initial training dataset into *training* and *validation*:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3vGWUvO9hkaP",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "967ec260-b881-4249-bffb-ad1be94bd2e6"
      },
      "source": [
        "history = model2.fit(X_train, Y_train, batch_size=BATCH_SIZE, epochs=EPOCHS, verbose=VERBOSE, validation_split=VALIDATION_SPLIT, callbacks=[lrd,es])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "387/387 [==============================] - 3s 7ms/step - loss: 2.1647 - accuracy: 0.2506 - val_loss: 1.8088 - val_accuracy: 0.2160\n",
            "Epoch 2/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.7198 - accuracy: 0.3142 - val_loss: 1.7796 - val_accuracy: 0.2680\n",
            "Epoch 3/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.6716 - accuracy: 0.3374 - val_loss: 1.7707 - val_accuracy: 0.2775\n",
            "Epoch 4/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.6587 - accuracy: 0.3443 - val_loss: 1.6603 - val_accuracy: 0.3481\n",
            "Epoch 5/50\n",
            "387/387 [==============================] - 3s 6ms/step - loss: 1.6344 - accuracy: 0.3572 - val_loss: 1.6068 - val_accuracy: 0.3744\n",
            "Epoch 6/50\n",
            "387/387 [==============================] - 3s 7ms/step - loss: 1.6244 - accuracy: 0.3649 - val_loss: 1.6419 - val_accuracy: 0.3711\n",
            "Epoch 7/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.6185 - accuracy: 0.3637 - val_loss: 1.6247 - val_accuracy: 0.3693\n",
            "Epoch 8/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.6046 - accuracy: 0.3667 - val_loss: 1.8125 - val_accuracy: 0.3154\n",
            "Epoch 9/50\n",
            "387/387 [==============================] - 3s 7ms/step - loss: 1.6141 - accuracy: 0.3674 - val_loss: 1.6110 - val_accuracy: 0.3711\n",
            "Epoch 10/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5960 - accuracy: 0.3695 - val_loss: 1.5794 - val_accuracy: 0.3962\n",
            "Epoch 11/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5824 - accuracy: 0.3804 - val_loss: 1.6231 - val_accuracy: 0.3736\n",
            "Epoch 12/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5777 - accuracy: 0.3793 - val_loss: 1.5834 - val_accuracy: 0.3882\n",
            "Epoch 13/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5765 - accuracy: 0.3805 - val_loss: 1.5971 - val_accuracy: 0.3853\n",
            "Epoch 14/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5680 - accuracy: 0.3849 - val_loss: 1.5834 - val_accuracy: 0.3878\n",
            "Epoch 15/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5620 - accuracy: 0.3893 - val_loss: 1.5796 - val_accuracy: 0.3871\n",
            "Epoch 16/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5633 - accuracy: 0.3894 - val_loss: 1.5768 - val_accuracy: 0.3937\n",
            "Epoch 17/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5513 - accuracy: 0.3909 - val_loss: 1.5901 - val_accuracy: 0.3878\n",
            "Epoch 18/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5470 - accuracy: 0.3926 - val_loss: 1.5947 - val_accuracy: 0.3911\n",
            "Epoch 19/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5429 - accuracy: 0.3942 - val_loss: 1.5660 - val_accuracy: 0.3926\n",
            "Epoch 20/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5389 - accuracy: 0.3979 - val_loss: 1.7228 - val_accuracy: 0.3631\n",
            "Epoch 21/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5442 - accuracy: 0.3935 - val_loss: 1.6210 - val_accuracy: 0.3664\n",
            "Epoch 22/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5225 - accuracy: 0.4020 - val_loss: 1.6221 - val_accuracy: 0.3733\n",
            "Epoch 23/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5231 - accuracy: 0.4031 - val_loss: 1.5576 - val_accuracy: 0.3995\n",
            "Epoch 24/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.5117 - accuracy: 0.4085 - val_loss: 1.6010 - val_accuracy: 0.3853\n",
            "Epoch 25/50\n",
            "387/387 [==============================] - 3s 7ms/step - loss: 1.5052 - accuracy: 0.4114 - val_loss: 1.5874 - val_accuracy: 0.3937\n",
            "Epoch 26/50\n",
            "387/387 [==============================] - 3s 7ms/step - loss: 1.5014 - accuracy: 0.4142 - val_loss: 1.5833 - val_accuracy: 0.3849\n",
            "Epoch 27/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4926 - accuracy: 0.4154 - val_loss: 1.5941 - val_accuracy: 0.3813\n",
            "Epoch 28/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4861 - accuracy: 0.4189 - val_loss: 1.6200 - val_accuracy: 0.3729\n",
            "Epoch 29/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4816 - accuracy: 0.4211 - val_loss: 1.5893 - val_accuracy: 0.3918\n",
            "Epoch 30/50\n",
            "387/387 [==============================] - 3s 7ms/step - loss: 1.4805 - accuracy: 0.4221 - val_loss: 1.5898 - val_accuracy: 0.3846\n",
            "Epoch 31/50\n",
            "387/387 [==============================] - 3s 7ms/step - loss: 1.4690 - accuracy: 0.4267 - val_loss: 1.6614 - val_accuracy: 0.3463\n",
            "Epoch 32/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4869 - accuracy: 0.4173 - val_loss: 1.6553 - val_accuracy: 0.3682\n",
            "Epoch 33/50\n",
            "387/387 [==============================] - 3s 7ms/step - loss: 1.4660 - accuracy: 0.4260 - val_loss: 1.5923 - val_accuracy: 0.3857\n",
            "Epoch 34/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4574 - accuracy: 0.4285 - val_loss: 1.6260 - val_accuracy: 0.3769\n",
            "Epoch 35/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4492 - accuracy: 0.4375 - val_loss: 1.5947 - val_accuracy: 0.3918\n",
            "Epoch 36/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4298 - accuracy: 0.4420 - val_loss: 1.6442 - val_accuracy: 0.3664\n",
            "Epoch 37/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4319 - accuracy: 0.4407 - val_loss: 1.6825 - val_accuracy: 0.3693\n",
            "Epoch 38/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4224 - accuracy: 0.4434 - val_loss: 1.6057 - val_accuracy: 0.3911\n",
            "Epoch 39/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4080 - accuracy: 0.4490 - val_loss: 1.6498 - val_accuracy: 0.3736\n",
            "Epoch 40/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.3963 - accuracy: 0.4560 - val_loss: 1.6220 - val_accuracy: 0.3889\n",
            "Epoch 41/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.4008 - accuracy: 0.4519 - val_loss: 1.6620 - val_accuracy: 0.3798\n",
            "Epoch 42/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.3818 - accuracy: 0.4635 - val_loss: 1.6365 - val_accuracy: 0.3791\n",
            "Epoch 43/50\n",
            "387/387 [==============================] - 2s 6ms/step - loss: 1.3682 - accuracy: 0.4678 - val_loss: 1.6711 - val_accuracy: 0.3824\n",
            "\n",
            "Epoch 00043: ReduceLROnPlateau reducing learning rate to 0.0005000000237487257.\n",
            "Epoch 00043: early stopping\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xUiukS73Z4B1"
      },
      "source": [
        "Here we plot the information about *loss* and *accuracy* of our trained model:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "Xuj78GxHaBVb",
        "outputId": "57057709-09d6-49cf-9c4f-96a183a35917"
      },
      "source": [
        "plotAccuracy(history)\n",
        "plotLoss(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3iV5fnA8e99kpAEMggZbAhLQDZEHKiAuLWOFkXrrpY6WrW1raPD2mprf20ddVGtaFXUKm7rZoqCMmTvFQgji5Edkpz798dzAoHscXKSnPtzXbmS877POefJgbz3+6z7EVXFGGNM8PIEugLGGGMCywKBMcYEOQsExhgT5CwQGGNMkLNAYIwxQc4CgTHGBDkLBMbUQkSSRURFJLQOZa8XkQXNUS9jmooFAtOmiMh2ETkkIgnHHP/OdzFPDkzN6hdQjGlOFghMW7QNuLL8gYgMA9oHrjrGtGwWCExb9DJwbYXH1wEvVSwgIrEi8pKIZIpIqoj8VkQ8vnMhIvJ3EckSka3ABVU893kR2SMiu0TkQREJaUyFRaSbiLwvIvtEZLOI/LjCubEiskREckQkXUQe8R2PEJFXRCRbRA6IyGIR6dyYepjgZIHAtEWLgBgRGey7QF8BvHJMmSeAWKAvMB4XOG7wnfsxcCEwCkgBJh/z3BeBUqC/r8zZwE2NrPPrQBrQzfd+fxaRM3znHgceV9UYoB/whu/4db7foScQD9wMFDayHiYIWSAwbVV5q+AsYB2wq/xEheBwr6rmqup24B/ANb4ilwOPqepOVd0H/KXCczsD5wN3qmq+qmYAj/per0FEpCcwDrhbVYtUdTnwb460akqA/iKSoKp5qrqowvF4oL+qlqnqUlXNaWg9TPCyQGDaqpeBHwLXc0y3EJAAhAGpFY6lAt19P3cDdh5zrlxv33P3+LpjDgD/ApIaUdduwD5Vza2mPjcCxwHrfd0/F/qOvwx8CrwuIrtF5P9EJKwR9TBBygKBaZNUNRU3aHw+8PYxp7Nwd9O9KxzrxZFWwx5cd0vFc+V2AsVAgqp29H3FqOqQRlR3N9BJRKKrqo+qblLVK3HB5q/ATBHpoKolqvqAqh4PnILrzroWY+rJAoFpy24EzlDV/IoHVbUM18/+kIhEi0hv4BccGUd4A7hdRHqISBxwT4Xn7gE+A/4hIjEi4hGRfiIyvh71CvcN9EaISATugv818BffseG+ur8CICJXi0iiqnqBA77X8IrIRBEZ5uvqysEFN2896mEMYIHAtGGqukVVl1Rz+mdAPrAVWAC8Ckz3nXsO1+WyAlhG5RbFtUA7YC2wH5gJdK1H1fJwg7rlX2fgprsm41oH7wD3q+oXvvLnAmtEJA83cHyFqhYCXXzvnYMbB5mH6y4ypl7ENqYxxpjgZi0CY4wJchYIjDEmyPktEIhITxGZIyJrRWSNiNxRRZmrRGSliKwSka9FZIS/6mOMMaZqfhsjEJGuQFdVXeabFrcUuERV11YocwqwTlX3i8h5wB9U9US/VMgYY0yV/JYF0TfNbo/v51wRWYdbILO2QpmvKzxlEdCjttdNSEjQ5OTkpq2sMca0cUuXLs1S1cSqzjVLOlxf6t9RwDc1FLsR+Lia508FpgL06tWLJUuqmxFojDGmKiKSWt05vw8Wi0gU8BYuN0uVeVBEZCIuENxd1XlVfVZVU1Q1JTGxyoBmjDGmgfzaIvDlPXkLmKGqxy7KKS8zHJdg6zxVzfZnfYwxxlTmz1lDAjyPGwx+pJoyvXCrNq9R1Y3+qosxxpjq+bNFMA6X1neViCz3HbsPXwIvVZ0G/B6XRvdpFzcoVdWU+r5RSUkJaWlpFBUVNUnFDURERNCjRw/CwiyZpTFtnT9nDS0ApJYyN9H4DT1IS0sjOjqa5ORkfAHFNIKqkp2dTVpaGn369Al0dYwxftYmVhYXFRURHx9vQaCJiAjx8fHWwjImSLSJQABYEGhi9nkaEzzaTCCoTWFJGXsPFlJaZunajTGmoqAJBIdKy8jILabED4HgwIEDPP300/V+3vnnn8+BAwdqL2iMMX4UNIEgxNfV4Y8GQXWBoLS0tMbnffTRR3Ts2LHpK2SMMfXQLCkmWoIQjy8Q+CHJ3j333MOWLVsYOXIkYWFhREREEBcXx/r169m4cSOXXHIJO3fupKioiDvuuIOpU6cCkJyczJIlS8jLy+O8887j1FNP5euvv6Z79+689957REZGNnldjTHmWG0uEDzwwRrW7q6cyUJVKThURnhYCKGe+g2EHt8thvu/V/3e5A8//DCrV69m+fLlzJ07lwsuuIDVq1cfnno5ffp0OnXqRGFhISeccAI/+MEPiI+PP+o1Nm3axGuvvcZzzz3H5ZdfzltvvcXVV19dr3oaY0xDtLlAUJ3yWTAu7bZ/Z8SMHTv2qPn3//znP3nnnXcA2LlzJ5s2baoUCPr06cPIkSMBGDNmDNu3b/drHY0xplybCwTV3bmrKqt3HSQxOoIusRF+rUOHDh0O/zx37ly++OILFi5cSPv27ZkwYUKV8/PDw8MP/xwSEkJhYaFf62iMMeWCZrBYRPB4xC9jBNHR0eTm5lZ57uDBg8TFxdG+fXvWr1/PokWLmvz9jTGmMdpci6AmIR6hzNv0gSA+Pp5x48YxdOhQIiMj6dy58+Fz5557LtOmTWPw4MEMHDiQk046qcnf3xhjGsNvW1X6S0pKih67Mc26desYPHhwrc/dlJFLqMdDn4QOtZY1df9cjTEtn4gsrS6pZ9B0DYFbS+CPFoExxrRmwRUI/NQ1ZIwxrZkFAmOMCXLBFwhUaW3jIsYY40/+3Kqyp4jMEZG1IrJGRO6ooswgEVkoIsUi8kt/1aVciEdQVSwOGGPMEf6cPloK3KWqy0QkGlgqIp+r6toKZfYBtwOX+LEehx1JPKd46plmwhhj2iq/tQhUdY+qLvP9nAusA7ofUyZDVRcDJf6qR0X+TDxXH1FRUQDs3r2byZMnV1lmwoQJHDtN9liPPfYYBQUFhx9bWmtjTEM0yxiBiCQDo4BvGvj8qSKyRESWZGZmNrgehwNBCxkw7tatGzNnzmzw848NBJbW2hjTEH4PBCISBbwF3KmqldOC1oGqPquqKaqakpiY2OC6+CsQ3HPPPTz11FOHH//hD3/gwQcfZNKkSYwePZphw4bx3nvvVXre9u3bGTp0KACFhYVcccUVDB48mEsvvfSoXEO33HILKSkpDBkyhPvvvx9wiex2797NxIkTmThxIuDSWmdlZQHwyCOPMHToUIYOHcpjjz12+P0GDx7Mj3/8Y4YMGcLZZ59tOY2MMf5NMSEiYbggMENV3/bnex328T2wd1WVpyJU6XuojPAwD3jqEQO7DIPzHq729JQpU7jzzju57bbbAHjjjTf49NNPuf3224mJiSErK4uTTjqJiy66qNq9gJ955hnat2/PunXrWLlyJaNHjz587qGHHqJTp06UlZUxadIkVq5cye23384jjzzCnDlzSEhIOOq1li5dygsvvMA333yDqnLiiScyfvx44uLiLN21MaYSf84aEuB5YJ2qPuKv96mPw9fgJu4ZGjVqFBkZGezevZsVK1YQFxdHly5duO+++xg+fDhnnnkmu3btIj09vdrXmD9//uEL8vDhwxk+fPjhc2+88QajR49m1KhRrFmzhrVr11b3MgAsWLCASy+9lA4dOhAVFcX3v/99vvzyS8DSXRtjKvNni2AccA2wSkSW+47dB/QCUNVpItIFWALEAF4RuRM4vqFdSECNd+6osnXXQTrHRNA5pmlTUV922WXMnDmTvXv3MmXKFGbMmEFmZiZLly4lLCyM5OTkKtNP12bbtm38/e9/Z/HixcTFxXH99dc36HXKWbprY8yx/DlraIGqiqoOV9WRvq+PVHWaqk7zldmrqj1UNUZVO/p+bngQqIWI+C3f0JQpU3j99deZOXMml112GQcPHiQpKYmwsDDmzJlDampqjc8//fTTefXVVwFYvXo1K1euBCAnJ4cOHToQGxtLeno6H3/88eHnVJf++rTTTuPdd9+loKCA/Px83nnnHU477bQm/G2NMW1JUKWhBv+lmRgyZAi5ubl0796drl27ctVVV/G9732PYcOGkZKSwqBBg2p8/i233MINN9zA4MGDGTx4MGPGjAFgxIgRjBo1ikGDBtGzZ0/GjRt3+DlTp07l3HPPpVu3bsyZM+fw8dGjR3P99dczduxYAG666SZGjRpl3UDGmCoFVRpqgE3puYSFeEi2VNS1sjTUxrQdloa6Ao8lnjPGmKMEXSAI9dN2lcYY01q1mUBQ1y4u25ymblpbl6ExpuHaRCCIiIggOzu7Thcv25OgdqpKdnY2ERFNO8XWGNMytYlZQz169CAtLY265CHKLSrhYGEpnoMR1a7yNS649ujRI9DVMMY0gzYRCMLCwujTp0+dyr60cDu/f38Ni39zJonR4bWWN8aYtq5NdA3VR2xkGAAHC5sl87UxxrR4QRcIYnyBIKfIAoExxkAQBgJrERhjzNGCNhDkWCAwxhggCANBTIS1CIwxpqKgCwSHu4YKLBAYYwwEYSBoF+ohMizEWgTGGOMTdIEAXKvAZg0ZY4zjz60qe4rIHBFZKyJrROSOKsqIiPxTRDaLyEoRGV3VazW12MgwaxEYY4yPP1cWlwJ3qeoyEYkGlorI56paccPd84ABvq8TgWd83/0qJjLUAoExxvj4c6vKPaq6zPdzLrAO6H5MsYuBl9RZBHQUka7+qlM51yIo9ffbGGNMq9AsYwQikgyMAr455lR3YGeFx2lUDhaIyFQRWSIiS+qSWK42MZFhto7AGGN8/B4IRCQKeAu4s6Eb06vqs6qaoqopiYmJja5TrAUCY4w5zK+BQETCcEFghqq+XUWRXUDPCo97+I75VWxkGLnFpbYvgTHG4N9ZQwI8D6xT1UeqKfY+cK1v9tBJwEFV3eOvOpUrX11srQJjjPHvrKFxwDXAKhFZ7jt2H9ALQFWnAR8B5wObgQLgBj/W57CKiefiOrRrjrc0xpgWy2+BQFUXADVuAaZub8nb/FWH6lgGUmOMOSI4Vxa3tz0JjDGmXHAGAmsRGGPMYUEZCCwVtTHGHBGUgcBaBMYYc0RQBoKIMA/tQjwWCIwxhiANBCLiSzNh+YaMMSYoAwFAbGSoLSgzxhiCOBDE2J4ExhgDBHEgsM1pjDHGsUBgjDFBLqgDga0sNsaYYA8EhSV4LRW1MSbIBW0giIkIw6uQd8imkBpjglvQBoLDq4sLrHvIGBPcgjYQxFiaCWOMAYI4EJS3CGzA2BgT7Py5VeV0EckQkdXVnI8TkXdEZKWIfCsiQ/1Vl6ocDgTWIjDGBDl/tgheBM6t4fx9wHJVHQ5cCzzux7pUEhPpNmezriFjTLDzWyBQ1fnAvhqKHA/M9pVdDySLSGd/1edYloraGGOcQI4RrAC+DyAiY4HeQI+qCorIVBFZIiJLMjMzm+TNo8JDCfGIBQJjTNALZCB4GOgoIsuBnwHfAWVVFVTVZ1U1RVVTEhMTm+TNRYSYiFBLRW2MCXqhgXpjVc0BbgAQEQG2AVubsw6Wb8gYYwLYIhCRjiLSzvfwJmC+Lzg0G0tFbYwxfmwRiMhrwAQgQUTSgPuBMABVnQYMBv4jIgqsAW70V12qYy0CY4zxYyBQ1StrOb8QOM5f718XMZFh7NpfGMgqGGNMwAXtymKwVNTGGAMWCDhYWIKqpaI2xgSvoA4EMRFhlJQphSVVzlo1xpigENSBwFYXG2OMBQLAAoExJrhZIABbXWyMCWoWCLAWgTEmuFkgwAKBMSa4BXUgsD0JjDEmyANBdIS1CIwxJqgDQYhHiI4Ite0qjTFBLagDAfjSTFggMMYEMQsEloHUGBPkgj4QxERYIDDGBLegDwTWIjDGBLvgCQSqsGNRpcOtMhX1nL/AsxMCXQtjTBvht0AgItNFJENEVldzPlZEPhCRFSKyRkRu8FddAPjuZZh+DiyZftTh2PatsEWQ+hXsXg4lRYGuiTGmDfBni+BF4Nwazt8GrFXVEbgtLf9RYQ/jpjd8Cgw4Bz78+VHBIDYyjKISL8WlrSgVddYmQGH/tkDXxBjTBvgtEKjqfGBfTUWAaBERIMpX1n/Z30LDYcrLlYJBTEQrW11cdBDy9rqfs7cEti7GmDYhkGMET+I2sN8NrALuUFVvVQVFZKqILBGRJZmZmQ1/xyqCQczhDKStJBBkbT7y8z4LBMaYxgtkIDgHWA50A0YCT4pITFUFVfVZVU1R1ZTExMTGvesxwWDQrpkAHGwtqaizNrrv4rEWgTGmSdQpEIjIHSISI87zIrJMRM5u5HvfALytzmZgGzCoka9ZNxWCwcDFv+eHIbNaT4sgexN4QqHbKNi3NdC1Mca0AXVtEfxIVXOAs4E44Brg4Ua+9w5gEoCIdAYGAs13ZfMFg/zek/hz2PPErXul2d66UbI2QlwfSBhoLQJjTJOoayAQ3/fzgZdVdU2FY1U/QeQ1YCEwUETSRORGEblZRG72FfkTcIqIrAJmAXeralb9f4VGCA2n8NIXmVU2ipErHoAZl8OSFyBnT7NWo16yNkHCcRDfF3J3w6H8QNfIGNPKhdax3FIR+QzoA9wrItFAlQO75VT1ylrO78a1MAIqNjqKW0ru5NXes0nJnAubPnUnuo2Ggee5r85DQWqMe82jrNS1Ao47Fzr1c8f2bYUuwwJbL2NMq1bXFsGNwD3ACapaAITh+vhbvbAQD6HtIviky81wxwq4dRFM+j14QmDOn2HaqfDYMFj3YaCrCgdSwVviaxH4AoF1DxljGqmugeBkYIOqHhCRq4HfAgf9V63mdTjfkAgkDYbT7oKbvoBfboSLnnSDs/MaOyTSBLI2ue8Jx0Gnvu5nm0JqjGmkugaCZ4ACERkB3AVsAV7yW62aWbWJ56KSYPQ1MOpq2LsK8pt3CKOS8qmjCf0hPBqiOkO2zRwyxjROXQNBqaoqcDHwpKo+BUT7r1rNq9ZU1H0nuO/b5jVHdaqXtRE6JEJknHsc399aBMaYRqtrIMgVkXtx00b/JyIe3DhBmxBTWyrqriMhPBa2zm22OlWpfMZQuU59bYzAGNNodQ0EU4Bi3HqCvUAP4G9+q1Uzi40MI7eohpXFIaHQ5zTY2gJaBAkDjjyO7wf5GVCUE7g6GWNavToFAt/FfwYQKyIXAkWq2vbHCCrqO8HN2tkXoIyf+dlQuO+YFkGFKaTGGNNAdU0xcTnwLXAZcDnwjYhM9mfFmlNsZBh5xaWUltWwNKLPePc9UN1D2b4ZQ/HHtAgAsjdXLm+MMXVU166h3+DWEFynqtcCY4Hf+a9azSsm0q2ry6mpeyhhAER3C9yA8eEZQxUCQVwf991aBMaYRqhrIPCoakaFx9n1eG6LF+tLRV1j95CI6x7aOg+8NS6q9o+sjRASDh17HTnWrj3EdLcBY2NMo9T1Yv6JiHwqIteLyPXA/4CP/Fet5hVb1z0J+o53/fTpq5qhVsfI2uSmi3pCjj7eqa9NITXGNEpdB4t/BTwLDPd9Pauqd/uzYs2pTi0CCOw4wbEzhsrF97MWgTGmUercvaOqb6nqL3xf7/izUs2tzoEgpiskDmr+aaSlxbA/9egZQ+U69XOtlML9zVuntuq7GfDZbwNdC2OaVY2BQERyRSSniq9cEWkzk9c7tm8HwNbMOqR07jsBUr92F+fmsm8baFn1LQKwVBNNZdlLsPApC6wmqNQYCFQ1WlVjqviKVtUqt5VsjRKjwxl/XCLPfbmV9Jyimgv3GQ+lhbDz2+apHFQ9Y6jc4bUE1j3UaF4vpK8B9cKW2YGujTHNxm8zf0RkuohkiMjqas7/SkSW+75Wi0iZiHTyV31q88BFQzhU5uXB/62ruWDyOJCQ5h0nKA8E8VUFgj6A2DhBUziwHQ7lup83fRHQqhjTnPw5BfRF4NzqTqrq31R1pKqOBO4F5qnqPj/Wp0bJCR24ZXw/Plixm68215BlNCIWuo9p3vUEWZvcNNHwqMrnQsOhY09rETSFvb7ZYPEDYPPngZkmbEwA+C0QqOp8oK4X9iuB1/xVl7q6ZUI/enVqz+/fW82h0houAn0nwK6lUNRMWzJkb6q6W6hcp362urgp7F3lWnsn3wb5mbB3RaBrZEyzCPiiMBFpj2s5vBXoukSEhfDARUPYkpnPvxfUMPjad7zrR96+wP+VUvWtIaghEMT3c4PFqv6vT1u2d7WbmTXoQvd40+eBrY8xzSTggQD4HvBVTd1CIjJVRJaIyJLMzEy/VmbioCTOGdKZf87aRNr+gqoL9TgBwto3zzTSvHQozql66mi5Tv2g+CAUZPu/Pm3Z3lXQZShEJbo9qy0QmCDREgLBFdTSLaSqz6pqiqqmJCYm+r1Cv//eEAThjx+srbpAaDj0PqV5BoxrmjFUzvYvbryCfZCTBl2GuccDzoK0xe64MW1cQAOBiMQC44H3AlmPY3XvGMnPJvXns7XpzFmfUXWhvhMgawPk7PZvZQ4HglpaBGADxo2R7pvcdjgQnA2oTSM1QcGf00dfAxYCA0UkTURuFJGbReTmCsUuBT5T1Tqs5GpeN53al36JHbj//TUUlZRVLnA43YSfu4eyNkFYB4jpVn2ZuN5ukNNaBA1XPmOosy8QdBsFkZ1g02eBq5Pxr02fw1s3Ne/i0BbKn7OGrlTVrqoapqo9VPV5VZ2mqtMqlHlRVa/wVx0ao12ohz9dPJQd+wp4em4VF9jOQ6F9vP+nkWb5ZgyJVF8mJMxlJbUWQcPtXQVRXdz4ALjkfv3PhM1f2DTStih7C7x5A6x6EzZ+EujaBFxLGCNosU7pn8BFI7oxbd4Wtmcd02jxeFyrYOtc/87Wyapl6mg5Sz7XOHtXH+kWKjfgLDcAv/u7wNTJ+EdJIbxxrQv2HZJcfqnWIGeP3641Fghq8dsLBtMuxMMPnvmaRz/fSFZehWZk3/GQu+dIP35TO1QAB3fUPD5QrlM/t0GNTSGtv9JDkLnezRiqqN8kQNziMtN2fPQrNyb0/Wdh1NXu3zdnT6BrVbOc3fDcRPj89355eQsEtUiKiWDGTScyqldHHp+1iVMens3dM1eyYW+uGzAG/80eKl8kVtcWwaE8yKtmcNtUL3M9eEsqtwg6xLtV5DZO0HYsfxW+exlO/QUcdw6MvMqtCVr5eqBrVr1D+fDaFVCcC8On+OUtLBDUwYieHfn3dScw667xXJ7Sg/dW7OKcx+ZzzdvpFEb1RP0VCOoyY6hcJ9u/uMEOzxgaXvncgLNh1zLIryHtiDli8fPw5T9gyxwoPBDo2hwtfQ18+AvofSpM/I07ltAfep3suoeaozWduhDe+2ndpyV7vfDOT9wY1uTplVutTcQCQT30S4ziwUuGsfCeSfzqnIFs2JvLzAMDObRhFnvStjX9G2ZvBuTIRb4m8TaFtMH2roLQSLfb27EGnAkobJ7V7NVqdbK3wP9+AbP+CC9fAn/tDU+MgbenwqJpsHOx64YLhOJceOM6CI+Gyc9DSOiRcyOvcmlc/J1ROC/DjU189zL8+8y67TU++0+w7gM4+0HXgvETCwQNENehHbdN7M+Cu88g/qy78FDKwud/ydLUJs5hn7XRzQYKi6i9bGxP8ITZgHFD7F0FnYdU3gYUoOsoaJ9g4wR1sfQF8ITCbd/CNe/CGb87spHTJ3fD82fCf77X/ONYqvD+7e4mafJ0iO5y9Pkhl7gp2stf8V8dyu/si3PhoifcZlL/PgvSllT/nOWvwYJHYMz1cNKt/qsbFggapV2oh/PHn0L+8Ou5WGfz22ff4u1laU33Blkb69YtBO4OJy7ZWgT1pQp7V1YeHyjn8VSYRlrFehLjlBS57pVBF0DiQOg3EU7/JVwxA365AX6xDibcCzsXwZZmbl0t/jeseRvO+C30Oa3y+fBoFwxWv+P642vj9cJnv4O179e9DgufcIsTz/0LjL4WbvzCZRN+8UJY92Hl8qlfw/s/gz6nw/l/r3n6eBOwQNAEOp7zGzztOvCn6Df5xRsrePjj9Xi9jbzr8Xoha3PdAwEcST5n6u5gmssiW1Pf64Cz3I5lu5Y1X71am7XvubvclB9VfT6mmxugje4GCx5rvnrtWgqf3OvGesb9vPpyI69ye1HU5eK++N/w9T9dN8/S/9RePm2J6y47/hJ3dw9ubOLGL1xL9L9Xwzf/OlJ+3zZ4/Sq3UPTyl9w6IT+zQNAUOsQjp/2clKJF3DdkH9PmbWHqy0vJKy6t/jmF++GtH7tZDFUtWMpJczuh1WXGULnyKaS2AKruylcUVzVQXK7fGSAemz1UkyXT3f+/5NOrLxPaDk6+FbZ/6S7Q/lZWAu/cAlGd4dJ/udZddXqf4saIlteypmDfNvjiD9B3IvSfBB/cDgufrr580UGY+SMXAL/3+NF39lGJcN0HrhX18a/h09+468KrU9xMph++AZFx9fqVG8oCQVM58RaI7saPi17kDxcOZvb6dCY/83XVGUzLSuDN62HVG/DuLTD9nMqLlrI2ue/1ahH0dcEjt4XPiW5J9q4CBJKOr75M+04u46yNE1QtfY3r8kn5Uc0XW3B3xBGx8NXj/q/XN/9y+cAu+Lv7N6yJCIz8oQtS+6qZ+OH1uu4a8cDFT8IVr8Lg78Gn98L8v1Ue+1CFD3/uWp2Tn4fIjpVfs117d9d/4i2w8En452jXvTvllSMTQJqBBYKm0q49nPEbZNcSro9byYs3jGXXgUIufGIBf/pwLct3HkDL/6N8cq9be3DRE3Dx07B/Gzw70Q1olU9TPBwI6tkiABsnqI/0Ve5OsKrd3yrqf5YL1rZOo7Il0yEk3F1IaxMeDSfc5Lpg/DmxITcd5j7suoSOq3ajxKONuBIQ10qvytIXXKA450GI7eGyEE9+0c3tn/0gzHrg6GDw3Suw+i2YeB/0HFv9+3pC4LyH4Zy/uLVAFz5a9ViGH1kgaEojrnR3lrMe4PS+sbx72zhOSO7EywtTueSprzj9b3P47MUHYfFz6Mk/dYNGo66Cny11u2ItnwFPjHZ3Mhlr3Z1Th3qk3bZ01PW3d1X1A8UVDTjLfbdppEcrzoMV/4Uhl9Z+113uxJshpJ3rZ/eXL/4AZcVw7sN1H2iN7eG6AaBQB6cAAB8nSURBVJe/WnliwIEdblVv3wkw+rojx0NC4ZJpMOYGWPCo6+LxeiFzg/u5z+lwag1jExWdfCvcu8tdF5qZBYKm5AmBMx9w/fRLX6RfYhTPXZvC4t+eyd8mD+fCqI2cse0fzCobxdmrzuDRzzeyaGs2ubSHcx6CW752WS8//jUs+4/rFqrPbIGYHu7OrKEtAm9ZcKWoKMqB/dvrFgi6DHd5aax76Gir3nSDrCfcWPfnRCW51sPy19yde232b4fXfghpdRxX2PENrHjV3VzVt3tl1FVufK5iMsny6afgWvHH/k16PO4u/uSfwrfPwvs/deMCYZFw6bNVT0uuTmi7+tW3iYTWXsTUy4CzIPk0mPdXGHEFRMQQGxnGZcnF8MWfKU04jqxRTxG/Nod/zt7E47NcF1DfhA4M7R7LsN6PcHrPb+i/6h+E9JtUv/f2eKBTn4a1CA7lw3OTIGmwm2vt5+lqLUL6Gve9LoHA43H/tmvfh3dvhZICl7ys/Ku00N0JXvwkdB/t33o3hdy97t+8Mf3Qqq5bqPNQN4ZSH6f8DJa+CN9MgzPvr75cwT54ZbJb8JX6FdzwMXSuYTzHWwYf/8oNzp72y/rVCWDgBa4l/t0M1zoAWPYSbJ0DF/zDreupiohb9NUuCuY97I798E2I6Vr/OgSAtQiamgic9UcoyDrS9C3cD69NAU8ooVf/lymnDuH1qSez5Ddn8sINJ3DXWcfRPymKJdv38dDH6znn01j67f4jUzZOYN7GzCNjC3XRqYFZSL/4A2Suc/Otv322/s9vjQ7PGKpDIAAX2MOjYNt8l600ZzeUFrnxoZgeblBw3v/5r75NJW0JPHMKTDsVdixq+OvsWubWYKTcUP8bh/h+cPxFLiVFUU7VZUqL3dTKA6nuzjosEl6+tPrBXHAX7T0r4Ow/1T7uU5WwCBh2uVvNW7jf/Zt++ht3czemmqmx5URg4r1w0ZNwwSNw3Nn1f/8AsRaBP3QfDUN/AF8/6foT3/8p7E+F6953i7584qPCmTgwiYkDkw4fy84rZtWugyzfeYDXv93JddO/ZWj3GG6b0J9zhnTB46n6D66kzMvXW7KR7BhOztrKl2t3c8bxNWxmU9G2L93F/8SbXTP8s99Cr5Og64hGfAitwN6Vbk+J6DretfU5He5aX/35OX92LcG6pg4PhA0fuzz80Z3dne+My9wUxm4j6/9aS553d8ANTYQ27k63/mDpizDu9qPPeb3w3m2uFfCD52HYZPf/8YVz4aWL4UefVr7bLtjn5uv3Huf+/hpq1FWw+Dk30LvhY9Ay1yVU24yocqOvafh7B4g/dyibLiIZIrK6hjITRGS5iKwRkWbYCb4ZnfE78JbCc2e4GUIXPurmKtciPiqcCQOTuPPM45j36wn89QfDyCsq5ZYZyzjr0Xm8tTSNkjK3TqCopIzP16bzizeWM+ZPn3Pd9G/5LCOGMEp45ZXnufnlpew5WFjzGxbnwXu3upkzk+53s5jax7uLRXFuE3wQLVj6atet0VTdYCf82I3RLHyyaV6vqS2ZDq//EJIGwY2fw7Xvu2Dw8qWQsa5+r1W4310oh13mZgI1RPfRLrguerryLmFzHnLjD5N+74IAuHpf/ZbbI+LlSysnbpvzZyg6AOf9tXH/pl1Huv8XXzzgVpSf+YDrcm3D/Nk19CJQ7bwtEekIPA1cpKpDgMv8WJfm16mPmyaXn+EGkRpwlxAeGsKUE3ox664JPHHlKNqFhnDXmyuY8Le53PzyUkb/6XN+/NISvlibzpnHd+a5a1P47a/vxdt5KP+KeJIDGxdw5j/m8fyCbZSWVbPI7Iv74cBOFwDatXepl3/wbzel9X8N6GNtLcpKIX1t3buF6iIqEUZe6QZB8zKb7nUbS9VNb/zw5y5dxnUfugHbjj3h2vfcytWXLqlfl+Ly11y3WH0Giasy7g637mXVm0eOLXsJvvy7a02f+oujy3cfA1e+5iZkzJh85GZl7yrXQkm5sfH/piJupXFxjmtdnHBT416vNVBVv30BycDqas7dCjxY39ccM2aMthrF+apr3lMtK22Sl/N6vTp7XbpOfuYrHfvQ53rPWyt13oYMPVRadnTBnL2qj43Q0j/30vumva697/5Qz3tsvi5L3Xd0uS1zVe+PUf34nspvNucv7tx3rzZJ3Vuc9LXu91v+etO+buZG97qzH6r7c7bMVf3qn6oF+2ovW1+lh1TfucXV6d3bVEtLKpdJX6v6cLLqI0NU9++o/TW9XtV/jlF9blLj6+f1qj49TvWJFNWyMtVNX6j+IU71pUtd3auz7n+u3IsXqh4qVH3+XNW/9lHNz258nVRVC/arvv0T1X3bmub1WgBgiVZzXRX143RBEUkGPlTVSolcROQxIAwYAkQDj6vqS9W8zlRgKkCvXr3GpKam+qvKbcf+7TD9XFSVeeNe5u45OWTkFvPDsb34+VnHkRBWAs+c7DKW3rzAtQYq8pbBfy5yi6h+Mq/6Pm+vFzZ+7JJkDZvspr+2BF5vzX26K9+At3/spux2HtK07/3qFZD2Ldy5uvLneqyc3fDUSVB80GXAHHO9m08e26Px9ShPvbxllkv4Nv7u6rtMdi93mUE7JLqZOdGdq3/dbV/Cfy6ES56p2yKy2qx8E96+CSb+1q04juvt6hARU/PzVvwX3pnqMpxmrncpHMpz+ZhKRGSpqqZUeS6AgeBJIAWYBEQCC4ELVLXGfR9TUlJ0yZIaUreaI9LXwgvnQWQc+Vf/j398fZAXv95GWIiHlzr/l7HZ7yI/+sQNDFclZzc8Mw5iusNNXxydDrusBFbNhK8ec3+ECKAuB8tpv3CzLPw9BVUV8tJdltasjW6QNnOD+56f4S5+p/2i6ud+9js3dfG+3U2f1Gv7V/Di+W7mSE1dJ6queyP1azdld/Xbrt9dxPW9j7vDTeetr9y9rqtlyQvuhuB7j9VtkdKOb9w+AnHJcP3/ql8g9ub1buOZu9a7mTyNVVYKT4xyi7aiu7n/a7Hd6/bcb/7l1t10HQk/nl2/OftBpqUGgnuASFW93/f4eeATVX3z2LIVWSCop52L3SyLTn3h+g/ZnBvK7I/eYOr2n/Pv0vNZOeTX3Dy+H8d3q+bua+On8OrlMHYqnP83t4/ydy/D10/AwZ2QNMStnOw/yS2CW/i0uwh3H+P6dweeX/fZFnWVvcUNyK5+yyX1KtcuyrVcEo5zA4mbP4cJ98GEuyu/xkuXuGyZP5nftHUDd4F/7gxXt58urv7itOwll7vmvL/BiVPdsQM7YOFT7lxJgUuPcOJPoMsId2GuLriWFML6/8GK110LQL3QPcWlN+hfj/UoW+fCjMtdttC4ZDfhofyrrMS1FDPWuhlm5/65Pp9KzVa8Dp/fD1fPrH8f/8bP3NqCpmhFtWEtNRAMBp4EzgHaAd8CV6hqtbOMwAJBg2yZ7f64u4+BKS/Dc5Mo9YTxSL/n+c+36eQfKuP04xK5ZXw/TurbCTn2YvPJfbDoKXdXuf4jt0ai50nubnvA2UdfnEoK3RL9rx53878TBrqpgcmn+lY+N2LG8s7F8PXjLn97SBgM+b77nRIGuBz40V2P1MVb5qYfrngNTv+1uyCWn1OFv/V3F9lLnmp4fWqy+m2YeQNMmQGDL6x8/sBOePpkN23z2vcrB8uCffDtc/Dtv9wsGYDwWOiU7IJ6+Vf7eNjwEax51w1uxvSAEVNg+BWQWI+EhRVt+hzm/90Fk5AwF8g8ob6vMNcKOOuPdb9rryvV4FjIGCABCQQi8howAUgA0oH7cWMCqOo0X5lfATcAXuDfqlpronILBA205l13YQqPcXeqP/oUep3IwcISXlmUygtfbSMr7xAJUeEMSIqif1IUAzpH0T8xiv7xYSS+cRGyZ7m78J/689qnwpaVwtp3Xf6V8j2BJcTdtcUlu37guGTo2Nsdi0py6YLbdTj6dbxe2PiJW5y3YyFEdHTdLWN/UnM/Nrhg8MHtLvnXaXe5Kb0iruvkHwPh3L/CSTc39BOtWXl3R0x3+NEnR59TdV0waUvcGEVc7+pf51CBW8C2b+vRXwd2uPnt4MYWjr/I5bpKPq3pW2CmTQhYi8AfLBA0wtIX4YM73PL+sx886lRRSRnvLd/Fku372ZSRx5aMPHIr7KfQPaKYwTGHOBDZi8h2IUSEhRBZ/tUuhI7twzilXwKje3UkNKTChUgVdn7j+vD3b3cL6/Zvd62F/CqmWLaLcgEhqrMLDulrXHqB2F5uEHXUNfVbMer1wod3um6rcXfCmX9wc8NnTIbrP4LkcfX4AOtp0TPwyT1uA5KeFVIwLH7e7e174aPVb+RSm7IS1zWXs9v1jzdkFa0JKhYIzBH7trm78FruGlWVjNxiNqXnsTkjl00ZeaTnFFFYUkbhoTIKS7wUlZRRcKiUwkNl5BWX4lWIiQjl9OMSOWNQEuOPSyQ+Krz6NynOc3e2ubtdeue8dJeELC/9yOPIjq4/+vhLGt6t5PXCR79088xP/qnra5/1R7g7teoc8U2lOBceHeIyVl7umxC3fzs8fYpLS3zNO9YVYppNTYHAUkwEmzqukBQROsdE0DkmglMHJNRaPqeohAWbspizPoM5GzL5cOUeRGBEj45MHJjECclxDO0RS0xEhRk64VFukK+mJGJNweNxCcM8oW6QOTzGJQ/zZxAAt+I25UduvGTfVuiYDO/e5vrcL37SgoBpMSwQmCYRExHG+cO6cv6wrni9yprdOcxen8GcDRk8Nmvj4ezWfRM7MKJHR4b3iGV4j44M6RZDRFgIqkphSRn5xa6VUf49ISqc5IQONb95XYi41AOeEJfSILmZNv4Y+xOXc2rRMxDfH1IXuKRkNsPFtCDWNWT87kDBIVakHWTlzgPue9oBMnJdbplQjxAe6qGgpKzarRBO6tuJa05K5uwhnQkLaeRAqKqb1dR1eNOml6jJu7fCmnfce/c5ze1Fa60B08xsjMC0OHsPFrEi7QCr0g5SWFJGh3YhtA8Pdd/bhdIhPITIdqGs3Z3DjG9SSdtfSFJ0OFeM7cWVY3vSNbYJFjI1l/Q1Lu1zRCzcusjN0TemmVkgMK1amVeZvzGTlxelMmdDBh4RzhycxJQTenJ811g6x4RXXvvQ0iya5sZC+pwe6JqYIGWBwLQZO/cVMOObHbyxZCf78g8BEBkWQnJCB/oktCc5vgN9EtxXckIH4ju0a/lBwphmYIHAtDnFpWUs3rafbVl5bMsqYFtWHtuzC9i5r4BS75H/09HhofROaE/v+A70ie9A7/j2JCe474lRdWtJHCr1sjUrj/V7cik4VMYPxnQnPNRy2pjWxQKBCRolZV527S9kW1Y+27PzSc0uYFtWPqnZ+ezcX0hZhSAREeahV6f29OrUwfc9kl7x7RGE9XtzWb83hw17c9mckXdUcBnaPYYnrxzdNLOZjGkmFgiMoUKQyM5n574CUrML2LGv4PDPhSVlR5XvFhvBoK4xDOwSzaAu0QzuGsO2rHx+PXMlZV7lz98fxkUjbODXtA62oMwYICzEQ7Jv7OBYqkpW3iF27CugzKsM7BxNbPvK6amP6xzN0O6x3P7ad9z+2ncs3JLN/d87nogw6yoyrZcFAmNwK6kTo8NJjK4hJYZP946RvD71JB75fCPPzN3CstT9PHXVKPonNXDvXmMCzLqGjGmEuRsyuOuNFRQcKuOBi4ZwUt94DpV5KSnzUlqmHCrzUlrmpdSrHNc5uk6Bxhh/sDECY/woPaeIO17/jkVb99VYLsQjTBqUxBVje3L6gMSjs7Qa42c2RmCMH3WOiWDGTSfx+dq95BeXERbqoV2IEOrxEBbqIcwjIDBvYyZvLU3js7XpdI4J57IxPbk8pSe94mvZ19gYP7MWgTHNqKTMy+z1Gfx38U7mbsjAq3By33guHtmNlOQ4+iZE4fHYAjjT9AK1Q9l04EIgo5qtKicA7wHbfIfeVtU/1va6FghMW7HnYCFvLU3jv0t2snNfIQCxkWGM6tWR0b3iGN0rjhE9Y4mOqDx7yZj6ClQgOB3IA16qIRD8UlWr2NC1ehYITFvj9Spbs/JYlnqAZTv2s2zHfjam5wEuSWn/xCiSE9yit55xbtFbr07t6RHX3qatmjoLyBiBqs73bV5vjKmBxyP0T4qmf1I0l5/QE4CDhSUs33mAZan7WbP7IKnZ+Xy5KZOiEu9Rz+0cE87JfeOZNLgz4wcmHr3xjzF1FOjB4pNFZAWwG9c6WFNVIRGZCkwF6NWrVzNWz5jAiI0MY/xxiYw/LvHwsYqL3nb6vjZl5DFvYybvLt9NqEcY26cTkwZ35szBSfSOtxQYpm78OljsaxF8WE3XUAzgVdU8ETkfeFxVB9T2mtY1ZMzRyrzKdzv288W6DGatS2dThutW6p8Uxbh+8Yzu7cYbesRFWibWIBawdQQ1BYIqym4HUlQ1q6ZyFgiMqVlqdj6z1mUwe30GS1P3H86hlBgdzphecYzu7Qajh3aPbdYxhszcYopLy+gRZ9NlA6FFriMQkS5AuqqqiIwFPEB2oOpjTFvRO74DPzq1Dz86tQ+lZV7W783lux37WbbDDUZ/smYv4Aaiu3eMpF9iFH0TOxz+3j8xisToptvsp6TMy4tfbefRLzZS6lV+c/5grj25t7VOWhC/BQIReQ2YACSISBpwPxAGoKrTgMnALSJSChQCV2hrW9RgTAsXGuJhaPdYhnaP5ZqT3bGsvGKWpe5n3Z5ctmblsSUzj8Xb91Fw6Ej21fBQD+3bhRAR5r7CQz2+nz1EhYdyxqDOXDSyG1HhNV9Clqbu4zfvrGb93lzOGJSEqnL/+2uYuyGDv102goQoS7nREtiCMmMMqsrenCK2ZOSzJTOPXQcKKTxURlFJGUWlXvfd95WRW0xqdgHt24Vw0YhuXDm2F8N7xB51h78//xAPf7ye/y7ZSbfYCO6/aAhnH98ZgJcWpvLQR+uIiQjlb5eNYOLApED92kHFcg0ZY5qMqrJsxwFe/3YHH67cQ2FJGYO7xnDl2J5cPKI7n67Zy18+XkduUSk3ntqH2ycNoMMxLYcNe3O5/bXv2JCey/WnJHPPeYNsTYSfWSAwxvhFTlEJ7y3fzevf7mDN7hw8Al6Fscmd+NMlQxnYpfrU3EUlZTz88Xpe/Ho7g7pE8/gVo2osbxrHAoExxu9WpR3k/RW7GNQlhu+P7l7nweA5GzL41ZsryM4/xMieHZk0KImJg5I4vmuMDSg3IQsExpgWLSuvmBmLdjB7fTor0g4C0DU2gomDkjhjYBLj+icQ2c66jhrDAoExptXIyC1i7oZMZq/L4MtNmeQfKiM81MOJfeMPr7bul9jBWgv1ZIHAGNMqFZeWsXjbfmatT2f+xky2ZOYDbv3D+IEuKJzSL94ytNaBBQJjTJuwc18B8zdlMm9DJl9vySavuJRQjzB5TA/uu2CwJd2rgQUCY0ybU1LmZVnqfj5atYeXF6XSOSaCP39/mK1LqEZNgcA2TTXGtEphIW7c4IGLh/L2reOICg/lhhcW86s3V3CwsCTQ1WtVLBAYY1q9kT078sHPTuXWCf14+7tdnP3oPOaszwh0tVoN6xoyxrQpK9MO8Ms3V7AxPY8fjO7BT8b3Jb+4lAMFJewvOMT+ghIOFBxif8EhBOGMQUmc0j+e8NC2PT3VxgiMMUGluLSMJ2Zt5pl5WyjzVr7GiUDHyDAOlXrJP1RGdHgokwYnce7QrkwYmNgm011YIDDGBKUNe3NZs/sgHduH0bF9O+LatyOufRgxEWF4PEJxaRlfb87m49V7+GxtOgcKSmjfLoSJA5M4b1gXzj6+C+1C20YPugUCY4ypRUmZl2+27uPj1Xv4dE06WXnFdIuN4JaJ/bk8pUer7zqyQGCMMfVQ5lXmb8zkn7M38d2OA3SNjeCWCf24PKVnq+02skBgjDENoKos2JzF419sYknqfjrHhHPz+H5cObZXqwsIAVlHICLTRSRDRFbXUu4EESkVkcn+qosxxjSEiHDagETevPlkXr3pRHrHd+CBD9Zy2v/N4cnZm8jIKQp0FZuE31oEInI6kAe8VN3m9SISAnwOFAHTVXVmba9rLQJjTCAt2prNk7M3s2BzFiEe4czBSVw5thenD0jE42m5ifACsnm9qs4XkeRaiv0MeAs4wV/1MMaYpnRS33hO6hvPtqx8Xv92B28uTePTNel07xjJlWN7cllKTzrHRJBfXMrWzHy3L3RGHluy8tmSkYcq3HP+oBaVCsOvYwS+QPBhVS0CEekOvApMBKb7ylXZIhCRqcBUgF69eo1JTU31V5WNMaZeikvL+HxtOq9+s4Ovt2QT4hESotqRnlN8uIxHoGen9vRLjCI1O58tmfl8f1R3fnfh8cR1aNcs9QxIi6AOHgPuVlVvbXnFVfVZ4FlwXUPNUDdjjKmT8NAQLhzejQuHd2NbVj7/XbyTjJwi+iVF0TehA/2Sougd3/7w9NPi0jKemr2Zp+duYf6mTB64aCjnD+sS0P0VAtki2AaU/+YJQAEwVVXfrek1bYzAGNMWrN2dw91vrWTVroOcM6Qzf7p4KEkxEX57vxaZfVRV+6hqsqomAzOBW2sLAsYY01Yc3y2Gd249hXvOG8ScDZmc+cg83liyE28VKTH8zZ/TR18DFgIDRSRNRG4UkZtF5GZ/vacxxrQmoSEebh7fj0/uOI2BXaL59cyVnPPYfGYuTeNQqbfZ6mELyowxpgXwepUPVu7mmblbWL83l66xEdx4ah+uHNuLDuGNH861lcXGGNNKqCpzN2Yybe4Wvtm2j9jIMK47uTfXnZJMfFR4g1/XAoExxrRCy3bsZ9rcLXy2Np2IMA+/PHsgN53Wt0Gv1VKnjxpjjKnB6F5xPHttCpsz8nh2/hZ6xEX65X0sEBhjTAvXPymK/5s8wm+v3zZ2XDDGGNNgFgiMMSbIWSAwxpggZ4HAGGOCnAUCY4wJchYIjDEmyFkgMMaYIGeBwBhjglyrSzEhIplAQ7coSwCymrA6bZF9RjWzz6d29hnVLFCfT29VTazqRKsLBI0hIkuqy7VhHPuMamafT+3sM6pZS/x8rGvIGGOCnAUCY4wJcsEWCJ4NdAVaAfuMamafT+3sM6pZi/t8gmqMwBhjTGXB1iIwxhhzDAsExhgT5IImEIjIuSKyQUQ2i8g9ga5PSyAi00UkQ0RWVzjWSUQ+F5FNvu9xgaxjIIlITxGZIyJrRWSNiNzhO26fESAiESLyrYis8H0+D/iO9xGRb3x/a/8VkXaBrmsgiUiIiHwnIh/6Hre4zycoAoGIhABPAecBxwNXisjxga1Vi/AicO4xx+4BZqnqAGCW73GwKgXuUtXjgZOA23z/b+wzcoqBM1R1BDASOFdETgL+Cjyqqv2B/cCNAaxjS3AHsK7C4xb3+QRFIADGAptVdauqHgJeBy4OcJ0CTlXnA/uOOXwx8B/fz/8BLmnWSrUgqrpHVZf5fs7F/TF3xz4jANTJ8z0M830pcAYw03c8aD8fABHpAVwA/Nv3WGiBn0+wBILuwM4Kj9N8x0xlnVV1j+/nvUDnQFampRCRZGAU8A32GR3m6/ZYDmQAnwNbgAOqWuorEux/a48Bvwa8vsfxtMDPJ1gCgWkAdXOLg35+sYhEAW8Bd6pqTsVzwf4ZqWqZqo4EeuBa3oMCXKUWQ0QuBDJUdWmg61Kb0EBXoJnsAnpWeNzDd8xUli4iXVV1j4h0xd3pBS0RCcMFgRmq+rbvsH1Gx1DVAyIyBzgZ6Cgiob673mD+WxsHXCQi5wMRQAzwOC3w8wmWFsFiYIBvtL4dcAXwfoDr1FK9D1zn+/k64L0A1iWgfP25zwPrVPWRCqfsMwJEJFFEOvp+jgTOwo2jzAEm+4oF7eejqveqag9VTcZdc2ar6lW0wM8naFYW+6LyY0AIMF1VHwpwlQJORF4DJuDS4qYD9wPvAm8AvXDpvi9X1WMHlIOCiJwKfAms4kgf7324cYKg/4xEZDhusDMEd1P5hqr+UUT64iZkdAK+A65W1eLA1TTwRGQC8EtVvbAlfj5BEwiMMcZULVi6howxxlTDAoExxgQ5CwTGGBPkLBAYY0yQs0BgjDFBzgKBMc1IRCaUZ6E0pqWwQGCMMUHOAoExVRCRq3259peLyL98ydXyRORRX+79WSKS6Cs7UkQWichKEXmnfH8CEekvIl/48vUvE5F+vpePEpGZIrJeRGb4VjAbEzAWCIw5hogMBqYA43wJ1cqAq4AOwBJVHQLMw63EBngJuFtVh+NWIZcfnwE85cvXfwpQnrF0FHAnbm+MvricNMYETLAknTOmPiYBY4DFvpv1SFxiOS/wX1+ZV4C3RSQW6Kiq83zH/wO8KSLRQHdVfQdAVYsAfK/3raqm+R4vB5KBBf7/tYypmgUCYyoT4D+qeu9RB0V+d0y5huZnqZhXpgz7OzQBZl1DxlQ2C5gsIklweI/i3ri/l/KskT8EFqjqQWC/iJzmO34NMM+3o1maiFzie41wEWnfrL+FMXVkdyLGHENV14rIb4HPRMQDlAC3AfnAWN+5DNw4ArhUwtN8F/qtwA2+49cA/xKRP/pe47Jm/DWMqTPLPmpMHYlInqpGBboexjQ16xoyxpggZy0CY4wJctYiMMaYIGeBwBhjgpwFAmOMCXIWCIwxJshZIDDGmCD3/y6sKkDDPbhTAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BjAAKryxaS4k"
      },
      "source": [
        "Finally we print performances on *test set*:\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TBtIl3e2hkaQ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "996b7c6a-c941-4aeb-fc16-09fbeab787c3"
      },
      "source": [
        "test_loss, test_acc = model2.evaluate(X_test, Y_test)\n",
        "print('Test accuracy:', test_acc)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "96/96 [==============================] - 0s 3ms/step - loss: 1.6322 - accuracy: 0.3872\n",
            "Test accuracy: 0.3872131109237671\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hoiygacWhkaR"
      },
      "source": [
        "### Third Model: a first version of a Convolutional Neural Network"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MGwOywx8fadG"
      },
      "source": [
        "First, we read again our dataset from binary files"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3k6fJByUffXi"
      },
      "source": [
        "X_train = np.load(\"/content/drive/MyDrive/Machine Learning Competition/datasetTrain.npy\")\n",
        "X_test_label = np.load(\"/content/drive/MyDrive/Machine Learning Competition/datasetTest.npy\")\n",
        "Y_train = readCSVLabels(training_csv_dir)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jvzu5qeIMUZ6"
      },
      "source": [
        "Here we define a function to Process the read train dataset to get a correct split:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ttLSf3EzLVmd"
      },
      "source": [
        "def preProcessCNN(X_train, Y_train):\n",
        "  items_size = X_train.shape[0]\n",
        "  test_size = round(items_size * TEST_SPLIT)\n",
        "  train_size = items_size - test_size\n",
        "  X_test = X_train[train_size:train_size+test_size]\n",
        "  X_train = X_train[:train_size]\n",
        "  Y_test = Y_train[train_size:train_size+test_size]\n",
        "  Y_train = Y_train[:train_size]\n",
        "  return X_train, X_test, Y_train, Y_test"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pt1-kNzgMiTc"
      },
      "source": [
        "Then we actually read the content into the variables"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j6xH_-I9L23z"
      },
      "source": [
        "X_train, X_test, Y_train, Y_test = preProcessCNN(X_train, Y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PIC7tlMFhkaT"
      },
      "source": [
        "Labels are managed using One hot encoding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hcxxIMPyhkaU"
      },
      "source": [
        "Y_train = keras.utils.to_categorical(Y_train, NB_CLASSES)\n",
        "Y_test = keras.utils.to_categorical(Y_test, NB_CLASSES)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yNNyHUS1M5Pr"
      },
      "source": [
        "Here we define an *ImageDataGenerator* to do an augmentation of our dataset through a series if transforms. The operation is executed on *train*, *validation* and *test* datasets."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GE0IDlEqMx_F"
      },
      "source": [
        "train_datagen = ImageDataGenerator(rescale = 1./255, validation_split = VALIDATION_SPLIT, rotation_range=5, width_shift_range=0.2, height_shift_range=0.2, shear_range=0.2, horizontal_flip=True, vertical_flip=True, fill_mode='nearest')\n",
        "valid_datagen = ImageDataGenerator(rescale = 1./255, validation_split = VALIDATION_SPLIT)\n",
        "test_datagen  = ImageDataGenerator(rescale = 1./255)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Mi8zoc4FOUyS"
      },
      "source": [
        "Next, we define the flow for our three datasets to charge data in batches:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Cvj2fu0DN6AS"
      },
      "source": [
        "iterator_train = train_datagen.flow(X_train, Y_train, batch_size=BATCH_SIZE, subset='training')\n",
        "iterator_valid = valid_datagen.flow(X_train, Y_train, batch_size=BATCH_SIZE, subset='validation')\n",
        "iterator_test = test_datagen.flow(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iMl0KBRHRx4q"
      },
      "source": [
        "Here we define our model:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g2BWg_1IhkaV"
      },
      "source": [
        "model3 = tf.keras.models.Sequential()\n",
        "model3.add(keras.layers.Conv2D(64, kernel_size=(3, 3), activation='relu', input_shape=(img_h, img_w, img_c)))\n",
        "model3.add(keras.layers.BatchNormalization())\n",
        "model3.add(keras.layers.MaxPooling2D(pool_size=(2,2), strides=(2,2)))\n",
        "model3.add(keras.layers.Dropout(0.30))\n",
        "\n",
        "model3.add(keras.layers.Conv2D(128, (3, 3), activation='relu'))\n",
        "model3.add(keras.layers.BatchNormalization())\n",
        "model3.add(keras.layers.MaxPooling2D(pool_size=(2,2), strides=(2,2)))\n",
        "model3.add(keras.layers.Dropout(0.30))\n",
        "\n",
        "model3.add(keras.layers.Conv2D(256, (3, 3), activation='relu'))\n",
        "model3.add(keras.layers.BatchNormalization())\n",
        "model3.add(keras.layers.MaxPooling2D(pool_size=(2,2), strides=(2,2)))\n",
        "model3.add(keras.layers.Dropout(0.30))\n",
        "\n",
        "model3.add(keras.layers.Flatten())\n",
        "model3.add(keras.layers.Dense(4096, activation='relu'))\n",
        "model3.add(keras.layers.Dropout(0.50))\n",
        "model3.add(keras.layers.Dense(1024, activation='relu'))\n",
        "model3.add(keras.layers.Dropout(0.50))\n",
        "\n",
        "model3.add(keras.layers.Dense(NB_CLASSES, activation='softmax'))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "giNzXLonhkaf",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "52948d74-8228-44e9-9d0d-56f8670992d2"
      },
      "source": [
        "model3.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_5\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_8 (Conv2D)            (None, 46, 46, 64)        1792      \n",
            "_________________________________________________________________\n",
            "batch_normalization_8 (Batch (None, 46, 46, 64)        256       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_4 (MaxPooling2 (None, 23, 23, 64)        0         \n",
            "_________________________________________________________________\n",
            "dropout_5 (Dropout)          (None, 23, 23, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_9 (Conv2D)            (None, 21, 21, 128)       73856     \n",
            "_________________________________________________________________\n",
            "batch_normalization_9 (Batch (None, 21, 21, 128)       512       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_5 (MaxPooling2 (None, 10, 10, 128)       0         \n",
            "_________________________________________________________________\n",
            "dropout_6 (Dropout)          (None, 10, 10, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_10 (Conv2D)           (None, 8, 8, 256)         295168    \n",
            "_________________________________________________________________\n",
            "batch_normalization_10 (Batc (None, 8, 8, 256)         1024      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_6 (MaxPooling2 (None, 4, 4, 256)         0         \n",
            "_________________________________________________________________\n",
            "dropout_7 (Dropout)          (None, 4, 4, 256)         0         \n",
            "_________________________________________________________________\n",
            "flatten_5 (Flatten)          (None, 4096)              0         \n",
            "_________________________________________________________________\n",
            "dense_10 (Dense)             (None, 4096)              16781312  \n",
            "_________________________________________________________________\n",
            "dropout_8 (Dropout)          (None, 4096)              0         \n",
            "_________________________________________________________________\n",
            "dense_11 (Dense)             (None, 1024)              4195328   \n",
            "_________________________________________________________________\n",
            "dropout_9 (Dropout)          (None, 1024)              0         \n",
            "_________________________________________________________________\n",
            "dense_12 (Dense)             (None, 7)                 7175      \n",
            "=================================================================\n",
            "Total params: 21,356,423\n",
            "Trainable params: 21,355,527\n",
            "Non-trainable params: 896\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Dz2HqQj7R7DV"
      },
      "source": [
        "Here we define the optimizer and also some tools to manage overfitting:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nLWuXSRihka3"
      },
      "source": [
        "lrd = tf.keras.callbacks.ReduceLROnPlateau(monitor = 'val_loss', patience = 20, verbose = 1, factor = 0.50, min_lr = 1e-10)\n",
        "es = tf.keras.callbacks.EarlyStopping(verbose=1, patience=20)\n",
        "METRICS = [tf.keras.metrics.BinaryAccuracy(name='accuracy')]\n",
        "model3.compile(optimizer='Adam', loss='categorical_crossentropy', metrics=METRICS)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_yDWPMjuSCHS"
      },
      "source": [
        "Here we train the model defined above, splitting the initial training dataset into *training* and *validation*:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SwIJ_b-zhka5",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "eb11f84a-39ae-4847-9cca-d2db5b01f1e9"
      },
      "source": [
        "history = model3.fit(iterator_train, validation_data=iterator_valid, epochs=EPOCHS, verbose=VERBOSE, callbacks=[lrd,es])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "387/387 [==============================] - 32s 84ms/step - loss: 1.5104 - accuracy: 0.8751 - val_loss: 2.2449 - val_accuracy: 0.8507\n",
            "Epoch 2/50\n",
            "387/387 [==============================] - 33s 85ms/step - loss: 1.5085 - accuracy: 0.8750 - val_loss: 1.6134 - val_accuracy: 0.8678\n",
            "Epoch 3/50\n",
            "387/387 [==============================] - 32s 83ms/step - loss: 1.5046 - accuracy: 0.8754 - val_loss: 1.4130 - val_accuracy: 0.8797\n",
            "Epoch 4/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.4913 - accuracy: 0.8759 - val_loss: 1.3372 - val_accuracy: 0.8848\n",
            "Epoch 5/50\n",
            "387/387 [==============================] - 32s 83ms/step - loss: 1.4924 - accuracy: 0.8760 - val_loss: 1.4279 - val_accuracy: 0.8758\n",
            "Epoch 6/50\n",
            "387/387 [==============================] - 32s 82ms/step - loss: 1.4880 - accuracy: 0.8760 - val_loss: 1.3757 - val_accuracy: 0.8820\n",
            "Epoch 7/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.4795 - accuracy: 0.8762 - val_loss: 1.3388 - val_accuracy: 0.8838\n",
            "Epoch 8/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.4839 - accuracy: 0.8757 - val_loss: 1.3813 - val_accuracy: 0.8832\n",
            "Epoch 9/50\n",
            "387/387 [==============================] - 31s 79ms/step - loss: 1.4808 - accuracy: 0.8763 - val_loss: 1.3123 - val_accuracy: 0.8850\n",
            "Epoch 10/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.4750 - accuracy: 0.8768 - val_loss: 1.3266 - val_accuracy: 0.8835\n",
            "Epoch 11/50\n",
            "387/387 [==============================] - 31s 80ms/step - loss: 1.4694 - accuracy: 0.8767 - val_loss: 1.3079 - val_accuracy: 0.8854\n",
            "Epoch 12/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.4550 - accuracy: 0.8776 - val_loss: 1.3048 - val_accuracy: 0.8847\n",
            "Epoch 13/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.4589 - accuracy: 0.8773 - val_loss: 1.3341 - val_accuracy: 0.8831\n",
            "Epoch 14/50\n",
            "387/387 [==============================] - 30s 79ms/step - loss: 1.4588 - accuracy: 0.8774 - val_loss: 1.3378 - val_accuracy: 0.8836\n",
            "Epoch 15/50\n",
            "387/387 [==============================] - 32s 82ms/step - loss: 1.4546 - accuracy: 0.8771 - val_loss: 1.3130 - val_accuracy: 0.8829\n",
            "Epoch 16/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.4581 - accuracy: 0.8771 - val_loss: 1.3262 - val_accuracy: 0.8836\n",
            "Epoch 17/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.4400 - accuracy: 0.8785 - val_loss: 1.2764 - val_accuracy: 0.8853\n",
            "Epoch 18/50\n",
            "387/387 [==============================] - 31s 80ms/step - loss: 1.4388 - accuracy: 0.8780 - val_loss: 1.3627 - val_accuracy: 0.8810\n",
            "Epoch 19/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.4389 - accuracy: 0.8779 - val_loss: 1.2779 - val_accuracy: 0.8858\n",
            "Epoch 20/50\n",
            "387/387 [==============================] - 30s 77ms/step - loss: 1.4343 - accuracy: 0.8784 - val_loss: 1.2635 - val_accuracy: 0.8869\n",
            "Epoch 21/50\n",
            "387/387 [==============================] - 31s 80ms/step - loss: 1.4263 - accuracy: 0.8792 - val_loss: 1.3050 - val_accuracy: 0.8832\n",
            "Epoch 22/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.4190 - accuracy: 0.8788 - val_loss: 1.2727 - val_accuracy: 0.8865\n",
            "Epoch 23/50\n",
            "387/387 [==============================] - 31s 80ms/step - loss: 1.4230 - accuracy: 0.8788 - val_loss: 1.2692 - val_accuracy: 0.8863\n",
            "Epoch 24/50\n",
            "387/387 [==============================] - 30s 77ms/step - loss: 1.4170 - accuracy: 0.8788 - val_loss: 1.2474 - val_accuracy: 0.8870\n",
            "Epoch 25/50\n",
            "387/387 [==============================] - 31s 80ms/step - loss: 1.4145 - accuracy: 0.8791 - val_loss: 1.2405 - val_accuracy: 0.8878\n",
            "Epoch 26/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.4037 - accuracy: 0.8793 - val_loss: 1.2916 - val_accuracy: 0.8830\n",
            "Epoch 27/50\n",
            "387/387 [==============================] - 31s 80ms/step - loss: 1.4136 - accuracy: 0.8789 - val_loss: 1.2441 - val_accuracy: 0.8878\n",
            "Epoch 28/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.3989 - accuracy: 0.8801 - val_loss: 1.2708 - val_accuracy: 0.8845\n",
            "Epoch 29/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.3919 - accuracy: 0.8800 - val_loss: 1.2208 - val_accuracy: 0.8884\n",
            "Epoch 30/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.3877 - accuracy: 0.8805 - val_loss: 1.2343 - val_accuracy: 0.8877\n",
            "Epoch 31/50\n",
            "387/387 [==============================] - 30s 79ms/step - loss: 1.3917 - accuracy: 0.8799 - val_loss: 1.2266 - val_accuracy: 0.8900\n",
            "Epoch 32/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.3911 - accuracy: 0.8797 - val_loss: 1.2143 - val_accuracy: 0.8889\n",
            "Epoch 33/50\n",
            "387/387 [==============================] - 31s 79ms/step - loss: 1.3769 - accuracy: 0.8807 - val_loss: 1.2249 - val_accuracy: 0.8887\n",
            "Epoch 34/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.3841 - accuracy: 0.8806 - val_loss: 1.2066 - val_accuracy: 0.8899\n",
            "Epoch 35/50\n",
            "387/387 [==============================] - 30s 79ms/step - loss: 1.3785 - accuracy: 0.8805 - val_loss: 1.3032 - val_accuracy: 0.8823\n",
            "Epoch 36/50\n",
            "387/387 [==============================] - 30s 77ms/step - loss: 1.3784 - accuracy: 0.8799 - val_loss: 1.2377 - val_accuracy: 0.8870\n",
            "Epoch 37/50\n",
            "387/387 [==============================] - 31s 80ms/step - loss: 1.3682 - accuracy: 0.8812 - val_loss: 1.5128 - val_accuracy: 0.8775\n",
            "Epoch 38/50\n",
            "387/387 [==============================] - 31s 79ms/step - loss: 1.3679 - accuracy: 0.8809 - val_loss: 1.2578 - val_accuracy: 0.8855\n",
            "Epoch 39/50\n",
            "387/387 [==============================] - 31s 80ms/step - loss: 1.3683 - accuracy: 0.8815 - val_loss: 1.2263 - val_accuracy: 0.8897\n",
            "Epoch 40/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.3605 - accuracy: 0.8818 - val_loss: 1.3890 - val_accuracy: 0.8844\n",
            "Epoch 41/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.3615 - accuracy: 0.8813 - val_loss: 1.1994 - val_accuracy: 0.8908\n",
            "Epoch 42/50\n",
            "387/387 [==============================] - 31s 79ms/step - loss: 1.3554 - accuracy: 0.8814 - val_loss: 1.3320 - val_accuracy: 0.8842\n",
            "Epoch 43/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.3596 - accuracy: 0.8813 - val_loss: 1.1925 - val_accuracy: 0.8908\n",
            "Epoch 44/50\n",
            "387/387 [==============================] - 31s 79ms/step - loss: 1.3552 - accuracy: 0.8819 - val_loss: 1.1981 - val_accuracy: 0.8898\n",
            "Epoch 45/50\n",
            "387/387 [==============================] - 30s 77ms/step - loss: 1.3528 - accuracy: 0.8815 - val_loss: 1.2055 - val_accuracy: 0.8906\n",
            "Epoch 46/50\n",
            "387/387 [==============================] - 31s 79ms/step - loss: 1.3508 - accuracy: 0.8818 - val_loss: 1.2423 - val_accuracy: 0.8861\n",
            "Epoch 47/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.3482 - accuracy: 0.8817 - val_loss: 1.2608 - val_accuracy: 0.8852\n",
            "Epoch 48/50\n",
            "387/387 [==============================] - 31s 81ms/step - loss: 1.3488 - accuracy: 0.8813 - val_loss: 1.2422 - val_accuracy: 0.8852\n",
            "Epoch 49/50\n",
            "387/387 [==============================] - 30s 78ms/step - loss: 1.3442 - accuracy: 0.8822 - val_loss: 1.2155 - val_accuracy: 0.8897\n",
            "Epoch 50/50\n",
            "387/387 [==============================] - 30s 77ms/step - loss: 1.3334 - accuracy: 0.8821 - val_loss: 1.2683 - val_accuracy: 0.8823\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ToBRt1gXPn5Z"
      },
      "source": [
        "Here we plot the information about *loss* and *accuracy* of our trained model:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "DVv4fcNMPuPG",
        "outputId": "4c055324-0696-462b-ea6e-36ce9d24087f"
      },
      "source": [
        "plotAccuracy(history)\n",
        "plotLoss(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7UCXFkn2QE_w"
      },
      "source": [
        "Finally we print performances on *test set*:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MLvE6pF0hka7",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d4c38644-eb53-4a04-8027-1998b5fd8bfd"
      },
      "source": [
        "test_loss, test_acc = model3.evaluate(iterator_test)\n",
        "print('Test accuracy:', test_acc)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "96/96 [==============================] - 1s 6ms/step - loss: 1.3814 - accuracy: 0.8835\n",
            "Test accuracy: 0.8835129141807556\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7PbvTOTwhka-"
      },
      "source": [
        "### Fourth Model: A Second Version of a Convolutional Neural Network"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "o9q4HmG6WaSB"
      },
      "source": [
        "First, we read again our dataset from binary files"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1iFGjr1oWegy"
      },
      "source": [
        "X_train = np.load(\"/content/drive/MyDrive/Machine Learning Competition/datasetTrain.npy\")\n",
        "X_test_label = np.load(\"/content/drive/MyDrive/Machine Learning Competition/datasetTest.npy\")\n",
        "Y_train = readCSVLabels(training_csv_dir)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kunleoWYWmy6"
      },
      "source": [
        "Then we actually read the content into the variables"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pn9XFqo8Wnhb"
      },
      "source": [
        "X_train, X_test, Y_train, Y_test = preProcessCNN(X_train, Y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qo-BS5sGWwny"
      },
      "source": [
        "Labels are managed using One hot encoding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ExWDmaoKW0xA"
      },
      "source": [
        "Y_train = keras.utils.to_categorical(Y_train, NB_CLASSES)\n",
        "Y_test = keras.utils.to_categorical(Y_test, NB_CLASSES)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0ewo5PeLW8Ap"
      },
      "source": [
        "Here we define an *ImageDataGenerator* to do an augmentation of our dataset through a series if transforms. The operation is executed on *train*, *validation* and *test* datasets."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MCgoRbtFW8-S"
      },
      "source": [
        "train_datagen = ImageDataGenerator(rescale = 1./255, validation_split = VALIDATION_SPLIT, rotation_range=5, width_shift_range=0.2, height_shift_range=0.2, shear_range=0.2, zoom_range=0.2, horizontal_flip=True, vertical_flip=True, fill_mode='nearest')\n",
        "valid_datagen = ImageDataGenerator(rescale = 1./255, validation_split = VALIDATION_SPLIT)\n",
        "test_datagen  = ImageDataGenerator(rescale = 1./255)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kd-25yy1XDF5"
      },
      "source": [
        "Next, we define the flow for our three datasets to charge data in batches:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Trqo3ePMXJUx"
      },
      "source": [
        "iterator_train = train_datagen.flow(X_train, Y_train, batch_size=BATCH_SIZE, subset='training')\n",
        "iterator_valid = valid_datagen.flow(X_train, Y_train, batch_size=BATCH_SIZE, subset='validation')\n",
        "iterator_test = test_datagen.flow(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "I6bTiW8ThkbA"
      },
      "source": [
        "We can create our Model as follows:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xRmH1iIZhkbE"
      },
      "source": [
        "model4 = tf.keras.models.Sequential()\n",
        "\n",
        "model4.add(keras.layers.Conv2D(32, kernel_size=(3, 3), padding='same', activation='relu', input_shape=(img_h, img_w, img_c)))  \n",
        "model4.add(keras.layers.BatchNormalization())\n",
        "model4.add(keras.layers.Conv2D(32, kernel_size=(3, 3), padding='same', activation='relu'))  \n",
        "model4.add(keras.layers.BatchNormalization())\n",
        "model4.add(keras.layers.MaxPooling2D(pool_size=(2,2)))\n",
        "\n",
        "model4.add(keras.layers.Conv2D(64, kernel_size=(3, 3), padding='same', activation='relu'))  \n",
        "model4.add(keras.layers.BatchNormalization())\n",
        "model4.add(keras.layers.Conv2D(64, kernel_size=(3, 3), padding='same', activation='relu'))  \n",
        "model4.add(keras.layers.BatchNormalization())\n",
        "model4.add(keras.layers.MaxPooling2D(pool_size=(2,2)))\n",
        "\n",
        "model4.add(keras.layers.Conv2D(128, kernel_size=(3, 3), padding='same', activation='relu'))  \n",
        "model4.add(keras.layers.BatchNormalization())\n",
        "model4.add(keras.layers.Conv2D(128, kernel_size=(3, 3), padding='same', activation='relu'))  \n",
        "model4.add(keras.layers.BatchNormalization())\n",
        "model4.add(keras.layers.MaxPooling2D(pool_size=(2,2)))\n",
        "\n",
        "model4.add(keras.layers.Conv2D(256, kernel_size=(3, 3), padding='same', activation='relu'))  \n",
        "model4.add(keras.layers.BatchNormalization())\n",
        "model4.add(keras.layers.Conv2D(256, kernel_size=(3, 3), padding='same', activation='relu'))  \n",
        "model4.add(keras.layers.BatchNormalization())\n",
        "model4.add(keras.layers.MaxPooling2D(pool_size=(2,2)))\n",
        "\n",
        "model4.add(keras.layers.Flatten())\n",
        "model4.add(keras.layers.Dense(2304, activation='relu'))\n",
        "model4.add(keras.layers.Dropout(0.30))\n",
        "model4.add(keras.layers.Dense(NB_CLASSES, activation='softmax'))\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OqmY_wf6hkbG",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5b1f9a45-3d25-469f-a22e-7fea9b1c133e"
      },
      "source": [
        "model4.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_16 (Conv2D)           (None, 48, 48, 32)        896       \n",
            "_________________________________________________________________\n",
            "batch_normalization_16 (Batc (None, 48, 48, 32)        128       \n",
            "_________________________________________________________________\n",
            "conv2d_17 (Conv2D)           (None, 48, 48, 32)        9248      \n",
            "_________________________________________________________________\n",
            "batch_normalization_17 (Batc (None, 48, 48, 32)        128       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_8 (MaxPooling2 (None, 24, 24, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_18 (Conv2D)           (None, 24, 24, 64)        18496     \n",
            "_________________________________________________________________\n",
            "batch_normalization_18 (Batc (None, 24, 24, 64)        256       \n",
            "_________________________________________________________________\n",
            "conv2d_19 (Conv2D)           (None, 24, 24, 64)        36928     \n",
            "_________________________________________________________________\n",
            "batch_normalization_19 (Batc (None, 24, 24, 64)        256       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_9 (MaxPooling2 (None, 12, 12, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_20 (Conv2D)           (None, 12, 12, 128)       73856     \n",
            "_________________________________________________________________\n",
            "batch_normalization_20 (Batc (None, 12, 12, 128)       512       \n",
            "_________________________________________________________________\n",
            "conv2d_21 (Conv2D)           (None, 12, 12, 128)       147584    \n",
            "_________________________________________________________________\n",
            "batch_normalization_21 (Batc (None, 12, 12, 128)       512       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_10 (MaxPooling (None, 6, 6, 128)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_22 (Conv2D)           (None, 6, 6, 256)         295168    \n",
            "_________________________________________________________________\n",
            "batch_normalization_22 (Batc (None, 6, 6, 256)         1024      \n",
            "_________________________________________________________________\n",
            "conv2d_23 (Conv2D)           (None, 6, 6, 256)         590080    \n",
            "_________________________________________________________________\n",
            "batch_normalization_23 (Batc (None, 6, 6, 256)         1024      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_11 (MaxPooling (None, 3, 3, 256)         0         \n",
            "_________________________________________________________________\n",
            "flatten_2 (Flatten)          (None, 2304)              0         \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 2304)              5310720   \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 2304)              0         \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 7)                 16135     \n",
            "=================================================================\n",
            "Total params: 6,502,951\n",
            "Trainable params: 6,501,031\n",
            "Non-trainable params: 1,920\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "22BVi26IXg5h"
      },
      "source": [
        "Here we define the optimizer and also some tools to manage overfitting:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mt7ysxwOhkbI"
      },
      "source": [
        "lrd = tf.keras.callbacks.ReduceLROnPlateau(monitor = 'val_loss', patience = 20, verbose = 1, factor = 0.50, min_lr = 1e-10)\n",
        "es = tf.keras.callbacks.EarlyStopping(verbose=1, patience=20)\n",
        "METRICS = [tf.keras.metrics.BinaryAccuracy(name='accuracy')]\n",
        "model4.compile(optimizer='Adam', loss='categorical_crossentropy', metrics=METRICS)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rJEhCB3WXnUY"
      },
      "source": [
        "Here we train the model defined above, splitting the initial training dataset into *training* and *validation*:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h3LqxuHShkbJ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9516c4e4-1b61-40cb-cae2-30e09ef2bb01"
      },
      "source": [
        "history = model4.fit(iterator_train, validation_data=iterator_valid, epochs=EPOCHS, verbose=VERBOSE, callbacks=[lrd,es])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "387/387 [==============================] - 23s 56ms/step - loss: 2.2201 - accuracy: 0.8466 - val_loss: 1.9071 - val_accuracy: 0.8571\n",
            "Epoch 2/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.7529 - accuracy: 0.8582 - val_loss: 1.7874 - val_accuracy: 0.8530\n",
            "Epoch 3/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.6443 - accuracy: 0.8649 - val_loss: 1.5422 - val_accuracy: 0.8686\n",
            "Epoch 4/50\n",
            "387/387 [==============================] - 21s 54ms/step - loss: 1.5573 - accuracy: 0.8698 - val_loss: 1.4279 - val_accuracy: 0.8772\n",
            "Epoch 5/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 1.4874 - accuracy: 0.8737 - val_loss: 1.5444 - val_accuracy: 0.8718\n",
            "Epoch 6/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.4176 - accuracy: 0.8776 - val_loss: 1.3669 - val_accuracy: 0.8756\n",
            "Epoch 7/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.3787 - accuracy: 0.8787 - val_loss: 1.3574 - val_accuracy: 0.8799\n",
            "Epoch 8/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.3408 - accuracy: 0.8812 - val_loss: 1.2843 - val_accuracy: 0.8848\n",
            "Epoch 9/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 1.3166 - accuracy: 0.8827 - val_loss: 1.3475 - val_accuracy: 0.8767\n",
            "Epoch 10/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.3159 - accuracy: 0.8824 - val_loss: 1.2735 - val_accuracy: 0.8845\n",
            "Epoch 11/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.2755 - accuracy: 0.8852 - val_loss: 1.3474 - val_accuracy: 0.8788\n",
            "Epoch 12/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.2602 - accuracy: 0.8854 - val_loss: 1.2599 - val_accuracy: 0.8883\n",
            "Epoch 13/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.2448 - accuracy: 0.8870 - val_loss: 1.2040 - val_accuracy: 0.8885\n",
            "Epoch 14/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.2303 - accuracy: 0.8883 - val_loss: 1.2300 - val_accuracy: 0.8864\n",
            "Epoch 15/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.2066 - accuracy: 0.8892 - val_loss: 1.2033 - val_accuracy: 0.8866\n",
            "Epoch 16/50\n",
            "387/387 [==============================] - 22s 57ms/step - loss: 1.1915 - accuracy: 0.8903 - val_loss: 1.1732 - val_accuracy: 0.8917\n",
            "Epoch 17/50\n",
            "387/387 [==============================] - 22s 57ms/step - loss: 1.1805 - accuracy: 0.8909 - val_loss: 1.1499 - val_accuracy: 0.8915\n",
            "Epoch 18/50\n",
            "387/387 [==============================] - 24s 61ms/step - loss: 1.1725 - accuracy: 0.8919 - val_loss: 1.1898 - val_accuracy: 0.8901\n",
            "Epoch 19/50\n",
            "387/387 [==============================] - 23s 59ms/step - loss: 1.1785 - accuracy: 0.8915 - val_loss: 1.1231 - val_accuracy: 0.8963\n",
            "Epoch 20/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 1.1529 - accuracy: 0.8929 - val_loss: 1.1374 - val_accuracy: 0.8938\n",
            "Epoch 21/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 1.1362 - accuracy: 0.8938 - val_loss: 1.1562 - val_accuracy: 0.8920\n",
            "Epoch 22/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.1209 - accuracy: 0.8954 - val_loss: 1.1061 - val_accuracy: 0.8970\n",
            "Epoch 23/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 1.1245 - accuracy: 0.8951 - val_loss: 1.2150 - val_accuracy: 0.8903\n",
            "Epoch 24/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.1054 - accuracy: 0.8964 - val_loss: 1.1207 - val_accuracy: 0.8971\n",
            "Epoch 25/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 1.0963 - accuracy: 0.8968 - val_loss: 1.0977 - val_accuracy: 0.8951\n",
            "Epoch 26/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.0879 - accuracy: 0.8982 - val_loss: 1.0731 - val_accuracy: 0.8995\n",
            "Epoch 27/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 1.0809 - accuracy: 0.8985 - val_loss: 1.1259 - val_accuracy: 0.8965\n",
            "Epoch 28/50\n",
            "387/387 [==============================] - 21s 54ms/step - loss: 1.0769 - accuracy: 0.8986 - val_loss: 1.1520 - val_accuracy: 0.8896\n",
            "Epoch 29/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.0619 - accuracy: 0.8997 - val_loss: 1.1415 - val_accuracy: 0.8943\n",
            "Epoch 30/50\n",
            "387/387 [==============================] - 23s 59ms/step - loss: 1.0499 - accuracy: 0.9007 - val_loss: 1.0811 - val_accuracy: 0.8985\n",
            "Epoch 31/50\n",
            "387/387 [==============================] - 24s 61ms/step - loss: 1.0417 - accuracy: 0.9012 - val_loss: 1.0614 - val_accuracy: 0.9008\n",
            "Epoch 32/50\n",
            "387/387 [==============================] - 22s 58ms/step - loss: 1.0385 - accuracy: 0.9012 - val_loss: 1.0592 - val_accuracy: 0.9011\n",
            "Epoch 33/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 1.0349 - accuracy: 0.9008 - val_loss: 1.0759 - val_accuracy: 0.8997\n",
            "Epoch 34/50\n",
            "387/387 [==============================] - 21s 56ms/step - loss: 1.0234 - accuracy: 0.9027 - val_loss: 1.0655 - val_accuracy: 0.8995\n",
            "Epoch 35/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 1.0158 - accuracy: 0.9031 - val_loss: 1.0373 - val_accuracy: 0.9003\n",
            "Epoch 36/50\n",
            "387/387 [==============================] - 21s 55ms/step - loss: 1.0004 - accuracy: 0.9044 - val_loss: 1.1616 - val_accuracy: 0.8891\n",
            "Epoch 37/50\n",
            "387/387 [==============================] - 22s 57ms/step - loss: 1.0086 - accuracy: 0.9039 - val_loss: 1.0531 - val_accuracy: 0.9007\n",
            "Epoch 38/50\n",
            "387/387 [==============================] - 23s 58ms/step - loss: 0.9980 - accuracy: 0.9043 - val_loss: 1.0833 - val_accuracy: 0.8978\n",
            "Epoch 39/50\n",
            "387/387 [==============================] - 23s 58ms/step - loss: 0.9828 - accuracy: 0.9052 - val_loss: 1.0835 - val_accuracy: 0.8991\n",
            "Epoch 40/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 0.9889 - accuracy: 0.9055 - val_loss: 1.0621 - val_accuracy: 0.8974\n",
            "Epoch 41/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 0.9726 - accuracy: 0.9060 - val_loss: 1.2620 - val_accuracy: 0.8836\n",
            "Epoch 42/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 0.9809 - accuracy: 0.9055 - val_loss: 1.0450 - val_accuracy: 0.9035\n",
            "Epoch 43/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 0.9660 - accuracy: 0.9068 - val_loss: 1.0731 - val_accuracy: 0.9002\n",
            "Epoch 44/50\n",
            "387/387 [==============================] - 22s 57ms/step - loss: 0.9653 - accuracy: 0.9072 - val_loss: 1.0781 - val_accuracy: 0.8997\n",
            "Epoch 45/50\n",
            "387/387 [==============================] - 22s 56ms/step - loss: 0.9525 - accuracy: 0.9079 - val_loss: 1.0828 - val_accuracy: 0.8995\n",
            "Epoch 46/50\n",
            "387/387 [==============================] - 22s 57ms/step - loss: 0.9480 - accuracy: 0.9080 - val_loss: 1.0714 - val_accuracy: 0.8993\n",
            "Epoch 47/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 0.9440 - accuracy: 0.9083 - val_loss: 1.0276 - val_accuracy: 0.9029\n",
            "Epoch 48/50\n",
            "387/387 [==============================] - 22s 58ms/step - loss: 0.9409 - accuracy: 0.9091 - val_loss: 1.0455 - val_accuracy: 0.9037\n",
            "Epoch 49/50\n",
            "387/387 [==============================] - 22s 57ms/step - loss: 0.9363 - accuracy: 0.9092 - val_loss: 1.0410 - val_accuracy: 0.9026\n",
            "Epoch 50/50\n",
            "387/387 [==============================] - 23s 59ms/step - loss: 0.9262 - accuracy: 0.9103 - val_loss: 1.0160 - val_accuracy: 0.9038\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZwzuxQRzX4nu"
      },
      "source": [
        "Here we plot the information about *loss* and *accuracy* of our trained model:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "cYDeSsvWX8iv",
        "outputId": "49ab8139-0acb-4ca8-ad07-7fa06b679ee3"
      },
      "source": [
        "plotAccuracy(history)\n",
        "plotLoss(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7JiOHpHaYA9g"
      },
      "source": [
        "Finally we print performances on *test set*:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Vg9XjtjuJC2u",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "3e0f3f24-400d-4d64-b4ac-7ffbef994aca"
      },
      "source": [
        "test_loss, test_acc = model4.evaluate(iterator_test)\n",
        "print('Test accuracy:', test_acc)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "96/96 [==============================] - 1s 6ms/step - loss: 1.0266 - accuracy: 0.9040\n",
            "Test accuracy: 0.9039813280105591\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6MyyayOvJXlW"
      },
      "source": [
        "### Fifth Model: Transfer Learning of a prebuilt learning model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "spIFSJMzftxZ"
      },
      "source": [
        "First, we read again our dataset from binary files"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hMp0sY8IZ65g"
      },
      "source": [
        "X_train = np.load(\"/content/drive/MyDrive/Machine Learning Competition/datasetTrain.npy\")\n",
        "X_test_label = np.load(\"/content/drive/MyDrive/Machine Learning Competition/datasetTest.npy\")\n",
        "Y_train = readCSVLabels(training_csv_dir)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WJ9oALkTf11x"
      },
      "source": [
        "Then we actually read the content into the variables"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xrnBX_gkf7cJ"
      },
      "source": [
        "X_train, X_test, Y_train, Y_test = preProcessCNN(X_train, Y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iazhOmIRf_6Q"
      },
      "source": [
        "Labels are managed using One hot encoding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sebgOYQrgFYA"
      },
      "source": [
        "Y_train = keras.utils.to_categorical(Y_train, NB_CLASSES)\n",
        "Y_test = keras.utils.to_categorical(Y_test, NB_CLASSES)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nPnt5hARgJHg"
      },
      "source": [
        "Here we define an *ImageDataGenerator* to do an augmentation of our dataset through a series if transforms. The operation is executed on *train*, *validation* and *test* datasets."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uUeABbHOgOOo"
      },
      "source": [
        "train_datagen = ImageDataGenerator(rescale = 1./255, validation_split = VALIDATION_SPLIT, rotation_range=5, width_shift_range=0.2, height_shift_range=0.2, shear_range=0.2, horizontal_flip=True, vertical_flip=True, fill_mode='nearest')\n",
        "valid_datagen = ImageDataGenerator(rescale = 1./255, validation_split = VALIDATION_SPLIT)\n",
        "test_datagen  = ImageDataGenerator(rescale = 1./255)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kKx4xNC_gSFY"
      },
      "source": [
        "Next, we define the flow for our three datasets to charge data in batches:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pw7CXcfOgXiu"
      },
      "source": [
        "iterator_train = train_datagen.flow(X_train, Y_train, batch_size=BATCH_SIZE, subset='training')\n",
        "iterator_valid = valid_datagen.flow(X_train, Y_train, batch_size=BATCH_SIZE, subset='validation')\n",
        "iterator_test = test_datagen.flow(X_test, Y_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e-m2M0SCglr5"
      },
      "source": [
        "We first import our pretrained Model *EfficientNetB7* trained on *imagenet*"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "740d6MZhxe2r",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "430c0ed3-9060-4755-9911-8d3340bbcb49"
      },
      "source": [
        "base_model = tf.keras.applications.EfficientNetB0(input_shape=(48,48,3), include_top=False, weights=\"imagenet\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/keras-applications/efficientnetb0_notop.h5\n",
            "16711680/16705208 [==============================] - 0s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "K5vq6VWng3_3"
      },
      "source": [
        "First, all the layers of our pretrained Model are set as not trainable"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G5Hfsp93mpvP"
      },
      "source": [
        "for layer in base_model.layers: \n",
        "    layer.trainable=False"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rNPF1SKjhXLP"
      },
      "source": [
        "Here we define our Model, adding the Fully Connected Network"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nhblor-ObTY4"
      },
      "source": [
        "model5 = tf.keras.models.Sequential()\n",
        "\n",
        "model5.add(base_model)\n",
        "model5.add(keras.layers.Flatten())\n",
        "model5.add(keras.layers.Dense(5120, activation='relu'))\n",
        "model5.add(keras.layers.Dropout(0.50))\n",
        "model5.add(keras.layers.Dense(NB_CLASSES, activation='softmax'))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "meYwfX_0m28F",
        "outputId": "038d0308-6a1d-495e-95e8-fef49de85aa8"
      },
      "source": [
        "model5.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_3\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "efficientnetb0 (Functional)  (None, 2, 2, 1280)        4049571   \n",
            "_________________________________________________________________\n",
            "flatten_3 (Flatten)          (None, 5120)              0         \n",
            "_________________________________________________________________\n",
            "dense_6 (Dense)              (None, 5120)              26219520  \n",
            "_________________________________________________________________\n",
            "dropout_3 (Dropout)          (None, 5120)              0         \n",
            "_________________________________________________________________\n",
            "dense_7 (Dense)              (None, 7)                 35847     \n",
            "=================================================================\n",
            "Total params: 30,304,938\n",
            "Trainable params: 26,255,367\n",
            "Non-trainable params: 4,049,571\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oxOgd14xhman"
      },
      "source": [
        "Here we define the optimizer and also some tools to manage overfitting:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bqYo1pxQm4Sw"
      },
      "source": [
        "lrd = tf.keras.callbacks.ReduceLROnPlateau(monitor = 'val_loss', patience = 20, verbose = 1, factor = 0.50, min_lr = 1e-10)\n",
        "es = tf.keras.callbacks.EarlyStopping(verbose=1, patience=20)\n",
        "METRICS = [tf.keras.metrics.BinaryAccuracy(name='accuracy')]\n",
        "model5.compile(optimizer='Adam', loss='categorical_crossentropy', metrics=METRICS)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bIGI-UIlh-8G"
      },
      "source": [
        "Here we train the model defined above, splitting the initial training dataset into *training* and *validation*:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FIVMiCPSh4bb",
        "outputId": "4dbd4580-a756-432f-c1e7-97816d4afb38"
      },
      "source": [
        "history = model5.fit(iterator_train, validation_data=iterator_valid, epochs=EPOCHS, verbose=VERBOSE, callbacks=[lrd,es])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "387/387 [==============================] - 30s 64ms/step - loss: 2.0787 - accuracy: 0.8520 - val_loss: 1.8096 - val_accuracy: 0.8571\n",
            "Epoch 2/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8275 - accuracy: 0.8571 - val_loss: 1.8004 - val_accuracy: 0.8571\n",
            "Epoch 3/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8268 - accuracy: 0.8571 - val_loss: 1.7973 - val_accuracy: 0.8571\n",
            "Epoch 4/50\n",
            "387/387 [==============================] - 23s 59ms/step - loss: 1.8250 - accuracy: 0.8571 - val_loss: 1.7992 - val_accuracy: 0.8571\n",
            "Epoch 5/50\n",
            "387/387 [==============================] - 23s 59ms/step - loss: 1.8216 - accuracy: 0.8571 - val_loss: 1.8034 - val_accuracy: 0.8571\n",
            "Epoch 6/50\n",
            "387/387 [==============================] - 24s 61ms/step - loss: 1.8199 - accuracy: 0.8571 - val_loss: 1.7967 - val_accuracy: 0.8571\n",
            "Epoch 7/50\n",
            "387/387 [==============================] - 24s 62ms/step - loss: 1.8182 - accuracy: 0.8571 - val_loss: 1.8011 - val_accuracy: 0.8571\n",
            "Epoch 8/50\n",
            "387/387 [==============================] - 24s 61ms/step - loss: 1.8173 - accuracy: 0.8571 - val_loss: 1.8030 - val_accuracy: 0.8571\n",
            "Epoch 9/50\n",
            "387/387 [==============================] - 24s 62ms/step - loss: 1.8141 - accuracy: 0.8571 - val_loss: 1.7999 - val_accuracy: 0.8571\n",
            "Epoch 10/50\n",
            "387/387 [==============================] - 24s 61ms/step - loss: 1.8127 - accuracy: 0.8571 - val_loss: 1.7985 - val_accuracy: 0.8571\n",
            "Epoch 11/50\n",
            "387/387 [==============================] - 23s 59ms/step - loss: 1.8124 - accuracy: 0.8571 - val_loss: 1.7978 - val_accuracy: 0.8571\n",
            "Epoch 12/50\n",
            "387/387 [==============================] - 24s 61ms/step - loss: 1.8118 - accuracy: 0.8571 - val_loss: 1.7976 - val_accuracy: 0.8571\n",
            "Epoch 13/50\n",
            "387/387 [==============================] - 24s 61ms/step - loss: 1.8118 - accuracy: 0.8571 - val_loss: 1.7972 - val_accuracy: 0.8571\n",
            "Epoch 14/50\n",
            "387/387 [==============================] - 24s 62ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7973 - val_accuracy: 0.8571\n",
            "Epoch 15/50\n",
            "387/387 [==============================] - 24s 61ms/step - loss: 1.8117 - accuracy: 0.8571 - val_loss: 1.7971 - val_accuracy: 0.8571\n",
            "Epoch 16/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.8007 - val_accuracy: 0.8571\n",
            "Epoch 17/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8120 - accuracy: 0.8571 - val_loss: 1.7968 - val_accuracy: 0.8571\n",
            "Epoch 18/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7969 - val_accuracy: 0.8571\n",
            "Epoch 19/50\n",
            "387/387 [==============================] - 23s 61ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7969 - val_accuracy: 0.8571\n",
            "Epoch 20/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7970 - val_accuracy: 0.8571\n",
            "Epoch 21/50\n",
            "387/387 [==============================] - 23s 59ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7970 - val_accuracy: 0.8571\n",
            "Epoch 22/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7970 - val_accuracy: 0.8571\n",
            "Epoch 23/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7971 - val_accuracy: 0.8571\n",
            "Epoch 24/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7969 - val_accuracy: 0.8571\n",
            "Epoch 25/50\n",
            "387/387 [==============================] - 23s 59ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7968 - val_accuracy: 0.8571\n",
            "Epoch 26/50\n",
            "387/387 [==============================] - 23s 60ms/step - loss: 1.8116 - accuracy: 0.8571 - val_loss: 1.7969 - val_accuracy: 0.8571\n",
            "\n",
            "Epoch 00026: ReduceLROnPlateau reducing learning rate to 0.0005000000237487257.\n",
            "Epoch 00026: early stopping\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EeCZjOa5iA-F"
      },
      "source": [
        "Here we plot the information about *loss* and *accuracy* of our trained model:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BPqs_te2iHOe",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "outputId": "32d7e6c6-4fff-4763-e848-39e226b768b5"
      },
      "source": [
        "plotAccuracy(history)\n",
        "plotLoss(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jjD2F164iJ-X"
      },
      "source": [
        "Finally we print performances on *test set*:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lIGvgHxXiMZm",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9e1d850b-454f-40a0-fd7d-18a86280adc5"
      },
      "source": [
        "test_loss, test_acc = model5.evaluate(iterator_test)\n",
        "print('Test accuracy:', test_acc)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "96/96 [==============================] - 2s 17ms/step - loss: 1.8061 - accuracy: 0.8571\n",
            "Test accuracy: 0.857142448425293\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "F9D5cQoykPmf"
      },
      "source": [
        "#Emotion assignment on *Testing set*"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6CoNlWC_kvdN"
      },
      "source": [
        "The evaluation of the models defined in the previous section shows that the best solution is given by **Fourth model**, a Convolutional Neural Network with an accuracy of about *90%* on testing. Now, using such a model we try to label images on *Testing set* to produce *emotion* column on the related *.csv* file."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lqirWnvUoEki"
      },
      "source": [
        "First, let's define an ImageDataGenerator for our Testing set:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mNktevqsoTat"
      },
      "source": [
        "test_label_datagen  = ImageDataGenerator(rescale = 1./255)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "K70beoX1owPu"
      },
      "source": [
        "Then the flow:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JVkjZgYTpF16"
      },
      "source": [
        "iterator_test_label = test_label_datagen.flow(X_test_label, shuffle=False)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2wm-qvWnv5ik"
      },
      "source": [
        "Here we compute our predictions:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wQtxsOWzsQT5"
      },
      "source": [
        "predictions = model4.predict(iterator_test_label).argmax(axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "W7Q3xMT7-xZf"
      },
      "source": [
        "In addition, we define a function *readCSVList* used to load our *test csv* from the directory into a numpy array and ordered by *image_id*."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kMmAoWjQ0odp"
      },
      "source": [
        "def readCSVList(directory):\n",
        "    df = pd.read_csv(directory, sep=',',header=0)\n",
        "    df_sorted = df.sort_values(\"image_id\")\n",
        "    npy = df_sorted.to_numpy()\n",
        "    npy = np.delete(npy, obj=2, axis=1)\n",
        "    return npy"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nGUTSUd-_8Hc"
      },
      "source": [
        "We actually load the data from the *.csv* file on the directory."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kVTZ3ltP6idb"
      },
      "source": [
        "test_npy = readCSVList(test_csv_dir)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SQ8k3iDpAMbt"
      },
      "source": [
        "And define a function which adds predicted labels from our model creating the original structure of the *.csv* file."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DBkBCv0o_SZb"
      },
      "source": [
        "def addLabels(numpy_data, labels):\n",
        "  labeled_npy = np.append(numpy_data, np.expand_dims(labels, axis = 1), axis=1)\n",
        "  labeled_df = pd.DataFrame({'': labeled_npy[:, 0], 'image_id': labeled_npy[:, 1], 'emotion': labeled_npy[:, 2]})\n",
        "  resorted_labeled_df = labeled_df.sort_values(\"\")\n",
        "  return resorted_labeled_df"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rUj12cOKAj__"
      },
      "source": [
        "Here we execute the function defined above."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_c5RDX5N_wCB"
      },
      "source": [
        "test_label = addLabels(test_npy, predictions)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FDcn5rFZAqge"
      },
      "source": [
        "And finally we save the labeled *.csv* Test set."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TrEr0qLd8McY"
      },
      "source": [
        "test_label.to_csv('/content/drive/MyDrive/Machine Learning Competition/challengeA_data/challengeA_data/Cavallaro,Salvatore_Answers.csv', index = False)"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}