competitions/SIIM_ISIC_Melanoma_Classification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Melanoma_final.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "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": "3.6.9"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "nPtzZfeclfLs"
      },
      "source": [
        "# SIIM-ISIC Melanoma Classification\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "kRNS-y6lni6Y"
      },
      "source": [
        "O [câncer de pele](https://saude.gov.br/saude-de-a-z/cancer-de-pele) é um grave problema de saúde, sendo o tipo de câncer mais comum no Brasil e no mundo. Dentro das categorias de câncer de pele, o Melanoma é o tipo mais perigoso por sua alta possibilidade de ocasionar metástase e se espalhar para outros órgãos.\n",
        "\n",
        "A competição [SIIM-ISIC Melanoma Classification](https://www.kaggle.com/c/siim-isic-melanoma-classification) tem o objetivo de classificar em benigna e maligna várias imagens suspeitas de possuir câncer de pele Melanoma. A submissão deve conter um arquivo .csv com uma coluna de dados referente ao nome das imagens e outra com as respectivas probabilidades de cada imagem conter uma amostra maligna.\n",
        "\n",
        "A base é composta por 33126 imagens de treinamento e 10982 imagens de teste. E também por dois arquivos .csv com algumas informações sobre os pacientes de cada imagem."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "AzYnjm99ntQy"
      },
      "source": [
        "## Análise da base de dados"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "pz43JPyeWdZp"
      },
      "source": [
        "Para todos os exemplos testados nesse notebook, admitimos que os arquivos train.csv, test.csv e as pastas /train e /test com as imagens .jpg estão no mesmo diretório que esse código."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "c5gcWraNoKXP"
      },
      "source": [
        "Primeiramente, devemos ler os arquivos .csv com as informações da base."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "DCxTfHM6n7Zt",
        "colab": {}
      },
      "source": [
        "import pandas as pd\n",
        "\n",
        "# Leitura dos dataframes dos arquivos csv\n",
        "train_data = pd.read_csv('train.csv')\n",
        "test_data = pd.read_csv('test.csv')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "CGz6BS3HpC5J"
      },
      "source": [
        "Características do arquivo de treinamento."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "35Rf643FpHpd",
        "colab": {},
        "outputId": "68fb04f4-058b-44ea-dd13-11cfe270482e"
      },
      "source": [
        "train_data.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>image_name</th>\n",
              "      <th>patient_id</th>\n",
              "      <th>sex</th>\n",
              "      <th>age_approx</th>\n",
              "      <th>anatom_site_general_challenge</th>\n",
              "      <th>diagnosis</th>\n",
              "      <th>benign_malignant</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>ISIC_2637011</td>\n",
              "      <td>IP_7279968</td>\n",
              "      <td>male</td>\n",
              "      <td>45.0</td>\n",
              "      <td>head/neck</td>\n",
              "      <td>unknown</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>ISIC_0015719</td>\n",
              "      <td>IP_3075186</td>\n",
              "      <td>female</td>\n",
              "      <td>45.0</td>\n",
              "      <td>upper extremity</td>\n",
              "      <td>unknown</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>ISIC_0052212</td>\n",
              "      <td>IP_2842074</td>\n",
              "      <td>female</td>\n",
              "      <td>50.0</td>\n",
              "      <td>lower extremity</td>\n",
              "      <td>nevus</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>ISIC_0068279</td>\n",
              "      <td>IP_6890425</td>\n",
              "      <td>female</td>\n",
              "      <td>45.0</td>\n",
              "      <td>head/neck</td>\n",
              "      <td>unknown</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>ISIC_0074268</td>\n",
              "      <td>IP_8723313</td>\n",
              "      <td>female</td>\n",
              "      <td>55.0</td>\n",
              "      <td>upper extremity</td>\n",
              "      <td>unknown</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "     image_name  patient_id     sex  age_approx anatom_site_general_challenge  \\\n",
              "0  ISIC_2637011  IP_7279968    male        45.0                     head/neck   \n",
              "1  ISIC_0015719  IP_3075186  female        45.0               upper extremity   \n",
              "2  ISIC_0052212  IP_2842074  female        50.0               lower extremity   \n",
              "3  ISIC_0068279  IP_6890425  female        45.0                     head/neck   \n",
              "4  ISIC_0074268  IP_8723313  female        55.0               upper extremity   \n",
              "\n",
              "  diagnosis benign_malignant  target  \n",
              "0   unknown           benign       0  \n",
              "1   unknown           benign       0  \n",
              "2     nevus           benign       0  \n",
              "3   unknown           benign       0  \n",
              "4   unknown           benign       0  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 2
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "7EJI6Q2fpLwK"
      },
      "source": [
        "Vemos que esse dataset possui as informações:\n",
        "\n",
        "* image_name: Nome da imagem referente a esse exemplo de treinamento.\n",
        "* patient_id: ID do paciente.\n",
        "* sex: Sexo do paciente.\n",
        "* age_approx: Idade aproximada do paciente\n",
        "* anatom_site_general_challenge: Parte do corpo onde a foto foi tirada.\n",
        "* diagnosis: Dianóstico.\n",
        "* benign_malignant: Classificação da mancha em maligna ou benigna.\n",
        "* target: 0 (benigno), 1 (maligno)."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "UR5KKzHvq0VA"
      },
      "source": [
        "Características do arquivo de teste."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "IZeabc25qwgW",
        "colab": {},
        "outputId": "8a11ef8e-15ca-46e4-a785-574b013eb05a"
      },
      "source": [
        "test_data.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>image_name</th>\n",
              "      <th>patient_id</th>\n",
              "      <th>sex</th>\n",
              "      <th>age_approx</th>\n",
              "      <th>anatom_site_general_challenge</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>ISIC_0052060</td>\n",
              "      <td>IP_3579794</td>\n",
              "      <td>male</td>\n",
              "      <td>70.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>ISIC_0052349</td>\n",
              "      <td>IP_7782715</td>\n",
              "      <td>male</td>\n",
              "      <td>40.0</td>\n",
              "      <td>lower extremity</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>ISIC_0058510</td>\n",
              "      <td>IP_7960270</td>\n",
              "      <td>female</td>\n",
              "      <td>55.0</td>\n",
              "      <td>torso</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>ISIC_0073313</td>\n",
              "      <td>IP_6375035</td>\n",
              "      <td>female</td>\n",
              "      <td>50.0</td>\n",
              "      <td>torso</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>ISIC_0073502</td>\n",
              "      <td>IP_0589375</td>\n",
              "      <td>female</td>\n",
              "      <td>45.0</td>\n",
              "      <td>lower extremity</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "     image_name  patient_id     sex  age_approx anatom_site_general_challenge\n",
              "0  ISIC_0052060  IP_3579794    male        70.0                           NaN\n",
              "1  ISIC_0052349  IP_7782715    male        40.0               lower extremity\n",
              "2  ISIC_0058510  IP_7960270  female        55.0                         torso\n",
              "3  ISIC_0073313  IP_6375035  female        50.0                         torso\n",
              "4  ISIC_0073502  IP_0589375  female        45.0               lower extremity"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "6kMVnFzGq48P"
      },
      "source": [
        "O arquivo de dados de teste possui quase as mesmas informações que o de treinamento, exceto as últimas colunas (diagnosis, benign_malignant, target), as quais queremos inferir."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "jWowG8Bxrbr9"
      },
      "source": [
        "Um dado interessante a ser observado em qualquer base é a distribuição dos exemplos em cada classe."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "wl4NBHsyTRwq",
        "colab": {},
        "outputId": "b2ac10d6-ee05-47b3-8e3b-0fd902abc0f8"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "labels=train_data['benign_malignant'].value_counts().index\n",
        "values=train_data['benign_malignant'].value_counts().values\n",
        "\n",
        "plt.title('Número de exemplos de cada classe')\n",
        "nclasses = plt.bar(labels, values)\n",
        "nclasses[0].set_color('tab:orange')\n",
        "plt.show()\n",
        "\n",
        "print(\"Número de exemplos benignos: {}\".format(values[0]))\n",
        "print(\"Número de exemplos malignos: {}\".format(values[1]))"
      ],
      "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": "stream",
          "text": [
            "Número de exemplos benignos: 32542\n",
            "Número de exemplos malignos: 584\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "5GPlUDGlUABt"
      },
      "source": [
        "Vemos que essa base possui um forte desbalanceamento, essa característica é muito importante e irá influenciar a maneira como os modelos de redes neurais serão treinados."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Nt17syw96Y6O",
        "colab_type": "text"
      },
      "source": [
        "Também é interessante visualizar algumas das imagens de ambas as classes para verificar as características de manchas que são Melanoma e manchas que não são Melanomas."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SPyHY54G6Y6O",
        "colab_type": "code",
        "colab": {},
        "outputId": "4ea274f3-aa52-4b92-f990-0616776e9aaa"
      },
      "source": [
        "import cv2\n",
        "import numpy as np\n",
        "\n",
        "benign_samples = train_data[train_data['benign_malignant']=='benign'].sample(20)\n",
        "\n",
        "fig, ax = plt.subplots(5,4, figsize=(10,8))\n",
        "\n",
        "for i in range(len(benign_samples)):\n",
        "    img=cv2.imread(str(\"train/\" + benign_samples['image_name'].iloc[i]+'.jpg'))\n",
        "    img = cv2.resize(img, (224,224))\n",
        "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
        "    img = img.astype(np.float32)/255.\n",
        "    ax[i//4, i%4].imshow(img)\n",
        "    ax[i//4, i%4].axis('off')\n",
        "\n",
        "fig.suptitle('Imagens de amostras benignas', fontsize=22)       \n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x576 with 20 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Oi3xyYUV6Y6T",
        "colab_type": "code",
        "colab": {},
        "outputId": "f4f31b51-59ba-4530-bcbd-0f9dec11df77"
      },
      "source": [
        "malignant_samples = train_data[train_data['benign_malignant']=='benign'].sample(20)\n",
        "fig, ax = plt.subplots(5,4, figsize=(10,8))\n",
        "\n",
        "for i in range(len(malignant_samples)):\n",
        "    img=cv2.imread(str(\"train/\" + malignant_samples['image_name'].iloc[i]+'.jpg'))\n",
        "    img = cv2.resize(img, (224,224))\n",
        "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
        "    img = img.astype(np.float32)/255.\n",
        "    ax[i//4, i%4].imshow(img)\n",
        "    ax[i//4, i%4].axis('off')\n",
        "        \n",
        "fig.suptitle('Imagens de amostras malignas', fontsize=22)       \n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x576 with 20 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "q9DKFm3r6Y6X",
        "colab_type": "text"
      },
      "source": [
        "Para uma pessoa sem treinamento profissional, identificar se uma amostra é maligna ou benigna é bastante difícil."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "aejAYBkRtn_p"
      },
      "source": [
        "## Avaliação e score\n",
        "\n",
        "Quando treinamos um modelo que será usado para fazer inferências é essencial estabelecer quais métricas serão usadas para avaliá-lo.\n",
        "\n",
        "O score oficial usado para testar as submissões na competição é a [área abaixo da curva ROC](https://developers.google.com/machine-learning/crash-course/classification/roc-and-auc) das probabilidades preditas e os valores esperados.\n",
        "\n",
        "### Curva ROC e AUC\n",
        "\n",
        "A curva ROC (Curva Característica de Operação do Receptor) é uma forma de representar os resultados de desempenho de um classificador baseado em parâmetros de taxa de resultados \"Verdadeiro Positivo\" e \"Falso Positivo\".\n",
        "\n",
        "Temos que:\n",
        "\n",
        ">$Taxa De Verdadeiros Positivos = \\frac{Verdadeiros Positivos}{Verdadeiros Positivos + Falsos Negativos}$\n",
        "\n",
        ">$Taxa De Falsos Positivos = \\frac{Falsos Positivos}{Falsos Positivos + Verdadeiros Negativos}$\n",
        "\n",
        "Para representação gráfica, são consideradas essas taxas referentes à diferentes valores de threshold do classificador, aumentando e/ou diminuindo os valores esperados como positivos ou negativos.\n",
        "\n",
        "Como ilustrado na imagem abaixo, temos a curva ROC representada a partir de diferentes pontos de treshhold do classificador (pontos A, B e C).\n",
        "\n",
        "<img src=\"https://drive.google.com/uc?id=1mA6SJEf27PR9VJVfonuCqKHED4yMknjP\">\n",
        "\n",
        "Esse tipo de representação é extremamente útil para avaliação de aplicações que trabalham em um ambiente com uma grande desproporção entre classes, possibilitando a consideração de um \"custo/benefício\" para os erros/acertos do classificador.\n",
        "\n",
        "E com o intuito de simplificar e quantizar o desempenho de um classificador representado no gráfico de curva ROC, é utilizada a área abaixo da curva ROC, chamada de AUC, como métrica. Seu valor pode variar entre 0 e 1, sendo que \"quanto maior for seu valor, melhor será o classificador\".\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "EfsW8PqkrlKd"
      },
      "source": [
        "## Construindo um simples modelo\n",
        "\n",
        "Para a criação do modelos de CNN, utilizamos a biblioteca [Keras](https://keras.io/).\n",
        "\n",
        "A primeira submissão para a competição tinha o objetivo de testar um modelo simples e verificar aproximadamente quanto tempo demoraria para a sua execução, com isso avaliaríamos a viabilidade de lidar com essa base tão extensa."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "_EDHt1rfzGWW",
        "colab": {}
      },
      "source": [
        "# Imports\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import Conv2D\n",
        "from tensorflow.keras.layers import MaxPooling2D\n",
        "from tensorflow.keras.layers import Flatten\n",
        "from tensorflow.keras.layers import Dense\n",
        "\n",
        "\n",
        "from tensorflow.keras.metrics import TruePositives\n",
        "from tensorflow.keras.metrics import FalsePositives\n",
        "from tensorflow.keras.metrics import TrueNegatives\n",
        "from tensorflow.keras.metrics import FalseNegatives\n",
        "from tensorflow.keras.metrics import BinaryAccuracy\n",
        "from tensorflow.keras.metrics import Precision\n",
        "from tensorflow.keras.metrics import Recall\n",
        "from tensorflow.keras.metrics import AUC\n",
        "\n",
        "from tensorflow.keras.callbacks import EarlyStopping\n",
        "\n",
        "from sklearn.metrics import roc_auc_score"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "3Ij-uNYisqgj"
      },
      "source": [
        "Primeiramente carregamos os aquivos de informações e adicionamos a extensão de formato das imagens às colunas image_name."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "E5C6daLAs6PO",
        "colab": {}
      },
      "source": [
        "# Leitura dos dataframes dos arquivos csv\n",
        "train_data = pd.read_csv('train.csv')\n",
        "test_data = pd.read_csv('test.csv')\n",
        "\n",
        "# Adicionando a extensão aos nomes das imagens\n",
        "train_data['image_name'] = train_data['image_name'].astype(str)+\".jpg\"\n",
        "test_data['image_name'] = test_data['image_name'].astype(str)+\".jpg\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "lQ7XYAmVs-Kq"
      },
      "source": [
        "Para lidar com imagens, o Keras possui a ferramenta [ImageDataGenerator](https://keras.io/api/preprocessing/image/). Essa classe foi criada para lidar com grandes bases de dados de imagens e permite um carregamento em pacotes de imagens para gerar menos impacto na memória RAM do computador. Além disso, ainda faz o pré-processamento das imagens, colocando-as em um determinado tamanho e podendo até mesmo aplicar filtros e outros métodos de visão computacional.\n",
        "\n",
        "Para facilitar ainda mais, essa classe possui o método [flow_from_dataframe](https://keras.io/api/preprocessing/image/#flowfromdataframe-method), que faz o carregamento das imagens por meio de dados de um dataframe pandas.\n",
        "\n",
        "Nessa primeira submissão, escolhemos um tamanho de imagem pequeno de 32x32 pixels e um tamanho de pacote de 32, para testar a construção do modelo. Além disso dividimos o dataset de treinamento em 70% de imagens de treinamento e 30% de imagens de validação."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "OfeCO0haurOl",
        "colab": {},
        "outputId": "81ec1bef-cfbe-4670-aea7-eb2f8384553a"
      },
      "source": [
        "image_height = 32\n",
        "image_width = 32\n",
        "batch_size = 32\n",
        "\n",
        "# Carregando as imagens de treinamento com ImageDataGenerator\n",
        "image_datagen = ImageDataGenerator(rescale = 1./255,          # normaliza valores dos pixels da imagem entre 0-1\n",
        "                                   validation_split = 0.3)    # divide os dados do dataset em uma proporção de treinamento e validação\n",
        "\n",
        "\n",
        "train_generator = image_datagen.flow_from_dataframe(dataframe=train_data,                   # dataframe com dados da imagem\n",
        "                                                    directory=\"train\",                      # diretório com as imagens\n",
        "                                                    x_col=\"image_name\",                     # nome da coluna do dataframe com os nomes das imagens\n",
        "                                                    y_col=\"benign_malignant\",               # nome da coluna do dataframe com a especificação das classes\n",
        "                                                    class_mode=\"binary\",                    # modo binário, irá selecionar as imagens em duas classes\n",
        "                                                    target_size=(image_height,image_width), # tamanho das imagens de treinamento\n",
        "                                                    batch_size=batch_size,                  # quantidade de imagens por pacote\n",
        "                                                    subset=\"training\",                      # subset de treinamento ou validação\n",
        "                                                    color_mode=\"rgb\")                       # modo de carregamento da imagem em 3 canais RGB\n",
        "\n",
        "validation_generator = image_datagen.flow_from_dataframe(dataframe=train_data,\n",
        "                                                         directory=\"train\",\n",
        "                                                         x_col=\"image_name\",\n",
        "                                                         y_col=\"benign_malignant\",\n",
        "                                                         class_mode=\"binary\",\n",
        "                                                         target_size=(image_height,image_width),\n",
        "                                                         batch_size=batch_size,\n",
        "                                                         subset=\"validation\",\n",
        "                                                         color_mode=\"rgb\")\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 23189 validated image filenames belonging to 2 classes.\n",
            "Found 9937 validated image filenames belonging to 2 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "BXEkRwN8xyIJ"
      },
      "source": [
        "Em seguida, construimos um modelo simples, com apenas uma camada de convolução usando a biblioteca Keras. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "T0g1km-nyORE",
        "colab": {},
        "outputId": "7facf7b5-42a6-4de1-80d1-bc541b86ec2c"
      },
      "source": [
        "model = Sequential()\n",
        "model.add(Conv2D(filters=32, kernel_size=(7,7), input_shape=(32,32,3), activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size = (2, 2)))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(units = 100, activation = 'relu'))\n",
        "model.add(Dense(units = 1, activation = 'sigmoid'))\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 26, 26, 32)        4736      \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 13, 13, 32)        0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 5408)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 100)               540900    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 1)                 101       \n",
            "=================================================================\n",
            "Total params: 545,737\n",
            "Trainable params: 545,737\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "ytfXXz_9z0vb"
      },
      "source": [
        "Para compilar o nosso modelo, escolhemos o otimizador [Adam](https://machinelearningmastery.com/adam-optimization-algorithm-for-deep-learning/) e a funcão de loss [binary_crossentropy](https://keras.io/api/losses/probabilistic_losses/#binarycrossentropy-class), baseado em modelos que obtiveram sucesso em submissões públicas dessa mesma competição no kaggle.\n",
        "\n",
        "Para avaliar o nosso modelo durante o treinamento, iremos observar:\n",
        "\n",
        "* True Positives: amostras com cancer corretamente classificadas.\n",
        "* False Positives: amostras sem câncer erroneamente classificadas.\n",
        "* True Negatives: amostras sem câncer corretamente classificadas.\n",
        "* False Negatives: amostras com câncer erroneamente classificadas.\n",
        "* Binary Accuracy: Acurácia do modelo.\n",
        "* Precision: Precisão do modelo.\n",
        "* Recall: Revocação do modelo.\n",
        "* AUC: Área aproximada abaixo da curva ROC"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "UyG33q7XnBzz",
        "colab": {}
      },
      "source": [
        "METRICS = [\n",
        "      TruePositives(name='tp'),\n",
        "      FalsePositives(name='fp'),\n",
        "      TrueNegatives(name='tn'),\n",
        "      FalseNegatives(name='fn'), \n",
        "      BinaryAccuracy(name='accuracy'),\n",
        "      Precision(name='precision'),\n",
        "      Recall(name='recall'),\n",
        "      AUC(name='auc'),\n",
        "]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "HzXMjHTGzw1l",
        "colab": {}
      },
      "source": [
        "model.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = METRICS)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "Mamxadka6mrH"
      },
      "source": [
        "Para fazer o treinamento, é útil usar uma função de callback para parar esse treinamento quando o modelo não está mais evoluindo. Por isso, utilizaremos a função [EarlyStopping](https://keras.io/api/callbacks/early_stopping/), que irá acompanhar a medida AUC do dataset de validação ('val_auc')."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "XnncNLUs6kN3",
        "colab": {}
      },
      "source": [
        "early_stopping = EarlyStopping(\n",
        "    monitor='val_auc', \n",
        "    verbose=1,\n",
        "    patience=3,\n",
        "    mode='max',\n",
        "    restore_best_weights=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "6KQW5Plgyc0K"
      },
      "source": [
        "Para fazer o treinamento, usamos o método fit com as imagens de treinamento e validação. O parâmetro workers faz paralelização do processamento quando múltiplos cores de processamento estão disponíveis."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "MrZbFiySy59S",
        "colab": {},
        "outputId": "df4c4fc2-3585-4e5c-d59f-d45a933eb3d6"
      },
      "source": [
        "# Fazer o Treinamento\n",
        "STEP_SIZE_TRAIN = train_generator.n//train_generator.batch_size\n",
        "STEP_SIZE_VALID = validation_generator.n//validation_generator.batch_size\n",
        "\n",
        "model.fit(train_generator,\n",
        "          steps_per_epoch=STEP_SIZE_TRAIN,\n",
        "          validation_data=validation_generator,\n",
        "          validation_steps=STEP_SIZE_VALID,\n",
        "          epochs=50,\n",
        "          callbacks=early_stopping,\n",
        "          workers = 16)\n",
        "\n",
        "model.save_weights('simple_model_weights.h5')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "724/724 [==============================] - 320s 442ms/step - loss: 0.0926 - tp: 1.0000 - fp: 31.0000 - tn: 22720.0000 - fn: 405.0000 - accuracy: 0.9812 - precision: 0.0312 - recall: 0.0025 - auc: 0.6217 - val_loss: 0.0875 - val_tp: 0.0000e+00 - val_fp: 0.0000e+00 - val_tn: 9743.0000 - val_fn: 177.0000 - val_accuracy: 0.9822 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00 - val_auc: 0.7200\n",
            "Epoch 2/50\n",
            "724/724 [==============================] - 306s 422ms/step - loss: 0.0871 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 22751.0000 - fn: 406.0000 - accuracy: 0.9825 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.6664 - val_loss: 0.0847 - val_tp: 0.0000e+00 - val_fp: 0.0000e+00 - val_tn: 9743.0000 - val_fn: 177.0000 - val_accuracy: 0.9822 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00 - val_auc: 0.7414\n",
            "Epoch 3/50\n",
            "724/724 [==============================] - 300s 415ms/step - loss: 0.0871 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 22751.0000 - fn: 406.0000 - accuracy: 0.9825 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.6548 - val_loss: 0.0852 - val_tp: 0.0000e+00 - val_fp: 0.0000e+00 - val_tn: 9743.0000 - val_fn: 177.0000 - val_accuracy: 0.9822 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00 - val_auc: 0.7443\n",
            "Epoch 4/50\n",
            "724/724 [==============================] - 300s 415ms/step - loss: 0.0853 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 22752.0000 - fn: 405.0000 - accuracy: 0.9825 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.6914 - val_loss: 0.0832 - val_tp: 0.0000e+00 - val_fp: 0.0000e+00 - val_tn: 9743.0000 - val_fn: 177.0000 - val_accuracy: 0.9822 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00 - val_auc: 0.7475\n",
            "Epoch 5/50\n",
            "724/724 [==============================] - 295s 408ms/step - loss: 0.0854 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 22750.0000 - fn: 407.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.6755 - val_loss: 0.0843 - val_tp: 0.0000e+00 - val_fp: 0.0000e+00 - val_tn: 9744.0000 - val_fn: 176.0000 - val_accuracy: 0.9823 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00 - val_auc: 0.7614\n",
            "Epoch 6/50\n",
            "724/724 [==============================] - 297s 411ms/step - loss: 0.0838 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 22750.0000 - fn: 407.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.7062 - val_loss: 0.0871 - val_tp: 0.0000e+00 - val_fp: 0.0000e+00 - val_tn: 9744.0000 - val_fn: 176.0000 - val_accuracy: 0.9823 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00 - val_auc: 0.7816\n",
            "Epoch 7/50\n",
            "724/724 [==============================] - 297s 411ms/step - loss: 0.0843 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 22750.0000 - fn: 407.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.6968 - val_loss: 0.0821 - val_tp: 0.0000e+00 - val_fp: 0.0000e+00 - val_tn: 9743.0000 - val_fn: 177.0000 - val_accuracy: 0.9822 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00 - val_auc: 0.7557\n",
            "Epoch 8/50\n",
            "724/724 [==============================] - 295s 408ms/step - loss: 0.0833 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 22763.0000 - fn: 405.0000 - accuracy: 0.9825 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.7061 - val_loss: 0.0853 - val_tp: 0.0000e+00 - val_fp: 0.0000e+00 - val_tn: 9743.0000 - val_fn: 177.0000 - val_accuracy: 0.9822 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00 - val_auc: 0.7807\n",
            "Epoch 9/50\n",
            "724/724 [==============================] - ETA: 0s - loss: 0.0827 - tp: 1.0000 - fp: 0.0000e+00 - tn: 22750.0000 - fn: 406.0000 - accuracy: 0.9825 - precision: 1.0000 - recall: 0.0025 - auc: 0.7279Restoring model weights from the end of the best epoch.\n",
            "724/724 [==============================] - 305s 421ms/step - loss: 0.0827 - tp: 1.0000 - fp: 0.0000e+00 - tn: 22750.0000 - fn: 406.0000 - accuracy: 0.9825 - precision: 1.0000 - recall: 0.0025 - auc: 0.7279 - val_loss: 0.0826 - val_tp: 0.0000e+00 - val_fp: 0.0000e+00 - val_tn: 9743.0000 - val_fn: 177.0000 - val_accuracy: 0.9822 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00 - val_auc: 0.7718\n",
            "Epoch 00009: early stopping\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "6b922Y8wztv8"
      },
      "source": [
        "Antes de fazer a inferência dos dados de teste, vamos analisar o resultado nos dados de validação."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "OcE0HavTz8ak",
        "colab": {},
        "outputId": "38bea391-e86c-4d25-b7a8-4aec55f6875c"
      },
      "source": [
        "validation_predictions = model.predict(validation_generator,\n",
        "                                       verbose=1,\n",
        "                                       workers=16)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "311/311 [==============================] - 86s 278ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "W8BGI4rO86m3"
      },
      "source": [
        "Usamos o cálculo do score AUC da biblioteca scikitLearn para obter um resultado mais fiel dessa estatística. Porque a métrica AUC da biblioteca keras utiliza uma aproximação por [Soma de Riemann](http://ecalculo.if.usp.br/integrais/soma_riemann/soma_riemann.htm#:~:text=Soma%20de%20riemann,x%2C%20para%20x%20variando%20em%20.) da área abaixo da curva e costuma gerar valores maiores que o real."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "AB6km99T85E5",
        "colab": {},
        "outputId": "79804a92-f2ff-4b50-9327-1693ff4158f7"
      },
      "source": [
        "val_auc = roc_auc_score(y_true = validation_generator.classes, y_score=validation_predictions)\n",
        "print(val_auc)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.5018135824766139\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GswUw24xah1e",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "Como vemos, a métrica do Scikit reduziu bastante o valor AUC em comparação com o os mostrados durante o treinamento."
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "UCHzfNF89R7P"
      },
      "source": [
        "Para visualizar graficamente, construimos a matriz de confusão das predições de validação."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "RtDvHmcqaqpF",
        "colab": {}
      },
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "from sklearn.metrics import accuracy_score\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "\n",
        "def plot_confusion_matrix(y_true, y_pred, classes, title):\n",
        "    acc = accuracy_score(y_true, y_pred)\n",
        "    title = title + \" (Acurácia: \" + str(\"{:10.4f}\".format(acc)) + \")\"\n",
        "\n",
        "    cm = confusion_matrix(y_true, y_pred, normalize='true')\n",
        "    cm_df = pd.DataFrame(cm, index = classes, columns = classes)\n",
        "    plt.figure(figsize=(5.5,4))\n",
        "    sns.heatmap(cm_df, annot=True, cmap=\"YlGnBu\")\n",
        "    plt.title(title)\n",
        "    plt.ylabel('Classe verdadeira')\n",
        "    plt.xlabel('Classe predita')\n",
        "    plt.show()\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "sL_wnfUg9QdV",
        "colab": {},
        "outputId": "121528ce-86e6-46e7-f69f-f55328e17ee8"
      },
      "source": [
        "# Plota a matriz de confusão dos dados de validação\n",
        "plot_confusion_matrix(validation_generator.classes,\n",
        "                      (validation_predictions>0.5).astype(int),\n",
        "                      ['benign','malignant'], \"Matriz de confusão\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 396x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "QkH40LusbIPP"
      },
      "source": [
        "Como podemos ver, apesar de apresentar uma alta acurácia, o modelo não classificou nenhuma imagem como melanoma. Isso provavelmente se deve a simplicidade do modelo, redução massiva do tamanho das imagens e ao grande desbalanceamento da base."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "9Ap8J_5LbiwF"
      },
      "source": [
        "Para fazer a primeira submissão, inferimos os dados de teste."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "RKhqrqtfbxNX",
        "colab": {},
        "outputId": "d07d58e4-4204-4709-b8c1-a6cce707f60d"
      },
      "source": [
        "test_datagen = ImageDataGenerator(rescale = 1./255)\n",
        "test_generator = test_datagen.flow_from_dataframe(dataframe=test_data,\n",
        "                                                  directory=\"test\",\n",
        "                                                  x_col=\"image_name\",\n",
        "                                                  y_col=None,\n",
        "                                                  class_mode=None,\n",
        "                                                  target_size=(image_height,image_width),\n",
        "                                                  batch_size=batch_size)\n",
        "\n",
        "# Realiza predição\n",
        "predictions = model.predict(test_generator,\n",
        "                            verbose=1,\n",
        "                            workers=16)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 10982 validated image filenames.\n",
            "344/344 [==============================] - 98s 285ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "yBAzFLzBega-",
        "colab": {}
      },
      "source": [
        "submission = pd.DataFrame(test_data['image_name'].str.replace(r'.jpg', ''))\n",
        "submission['target']=predictions\n",
        "\n",
        "# Cria arquivo de submissão\n",
        "submission.to_csv('simple_model_submission.csv',index=False)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HLCBwrW87GHW",
        "colab_type": "text"
      },
      "source": [
        "O score oficial dessa submissão foi 0.506."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "0bS-Q1vQ7O5Q"
      },
      "source": [
        "## Modelo de rede neural VGG\n",
        "\n",
        "A segunda submissão tinha como objetivo implementar um modelo de classificação de imagem já testado em bases de classificação famosas, uma rede neural [VGG](https://becominghuman.ai/what-is-the-vgg-neural-network-a590caa72643).\n",
        "\n",
        "Nossa implementação possui uma arquitetura semelhante a VGG Network, usando várias camadas de convolução seguidas de camadas de pooling, criando uma rede neural convolucional profunda.\n",
        "\n",
        "Essa rede terá como entrada imagens de 224x224 pixels."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "tummWV6-AGMy",
        "colab": {},
        "outputId": "ef5b0cad-078a-404d-fddb-999bc2acad06"
      },
      "source": [
        "# Carregamento das imagens\n",
        "\n",
        "train_data = pd.read_csv('train.csv')\n",
        "test_data = pd.read_csv('test.csv')\n",
        "\n",
        "train_data['image_name'] = train_data['image_name'].astype(str)+\".jpg\"\n",
        "test_data['image_name'] = test_data['image_name'].astype(str)+\".jpg\"\n",
        "\n",
        "batch_size = 16       # Tamanho do pacote\n",
        "image_height = 224    # Altura da imagem\n",
        "image_width = 224     # Largura da imagem\n",
        "\n",
        "# Realiza as augmentações nas imagens\n",
        "image_datagen = ImageDataGenerator(rescale = 1./255,\n",
        "                                   validation_split = 0.3)\n",
        "\n",
        "train_generator = image_datagen.flow_from_dataframe(dataframe=train_data,\n",
        "                                                    directory=\"train\",\n",
        "                                                    x_col=\"image_name\",\n",
        "                                                    y_col=\"benign_malignant\",\n",
        "                                                    class_mode=\"binary\",\n",
        "                                                    target_size=(image_height,image_width),\n",
        "                                                    batch_size=batch_size,\n",
        "                                                    subset=\"training\",\n",
        "                                                    color_mode=\"rgb\")\n",
        "\n",
        "validation_generator = image_datagen.flow_from_dataframe(dataframe=train_data,\n",
        "                                                         directory=\"train\",\n",
        "                                                         x_col=\"image_name\",\n",
        "                                                         y_col=\"benign_malignant\",\n",
        "                                                         class_mode=\"binary\",\n",
        "                                                         target_size=(image_height,image_width),\n",
        "                                                         batch_size=batch_size,\n",
        "                                                         subset=\"validation\",\n",
        "                                                         color_mode=\"rgb\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 23189 validated image filenames belonging to 2 classes.\n",
            "Found 9937 validated image filenames belonging to 2 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "sY79uCVGBAAv",
        "colab": {},
        "outputId": "32da8186-3754-4acb-e82a-2125743cc94f"
      },
      "source": [
        "# Criação do modelo\n",
        "\n",
        "model = Sequential()\n",
        "model.add(Conv2D(input_shape=(image_height,image_width,3),filters=64,kernel_size=(3,3),padding=\"same\", activation=\"relu\"))\n",
        "model.add(Conv2D(filters=64,kernel_size=(3,3),padding=\"same\", activation=\"relu\"))\n",
        "model.add(MaxPooling2D(pool_size=(2,2),strides=(2,2)))\n",
        "model.add(Conv2D(filters=128, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(Conv2D(filters=128, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(MaxPooling2D(pool_size=(2,2),strides=(2,2)))\n",
        "model.add(Conv2D(filters=256, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(Conv2D(filters=256, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(Conv2D(filters=256, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(MaxPooling2D(pool_size=(2,2),strides=(2,2)))\n",
        "model.add(Conv2D(filters=512, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(Conv2D(filters=512, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(Conv2D(filters=512, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(MaxPooling2D(pool_size=(2,2),strides=(2,2)))\n",
        "model.add(Conv2D(filters=512, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(Conv2D(filters=512, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(Conv2D(filters=512, kernel_size=(3,3), padding=\"same\", activation=\"relu\"))\n",
        "model.add(MaxPooling2D(pool_size=(2,2),strides=(2,2)))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(units=4096,activation=\"relu\"))\n",
        "model.add(Dense(units=4096,activation=\"relu\"))\n",
        "model.add(Dense(units=1, activation=\"softmax\"))\n",
        "model.summary()\n",
        "\n",
        "model.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = METRICS)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_1 (Conv2D)            (None, 224, 224, 64)      1792      \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 224, 224, 64)      36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 112, 112, 64)      0         \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 112, 112, 128)     73856     \n",
            "_________________________________________________________________\n",
            "conv2d_4 (Conv2D)            (None, 112, 112, 128)     147584    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 56, 56, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_5 (Conv2D)            (None, 56, 56, 256)       295168    \n",
            "_________________________________________________________________\n",
            "conv2d_6 (Conv2D)            (None, 56, 56, 256)       590080    \n",
            "_________________________________________________________________\n",
            "conv2d_7 (Conv2D)            (None, 56, 56, 256)       590080    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 28, 28, 256)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_8 (Conv2D)            (None, 28, 28, 512)       1180160   \n",
            "_________________________________________________________________\n",
            "conv2d_9 (Conv2D)            (None, 28, 28, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "conv2d_10 (Conv2D)           (None, 28, 28, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "max_pooling2d_4 (MaxPooling2 (None, 14, 14, 512)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_11 (Conv2D)           (None, 14, 14, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "conv2d_12 (Conv2D)           (None, 14, 14, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "conv2d_13 (Conv2D)           (None, 14, 14, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "max_pooling2d_5 (MaxPooling2 (None, 7, 7, 512)         0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 25088)             0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 4096)              102764544 \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 4096)              16781312  \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 1)                 4097      \n",
            "=================================================================\n",
            "Total params: 134,264,641\n",
            "Trainable params: 134,264,641\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "sAoyCfMDKM0f",
        "colab": {},
        "outputId": "b24d46a7-14d0-4a22-ab6f-c626a25a6225"
      },
      "source": [
        "# Realiza treinamento\n",
        "\n",
        "STEP_SIZE_TRAIN = train_generator.n//train_generator.batch_size\n",
        "STEP_SIZE_VALID = validation_generator.n//validation_generator.batch_size\n",
        "\n",
        "model.fit(train_generator,\n",
        "          steps_per_epoch=STEP_SIZE_TRAIN,\n",
        "          validation_data=validation_generator,\n",
        "          validation_steps=STEP_SIZE_VALID,\n",
        "          epochs=50,\n",
        "          callbacks=early_stopping,\n",
        "          workers = 16)\n",
        "\n",
        "model.save_weights('vgg_model_weights.h5')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "1449/1449 [==============================] - 349s 241ms/step - loss: 14.9815 - tp: 407.0000 - fp: 22766.0000 - tn: 9743.0000 - fn: 177.0000 - accuracy: 0.3067 - precision: 0.0176 - recall: 0.6969 - auc: 0.5230 - val_loss: 14.9775 - val_tp: 177.0000 - val_fp: 9759.0000 - val_tn: 0.0000e+00 - val_fn: 0.0000e+00 - val_accuracy: 0.0178 - val_precision: 0.0178 - val_recall: 1.0000 - val_auc: 0.5000\n",
            "Epoch 2/50\n",
            "1449/1449 [==============================] - 362s 250ms/step - loss: 14.9815 - tp: 407.0000 - fp: 22766.0000 - tn: 0.0000e+00 - fn: 0.0000e+00 - accuracy: 0.0176 - precision: 0.0176 - recall: 1.0000 - auc: 0.5000 - val_loss: 14.9775 - val_tp: 177.0000 - val_fp: 9759.0000 - val_tn: 0.0000e+00 - val_fn: 0.0000e+00 - val_accuracy: 0.0178 - val_precision: 0.0178 - val_recall: 1.0000 - val_auc: 0.5000\n",
            "Epoch 3/50\n",
            "1449/1449 [==============================] - 360s 249ms/step - loss: 14.9815 - tp: 407.0000 - fp: 22766.0000 - tn: 0.0000e+00 - fn: 0.0000e+00 - accuracy: 0.0176 - precision: 0.0176 - recall: 1.0000 - auc: 0.5000 - val_loss: 14.9775 - val_tp: 177.0000 - val_fp: 9759.0000 - val_tn: 0.0000e+00 - val_fn: 0.0000e+00 - val_accuracy: 0.0178 - val_precision: 0.0178 - val_recall: 1.0000 - val_auc: 0.5000\n",
            "Epoch 4/50\n",
            "1449/1449 [==============================] - ETA: 0s - loss: 14.9815 - tp: 407.0000 - fp: 22766.0000 - tn: 0.0000e+00 - fn: 0.0000e+00 - accuracy: 0.0176 - precision: 0.0176 - recall: 1.0000 - auc: 0.5000Restoring model weights from the end of the best epoch.\n",
            "1449/1449 [==============================] - 367s 253ms/step - loss: 14.9815 - tp: 407.0000 - fp: 22766.0000 - tn: 0.0000e+00 - fn: 0.0000e+00 - accuracy: 0.0176 - precision: 0.0176 - recall: 1.0000 - auc: 0.5000 - val_loss: 14.9775 - val_tp: 177.0000 - val_fp: 9759.0000 - val_tn: 0.0000e+00 - val_fn: 0.0000e+00 - val_accuracy: 0.0178 - val_precision: 0.0178 - val_recall: 1.0000 - val_auc: 0.5000\n",
            "Epoch 00004: early stopping\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4-Z06N3z6Y7j",
        "colab_type": "text"
      },
      "source": [
        "Como podemos perceber pela saída do método fit, as métricas de avaliação do modelo não mudaram com o passar das épocas. Isso indica que o modelo não convergiu, isso pode ser ter sido causado pela alta quantidade de camadas, pelo grande desbalanceamento da base, pela função de loss. \n",
        "\n",
        "Pelos testes realizados, mudar parâmetros como função de loss, tamanho do batch das imagens e função de saída da última camada densa não foram suficientes para evitar a não convergência."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "d0CzCvTxYrD3",
        "colab": {},
        "outputId": "39647e63-e20c-4159-9bd2-9803cf231ca5"
      },
      "source": [
        "validation_predictions = model.predict(validation_generator,\n",
        "                                       verbose=1,\n",
        "                                       workers=16)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "  2/622 [..............................] - ETA: 21sWARNING:tensorflow:Callbacks method `on_predict_batch_end` is slow compared to the batch time (batch time: 0.0124s vs `on_predict_batch_end` time: 0.0504s). Check your callbacks.\n",
            "622/622 [==============================] - 94s 152ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "STbHsr2B_9cY",
        "colab": {},
        "outputId": "52c6e144-d7c9-4d7a-8bba-4372a5d79e79"
      },
      "source": [
        "val_auc = roc_auc_score(y_true = validation_generator.classes, y_score=validation_predictions)\n",
        "print(val_auc)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.5\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "znAML0eu6Y7p",
        "colab_type": "text"
      },
      "source": [
        "A medida AUC dos dados de validação deu exatamente 0.5, indicando que o modelo pode estar classificando todos os exemplos em apenas uma classe.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "-P8A1NqrYsyy",
        "colab": {},
        "outputId": "19470f8a-c6d2-4107-c0c5-1629fac21513"
      },
      "source": [
        "# Plota a matriz de confusão dos dados de validação\n",
        "plot_confusion_matrix(validation_generator.classes,\n",
        "                      (validation_predictions>0.5).astype(int),\n",
        "                      ['benign','malignant'], \"Matriz de confusão\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVQAAAEWCAYAAAA5Am/SAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nO3debwdRZ338c83CwQkLCZMQHYkomyCIougIgMCoiCDoIAjKBB9xmUQhVEfDSE47joqohgQWWUdRyPgICLLAwgSAgTDGtkSliABQthJ+D1/VB3oe7hL36T7ntv3ft959Sunl1NVZ7m/U13VVa2IwMzMlt2IThfAzGyocEA1M6uIA6qZWUUcUM3MKuKAamZWEQdUM7OKOKAWSDpI0h8rSOdUSd+ookxVkLSCpN9LWijp/GVMa7KkByRtLunyZUzrU5J+tCxpLCtJG0q6X9I6JY9/WtKGdZdrKJC0vKQ7JK3e6bIMlEEfUCXdJ+lFSePbtt8kKSStXyKN9fOxo3o7LiLOioj3LVuJB6UPAxOAcRGx3zKmtQWwM/BfwP9b2kQkLQd8Dfhe2/aVctD6wzKVsrwTgUkRMbfMwRGxUkTcU3OZ+pSD1SmSnpL0iKQj+zj+C/m4p/Lzli/sO07SrZIWS5rS9ryv5s+jtTwn6eXW36Ok10s6V9ICSY9JOkvSygAR8QJwCvDlyt+AQWrQB9TsXuCA1oqkzYEVq8ygr2DbcOsBd0XE4mVNKCI+HBFzImKXiJi8DEntDdwREQ+2bd8XeAHYVdIay5B+t4qfc66VnhIRl1SdzwCYAkwkfbbvBY6WtHt3B0rajRTU/jkfvyFwbOGQOcDRwEXtz42Ib+YfkZUiYiXgO8AVEfFYPuQbwGrABsAbST/cUwpJ/Bo4uBjAh7KmBNQzgI8X1g8GTi8eIGnPXGt9StLctl/aq/L/T+Zf2e0lHSLpGkn/JWkBMCVvuzqnd3TbL/NLkk7trnCStpI0U9IiSecCY9r2f0DSzZKelHStpC16eqGSNpV0qaTHJc2X9NW8fXlJP5L0UF5+1PqSStpJ0jxJX5T0qKSHJX0i7zsWmAx8JL+OQyVNkXRmIc8uNfj8PtyTX8+9kg7K298o6c9ttZFVC+m8RdIV+XXOlrRXT68T2AO4spvtB5NqjbOAj7W9Nzvm9+/J/BkfkrdfIemwwnGvfI55PSR9RtLdwN1524+Ba4Fpkm6U9K7C8SNzzezv+T24MQffVlob5ce9fefqdjBwXEQ8ERG3AycBh/Ry7C8jYnZEPAEcVzw2Ik6LiD8Ai3rLUJJIf4enFTZvAPw2Ip6KiIXA/wCbFtKeBzwBbNe/l9dQETGoF+A+YBfgTuAtwEhgHumXNoD183E7AZuTfiS2AOYDH8r71s/HjiqkewiwGPgcMApYIW+7upsyrAM8BOzRzb7lgPuBLwCjSafXLwHfyPu3Ah4Fts1lPzi/puW7SWss8DDwRVJQHgtsm/dNBa4D/glYnRQMjiu89sX5mNHA+4FngdXy/inAmYV82tdfeX+A1wFPARvnfWsCm+bHGwG7AsvnMlwF/CjvG02q6Xw1vyc7k/5AN+7hc70B2K9t23rAy8Am+T2Y1bZvEelMZTQwDtgy77sCOKzts726sB7ApcDrgRXyto/nNEYBXwIeAcbkfUcBtwIbAwLeSmouaaW1UV/fubx/FnBgDX8Tq+VyTChs+zBwaw/H3wJ8pLA+Pj9/XNtxZwJTesn33cDTwEqFbR8ALs5lWg34M3BE2/OmA5/vdCwZiKUpNVR4tZa6K3A70OVUMSKuiIhbI+LliJgFnA28p480H4qI4yNicUQ8190BklYAfgv8ONKveLvtSH/gP4qIlyLiAlKwaJkE/CIiro+IJRFxGumUtrtf7A8Aj0TEDyLi+YhYFBHX530HAVMj4tGI+AfplO1fC899Ke9/KSIuJn3xN+7j9ffkZWAzSStExMMRMRsg0qn+pRHxQi7DD3n1Pd4OWAn4dkS8GBF/Bi6k0FTTZlVeWyP6V1IQvQ04B9hU0lZ534HAnyLi7PwaF0TEzf14Td+KiMdbn3NEnJ7TWBwR3yf9gLXer8OAr0XEnZHcEhEL2hPs6zsXEVtExK/7UcayVsr/LyxsW0j6Ae7p+PZj6eX4nhwMXBARTxe2zST9gC7IyxLgZ23PW0T6vIe8pgXUA0m1j9Pbd0raVtLlkv4haSHwadIvcW/KdET8ErgzIr7Tw/43AA9G/inO7i88Xg/4Yj5NfVLSk6Qa7xu6SWsd4O+95FNM9/62NBZE1zbSZ3n1D6+0iHgG+Ajp/XtY0kWS3gwgaYKkcyQ9KOkpUo2m9R6/AZgbES+3lXGtHrJ6gtf+QX8cOCuX40FSk8DBeV9v700ZXT5rSZ/Np+tzJd1Heq9ar6VUXkv5natCK6CtXNi2Mj2fsj/dzbH0cvxrSFoR2I+up/sA5wF3kT7LlUnv25ltx4wFniybV5M1JqBGxP2kzqn3A7/p5pBfk04t1omIVUjtcGo9vadke8tT0peBNwGH9nLYw8BauX2pZd3C47nAf0bEqoVlxYg4u5u05pI6DLrzECk4F/N4qLfy9+IZunbqden8iYhLImJX0un+HaT2OYBvkt6zzSNiZVIbZ+t1PwSsI6n4nVqXtjOJglmk9xYASe8kdbJ8Rak3+hFSM8mBuW13LqnTo9+vp/WyCnntAHwd2D8i1omI9UlBp/VaesurqLfvXG0itYM+TGqKaHkrMLuHp8zu5tj53dW6e7EP8DipeaVoS9IZ2DO55noi6W+06C2kZochrzEBNTsU2DnXotqNBR6PiOclbUOqzbb8g3QaW/r6QUl7AJ8H9umpOSD7C6n98vOSRkv6F2Cbwv6TgE/n2owkvS53ZnR3unUhsKakI3In1FhJ2+Z9ZwNfk7S60iUrk3ltTaCsm4F3S1pX0irAVwqve4KkvSW9jtQ08TTpvYP0Hj8NLJS0FqmtseV6Uq346Pw+7AR8kHTq3p2L6dokczCpnXMT0h/plsBmpLbtPUg1110k7S9plKRxkrYsvJ5/kbRi7jDq7QcQ0unny8AzkpaTNJmuteWTgeMkTcyf2RaSxnWTTm/fubqdTvo+rJbPIA4HTu3l2EMlbZI7Eb9WPDZ/XmNI8WCUpDGSRralcTBwetuZGKTmrcOUrnVegdTENauQ9lqktuvrlvJ1NkunG3H7WsidUt1sH0XXTqkPk04xF5EC00/p2vEylRRYnyS19x1CWwdUcRvpC/cSKYC0lhN7KOPWwE0573Pz8o3C/t1JX7wnSTWL84GxPaS1GXAZ6ZT4EeDLefsY4Cf5+Q/nx61OlJ2AeT29b7R1QuVtJ+TyzCH9MbY6pdYknWovzPuvADbJz9kUuDG/FzeTOo7mFdLctPDc20g/Rj19rqOBB0hNBWPy6/1gN8f9jNRuB/AuUuB+ilSLPDhvHw/8Mb//1+TX294ptVFhfSTp+sin8nt5dNv7NZIUdO7Nad4ArN2eFn1/52YDB9X0d7F84TXMB44s7Fs3f0brFrYdmY97CvgVhU5R0nc92pZDCvvXIlUaNuqmHBsAvye1nz4O/C8wsbD/KOCHnY4jA7Uov2izASdpEilYH9Hpslj1lC7ruwV4d0Q82unyDAQHVDOzijStDdXMbJkpDb99VNLfetgvST+RNEfSLElvK5OuA6qZDUenkvo2erIH6aqTiaSOtp+XSdQB1cyGnYi4itSJ1pO9yVc1RMR1wKqS1uwr3UE8IchdbtwdBlZY95hOF8EG0HMPnL3U1+musO4BpWPC83PP+RSpZtkyLSKm9SO7teg6GGRe3vZwb08axAHVzGzp5ODZnwBaCQdUM2uEroPwavcgaQhyy9r0POrvFW5DNbNGGKFRpZcKTAc+nnv7twMWRkSvp/vgGqqZNUSVNVRJZ5NGGI6XNA84hjR6j4g4kTQ0+v2kkYTPAp8ok64Dqpk1Qtf5h5ZNRPQ0rWRrfwCf6W+6Dqhm1hCDv4XSAdXMGmGAO6WWigOqmTWCA6qZWUUq6r2v1eAvoZkZrqGamVXGAdXMrCKq/3Zdy8wB1cwawTVUM7OKjBgx+MPV4C+hmRngC/vNzCriU34zs4o4oJqZVUQ+5Tczq4ZrqGZmFRkxYmSni9AnB1QzawSf8puZVcSn/GZmFXFANTOriE/5zcwqIg89NTOrRpU36auLA6qZNYJP+c3MKuJOKTOzqviU38ysIoO/guqAamYNMWLwR1QHVDNrhsEfTx1QzawZwm2oZmYVGfzx1AHVzBpixOCPqA6oZtYMPuU3M6vISAdUM7NquIZqZlaRwR9PHVDNrCEa0CnVgEtlzcxINdSyS5nkpN0l3SlpjqQvd7N/XUmXS7pJ0ixJ7+8rTddQzawRYmR19T9JI4ETgF2BecANkqZHxG2Fw74GnBcRP5e0CXAxsH5v6bqGambNUG0NdRtgTkTcExEvAucAe7cdE8DK+fEqwEN9JeqAambNIJVeJE2SNKOwTGpLbS1gbmF9Xt5WNAX4mKR5pNrp5/oqok/5zawZ+tEpFRHTgGnLmOMBwKkR8QNJ2wNnSNosIl7usYjLmKGZ2cCo9pT/QWCdwvraeVvRocB5ABHxF2AMML63RB1QzawZ+nHKX8INwERJG0haDvgoML3tmAeAf05Z6y2kgPqP3hL1Kb+ZNUOFQ08jYrGkzwKXACOBUyJitqSpwIyImA58EThJ0hdIHVSHRET0lq4Dqpk1Q8VDTyPiYlJnU3Hb5MLj24Ad+pOmA6qZNcPgHyjlNtTB4qqrbmS33T7NrrtOYtq08ztdHKvJid/7FPfPPJEZl36300VpnBih0kunOKAOAkuWLGHq1BM5+eQpXHTRCVx44VXMmfNAp4tlNTjj/CvZ++Pf7nQxmqnaTqlaOKAOArNm3c16663JOuuswXLLjWbPPd/NZZdd3+liWQ2u+esdPP7k050uRjNVPJa/DrW2oUpaHtiXNP71lbwiYmqd+TbN/PkLWGONVy9vmzBhHLNm3dXBEpkNQhWO5a9L3SX8HWl87GLgmcLSreJwsWnTzq25aGbWKMO9hgqsHRG7lz2463Cxu3q93msomTBhHI888tgr6/PnL2DChHEdLJHZIOT5ULlW0uY159F4m28+kfvue4i5cx/hxRdf4qKLrmLnnbfpdLHMBpcRKr90SN011B2BQyTdC7xAqoxHRGxRc76NMmrUSCZP/jSHHXYMS5a8zL777sLEiet1ulhWg9OO/xzv2v4tjF9tLHOu/ynH/fACTjv3ik4XqxFi8FdQaw+oe9Sc/pDxnvdszXves3Wni2E1O/hzx3e6CM3VgE6pugPqopLbzMx614A21LoD6kzSFFlPkE73VwUekTQfODwibqw5fzMbKgZ/BbX2Il4KvD8ixkfEOFITwIXAvwE/qzlvMxtKPFKK7SLiktZKRPwR2D4irgOWrzlvMxtK3MvPw5L+g3QDLICPAPPzHQd7vI2AmVm76GDNs6y6A+qBwDHAb/P6NXnbSGD/mvM2s6Fk1DAPqBHxGD3fKXBOnXmb2RAzXGuokn4UEUdI+j3p1gFdRMRedeRrZkPYML5s6oz8//drSt/MhpvBH0/LBVRJqwETSXf9AyAirurp+Nb1pRFx5bIW0MwM6OhM/GX1GVAlHQb8O+m+1TcD2wF/AXYu8dwdgCnAejmv1lj+DZe+yGY2LA2FgEoKpu8ArouI90p6M/DNkun/EvgCcCOwZOmKaGZGpbeRrkuZgPp8RDwvCUnLR8QdkjYumf7CiPjDshTQzAwYMr388yStSrqW9FJJTwD3l0z/cknfA35Dmr4PgIiY2e+SmtnwNhRO+SNin/xwiqTLgVWA/y2Z/rb5/+K8dEGJ9lczsy6aHlDzENHZEfFm6H+vfUS8dxnKZmb2iiYMPe11cpSIWALcKWndpUlc0gRJv5T0h7y+iaRDlyYtMxvmRqr80iFlZptaDZgt6TJJ01tLyfRPBS4B3pDX7wKO6H8xzWzYGyKzTX19GdIfHxHnSfoKQEQsluTLp8ys/5rehgrLPNrpGUnjyOP5JW0HLFyG9MxsuBr88bTngCrp6ojYUdIiuk5w0hrttHKJ9I8EpgMbSroGWB348LIU2MyGp0YPPY2IHfP/Y5ch/duA/wGeJd2c77ekdlQzs/5pei9/i6QdJX0iPx4vaYOS6Z8OtIaqHg+8iVdnojIzK68BvfxlJkc5hnRh/sbAr4DlgDOBHUqkv1lEbFJYv1zSbUtTUDMb3kYMkbue7gPsBTwDEBEPAWWbAWbmjigAJG0LzOhvIc3Mqr7pqaTdJd0paY6kL/dwzP6SbpM0W9Kv+0qzzGVTL0ZESGr11L+uREFvJXVkjQaulfRAXl8PuKNEnmZmXVTZhJpHgZ4A7ArMA26QND0ibiscMxH4CrBDRDwh6Z/6SrdMQD1P0i+AVSUdDnwSOKmP53ygRLpmZqWp2k6pbYA5EXFPTvscYG9SR3rL4cAJEfEEQEQ82leiZa5D/b6kXYGnSO2okyPi0j6eU3Y2KjOzUvrThippEjCpsGlaREwrrK8FzC2sz+PVyZxa3pTTuoZ0p+YpEdHrxFClboGSA2ivQdTMrE7qR0DNwXNanwf2bhTp1k87ke5YcpWkzSPiyd6e0K1uLujvouSF/WZmlaj4MtQHgXUK62vnbUXzgOsj4iXgXkl3kQLsDT0l2tuF/WMBJB0HPEy6flTAQcCaS/ECzMyWWsUDpW4AJuZr6h8EPgoc2HbMb4EDgF9JGk9qArin1zKWyHiviPhZRCyKiKci4uekxlszswFT5WVTEbEY+CxpNrzbgfMiYrakqZL2yoddAizI185fDhwVEQt6S7dMG+ozkg4CziE1ARxAvibVzGygVD3yNCIuBi5u2za58DhI85EcWTbNMjXUA4H9gfl52Y/XVo3NzGo1YqRKL51S5rKp+/Apvpl1WAPmRik1ln8McCiwKTCmtT0iPlljuczMumhCQC1zyn8GsAawG3Al6fKCRXUWysysXdVj+etQJqBuFBFfB56JiNOAPXntiAIzs1o14JZSpXr5X8r/PylpM+ARoM9JAszMqtSEU/4yAXWapNVIN+ubDqwETO79KWZm1epk731ZZXr5T84PrwQ2rLc4Zmbda3QNVVKvF7NGxA+rL46ZWfcaHVB5dVb+jYF3kE73AT4I/LXOQpmZtWt0QI2IYwEkXQW8LSIW5fUpwEUDUjozs6wBd5Eu1Sk1AXixsP5i3mZmNmBGjOx0CfpWJqCeDvxV0v/k9Q8Bp9ZWIjOzbjT6lB9A6SYupwN/AN6VN38iIm6qu2BmZkUV31OqFr0G1Hy304sjYnNg5gCVyczsNRoQT0sNPZ0p6R21l8TMrBdNGMtfpg11W+Bjku4jTSwtUuV1izoLZmZW1IQaapmAulvtpTAz68Ooftz1tFP6LGJE3E+6O+DO+fGzZZ5nZlalEYrSS6eUmWD6GGBr0oipXwGjgTOBHeotmpnZq4bKhf37AFuRe/kj4iFJY3t/iplZtZpwWlwmoL6YL58KAEmvq7lMZmav0clT+bLKBP3zJP0CWFXS4cCfgJPqLZaZWVdDYsb+iPi+pF2Bp0jtqJMj4tLaS2ZmVjBqKLSh5nlRz3UQNbNOUgNO+cu0oY4F/ijpceBc4PyImF9vsczMumpCL3+Z61CPjYhNgc8AawJXSvpT7SUzMysY0Y+lU8rUUFseJd3xdAG+66mZDbAm9PKXaUP9N2B/YHXgfODwiLit7oKZmRUNiU4p0rDTIyLi5roLY2bWkya0oZa5bOorA1EQM7PeDIlTfjOzwWBI1FDNzAaDoTKW38ys4xp9yi9pEdB6Ba3KdvDqjP0r11w2M7NXNHqC6YgYGxEr52VsYX2sg6mZDbSqL+yXtLukOyXNkfTlXo7bV1JI2rpMGctkvKOkT+TH4yVtULLMZmaVqHLGfkkjgROAPYBNgAMkbdLNcWOBfweuL1XGEhkfA/wH0Lp8ajnSjP1mZgOm4un7tgHmRMQ9EfEicA6wdzfHHQd8B3i+VBlLHLMPsBfpjqdExEOkCVPMzAZMf075JU2SNKOwTGpLbi1gbmF9Xt72CklvA9aJiIvKltEz9ptZI/TnOtSImAZMW9q8JI0Afggc0p/nlQmo7TP2fxLP2G9mA2zkiEovm3qQNKy+Ze28rWUssBlwhSSANYDpkvaKiBk9JdrfGfvfhGfsN7MOqPiqqRuAibmD/UHgo8CBrZ0RsRAY31qXdAXwpd6CKZS/sP9WYAXSdai39qvYZmYVqPLC/ohYLOmzwCXASOCUiJgtaSowIyKmL026ZabvOwyYDPyZdFH/8ZKmRsQpS5OhmdnSqHosf0RcDFzctm1yD8fuVCbNMjXUo4CtImIBgKRxwLWAA6qZDZihMjnKAmBRYX1R3mZmNmBGN3ws/5H54Rzgekm/I7Wh7g3MGoCymZm9ouk11NbF+3/PS8vv6iuOmVn3Gh1QI+LYgSyImVlvRjY5oLZIWh04GtgUGNPaHhE711guM7MumlBDLXOt7FnAHcAGwLHAfaSLYs3MBkyVs03VVsYSx4yLiF8CL0XElRHxScC1UzMbUKNVfumUMpdNvZT/f1jSnsBDwOvrK5KZ2Ws14ZS/TED9hqRVgC8CxwMrA0fUWiozszaNvqdUS0RcmB8uBN4LIMkB1cwGVBN6+Zd2Apcj+z7EzKw6Fc/YX4ulvY10A34rzGwoacJdT5c2oA7+xgwzG1JGNrkNVdIiug+cIs2NamY2YBpQQe116KlvxGdmg8ZQuWzKzKzjHFDNzCrS6DZUM7PBZCj38puZDSif8puZVaQJI6UcUM2sEZowlr8BrRLDw1VX3chuu32aXXedxLRp53e6OFaTE7/3Ke6feSIzLv1up4vSOCP6sXSKA+ogsGTJEqZOPZGTT57CRRedwIUXXsWcOQ90ulhWgzPOv5K9P/7tThejkZowlt8BdRCYNetu1ltvTdZZZw2WW240e+75bi677PpOF8tqcM1f7+DxJ5/udDEaafSIKL10Sq0BVdJ+ZbYNd/PnL2CNNca/sj5hwjjmz1/QwRKZDT6uocJXSm4DQNIkSTMkzZg27dwai2VmTdOEgFpLL7+kPYD3A2tJ+klh18rA4p6eFxHTgGlp7a7B36VXkQkTxvHII4+9sj5//gImTBjXwRKZDT5NaJ+sq4wPATOA54EbC8t0YLea8myszTefyH33PcTcuY/w4osvcdFFV7Hzztt0ulhmg4pUfumUWmqoEXELcIukX0fES30+YZgbNWokkyd/msMOO4YlS15m3313YeLE9TpdLKvBacd/jndt/xbGrzaWOdf/lON+eAGnnXtFp4vVCE0YKaWI+s6sJe0ATAHWIwVvARERG/b97OFzyj+crbDuMZ0ugg2g5x44e6nD4szHLiodE942fs+OhN+6R0r9EvgC6XR/Sc15mdkQpgaMlKo7oC6MiD/UnIeZDQMNOOOvPaBeLul7wG+AF1obI2Jmzfma2RDTyc6msuoOqNvm/7cubAtg55rzNbMhpup4Kml34MfASODkiPh22/4jgcNIl3r+A/hkRNzfW5q1BtSIeG+d6ZvZ8FHl9H2SRgInALsC84AbJE2PiNsKh90EbB0Rz0r6P8B3gY/0lm7t0/dJ2hPYFBjT2hYRU+vO18yGlopP+bcB5kTEPSltnQPsDbwSUCPi8sLx1wEf6yvRusfyn0iK6J8j1dj3I11CZWbWL+rPUhjGnpdJbcmtBcwtrM/L23pyKNBnB3vdNdR3RsQWkmZFxLGSflCmUGZm7fpTQe06jH0Z85U+RuoHek9fx9YdUJ/L/z8r6Q3AAmDNmvM0syGo4pFSDwLrFNbXztu6kLQL8H+B90TEC+3729UdUC+UtCrwPWAmqYf/5JrzNLMhqOJe/huAiZI2IAXSjwIHdslP2gr4BbB7RDxaJtG6e/mPyw//W9KFwJiIWFhnnmY2NFV5T6mIWCzps8AlpMumTomI2ZKmAjMiYjqpIrgScL5Sj9gDEbFXb+kORC//O4H1W3lJIiJOrztfMxtaqr6wPyIuBi5u2za58HiX/qZZa0CVdAbwRuBmXh3LH4ADqpn1SxPmQ627hro1sEnUOaWVmQ0LTRh6WnfQ/xuwRs15mNkw0J/rUDul7hrqeOA2SX+l6+QovTbsmpm1a8IE03UH1Ck1p29mw8SwD6gRcWWd6ZvZ8NGAeFp7L/8iUq9+0ULSDfy+2JqYwMysL56xH35EmnTg16QfmI+SLqOaCZwC7FRz/mY2RDShhlp3L/9eEfGLiFgUEU/lCQt2i4hzgdVqztvMhpAm3Ea67oD6rKT9JY3Iy/7A83nf4K+/m9mgMbIfS6fUHVAPAv4VeBSYnx9/TNIKwGdrztvMhpAm1FDr7uW/B/hgD7uvrjNvMxtqBn8rai0BVdLREfFdScfTzal9RHy+jnzNbOjScA2owO35/xk1pW9mw4w0+KdHqSWgRsTv8/+n1ZG+mQ1Hw7SGKun39NKL77H8ZtZfasAEfnWd8n+/pnTNbJgazqf8HsNvZhUbpqf8LZImAt8CNgHGtLZHxIZ15mtmQ08TevnrrkP/Cvg5sBh4L+nWJ2fWnKeZDUHqx79OqTugrhARlwGKiPsjYgqwZ815mtkQJI0svXRK3bNNvaDUknx3vmXrg6TbspqZ9ZNP+f8dWBH4PPB24GPAx2vO08yGoCac8tddQw3gDGA9YHTedhKwRc35mtmQM0wvmyo4CzgKuBV4uea8zGwIa0Ivf90B9R8RMb3mPMxsGFAn5+Urqe6Aeoykk4HL6Hob6d/UnK+ZDTHq6NTR5dQdUD8BvJnUfto65Q/AAdXM+sk11HdExMY152Fmw0ATTvnr7ja7VtImNedhZsOC+rF0Rt011O2AmyXdS2pDFRAR4cumzKxfhvP0fS2715y+mQ0bg/+Uv+6b9N1fZ/pmNnyMGK7zoZqZVc8B1cysEk0YKTX4Q76ZGVB1L7+k3SXdKWmOpC93s395Sefm/ddLWr+vNB1QzawRJJVeSqQ1EjgB2IN0R5EDurnE81DgiYjYCPgv4Dt9peuAamaNIEaWXkrYBpgTEfdExIvAOcDebcfsDZyWH18A/LP6iNaDuA31TYO/waQGkiZFxLROlzeGaesAAAaOSURBVGOgPPfA2Z0uQkcMt8+5GuVjgqRJwKTCpmlt7/dawNzC+jxg27ZkXjkmIhZLWgiMAx7rKV/XUAefSX0fYkOAP+caRcS0iNi6sAzIj5cDqpkNRw8C6xTW187buj1G0ihgFWBBb4k6oJrZcHQDMFHSBpKWAz4KtM/dPB04OD/+MPDniIjeEh3EbajDltvVhgd/zh2U20Q/C1wCjAROiYjZkqYCM/LE+L8EzpA0B3icFHR7pT4CrpmZleRTfjOzijigmplVxAG1QpLWl/S3CtLZWtJPqiiTDTxJO0m6MD/eq7thjTXmvaWk9w9UftaVO6UGoYiYAczodDls2eXOjYG88++WwNbAxQOYp2WuoVZvlKSzJN0u6QJJK0p6u6QrJd0o6RJJawJIukLSdyT9VdJdkt6VtxdrOKtLulTSbEknS7pf0vhcG75d0kl53x8lrdDJFz6U5Pf3Dkmn5s/mLEm7SLpG0t2StsnLXyTdJOlaSa+5f5qkQyT9ND9+o6TrJN0q6RuSns7bd8rfhQtynme1hjhKmizpBkl/kzStsP013518+c9U4COSbpb0kYF7xwwcUOuwMfCziHgL8BTwGeB44MMR8XbgFOA/C8ePiohtgCOAY7pJ7xjS9W+bksYTr1vYNxE4Ie97Eti36hczzG0E/IB05943AwcCOwJfAr4K3AG8KyK2AiYD3+wjvR8DP46IzUlDHYu2In0HNgE2BHbI238aEe+IiM2AFYAPFJ7T5buTx6RPBs6NiC0j4tyleM22DHzKX725EXFNfnwm6Q9vM+DSXLkYCTxcOL51S+0bgfW7SW9HYB+AiPhfSU8U9t0bETf38XxbevdGxK0AkmYDl0VESLqV9F6vApwmaSLp9uij+0hve+BD+fGvge8X9v01IublvG7O6V8NvFfS0cCKwOuB2cDv83P6+u7YAHNArV77hb2LgNkRsX0Px7+Q/19C/z+PFwqPl5BqMFad4vv7cmH9ZdJndRxweUTsk+fKvKKivJaQmo7GAD8Dto6IuZKmAGO6ec7SfHesBj7lr966klrB80DgOmD11jZJoyVt2o/0rgH2z899H7BalYW1ZbIKr47/PqTE8dfxarNMn6NueDV4PiZpJdLwx74sAsaWOM5q4IBavTuBz0i6nRT8jif9IXxH0i3AzcA7+5HescD78uVY+wGPkP5orPO+C3xL0k2UqyEeARwpaRapfXZhbwdHxJPAScDfSEMkbyiRx+XAJu6U6gwPPR3kJC0PLMljj7cHfh4RW3a6XNZ/klYEnsvtsB8FDoiI9kmNrcHc7jL4rQucJ2kE8CJweIfLY0vv7cBP86VPTwKf7HB5rGKuoZqZVcRtqGZmFXFANTOriAOqmVlFHFCN1pjyksdOkfSlutJfWj3N8CTpQ3rt/dbNauGAaoOapFI3WS+KiOkR8e28+iHS+Hiz2jmgWrckfVDS9XkmpT9JmlDY/dY8y9Ldkg4vPOeoPDPSLEnH9pF+azanLjNz5X335ZmUZgL7SXpfzm+mpPPzqCEk7Z7TmAn8SyHtQyT9VNI7gb2A7+UL3d8o6fBcxlsk/XcrT7MqOKBaT64GtsszKZ0DHF3YtwWwM2myj8mS3pCHxU4EtiHNyfl2Se/uI4/2mbn+rbBvQUS8DfgT8DVgl7w+gzTaaAxpFNEHSdd3rtGeeERcS5qL9Kg8+9Lfgd/k2ZveCtwOHFr+LTHrnS/st56sDZyrNHfrcsC9hX2/i4jngOckXU4KojsC7wNuysesRAqwV/WSR/vMXJ/n1RmYWlPPbUc6Zb8mz9a1HPAX0nR690bE3QCSzgQmlXhdm0n6BrBqLuMlJZ5jVooDqvXkeOCHETFd0k7AlMK+9tEgAQj4VkT8oh95dJdOyzP5fwGXRsQBxQMlLe3w21OBD0XELZIOAXZaynTMXsOn/NaT4kxKB7ft21vSGEnjSAHpBlJN75OF9s21JP1TH3m0z8x1dTfHXAfsIGmjnO7rJL2JNLnz+pLemI87oJvnwmtnXxoLPCxpNHBQH+Uz6xcHVANYUdK8wnIkqUZ6vqQbgcfajp9FmtXoOuC4iHgoIv5ImjT5L3kC5gvoexq59pm5ft5+QET8gzQ13tl5lqa/AG+OiOdJp/gX5U6pR3vI4xzgqNy59kbg68D1pGkR7+ijfGb94rH81hF5QuYL8609zIYE11DNzCriGqqZWUVcQzUzq4gDqplZRRxQzcwq4oBqZlYRB1Qzs4r8fwsdzJjXxGhuAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 396x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "teqF5l4H6Y7s",
        "colab_type": "text"
      },
      "source": [
        "Nesse caso, estranhamente todos os elementos foram classificados como malignos, confirmando a não convergência do modelo."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "BaqxvjI8uziK",
        "colab": {},
        "outputId": "e65105dc-e222-42f1-b1e9-3de19148c64f"
      },
      "source": [
        "test_datagen = ImageDataGenerator(rescale = 1./255)\n",
        "test_generator = test_datagen.flow_from_dataframe(dataframe=test_data,\n",
        "                                                  directory=\"test\",\n",
        "                                                  x_col=\"image_name\",\n",
        "                                                  y_col=None,\n",
        "                                                  class_mode=None,\n",
        "                                                  target_size=(image_height,image_width),\n",
        "                                                  batch_size=batch_size)\n",
        "\n",
        "predictions = model.predict(test_generator,\n",
        "                            verbose=1,\n",
        "                            workers=16)\n",
        "\n",
        "\n",
        "submission = pd.DataFrame(test_data['image_name'].str.replace(r'.jpg', ''))\n",
        "submission['target'] = predictions\n",
        "\n",
        "submission.to_csv('vgg_model_submission.csv',index=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 10982 validated image filenames.\n",
            "687/687 [==============================] - 104s 151ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rORWAL02BCGL",
        "colab_type": "text"
      },
      "source": [
        "O score oficial dessa submissão foi 0.500."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "W0uO7yY5vL4m"
      },
      "source": [
        "## Tentativa de realizar o pré-processamento da base e reduzir o desbalanceamento\n",
        "\n",
        "Vimos que aumentar a complexidade do modelo não alterou a eficácia em predizer os pacientes com tumor maligno.\n",
        "\n",
        "Um palpide do porquê o modelo está enviesado é o grande desbalanceamento da base, que tentaremos corrigir utilizando pesos para as classes. Também tentaremos melhorar o pré-processamento das bases usando [augmentations](https://www.pyimagesearch.com/2019/07/08/keras-imagedatagenerator-and-data-augmentation/). \n",
        "\n",
        "Quando um dataset é pequeno ou possui alguma classe com poucas amostras, é interessante usar augmentações para utilizar os dados já existentes para gerar mais amostras. Essas novas imagens serão as imagens originais com aplicação de rotações, 'shifts', inversões, zoom, etc. \n",
        "\n",
        "A classe imageDataGenerator já possui a funcionalidade de aplicar augmentações de forma aleatória nas imagens. Dessa forma, a cada época as imagens de treinamento e validação serão um pouco diferentes das anteriores, adicionando mais exemplos ao processo de treinamento.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "EFuL75osSVkb",
        "colab": {},
        "outputId": "56002428-9085-42c8-fe38-65f0b1c15c7b"
      },
      "source": [
        "# Carregamento das imagens\n",
        "train_data = pd.read_csv('train.csv')\n",
        "test_data = pd.read_csv('test.csv')\n",
        "\n",
        "train_data['image_name'] = train_data['image_name'].astype(str)+\".jpg\"\n",
        "test_data['image_name'] = test_data['image_name'].astype(str)+\".jpg\"\n",
        "\n",
        "\n",
        "batch_size = 16       # Tamanho do pacote\n",
        "image_height = 224    # Altura da imagem\n",
        "image_width = 224     # Largura da imagem\n",
        "\n",
        "# Realiza as augmentações nas imagens\n",
        "image_datagen = ImageDataGenerator(rescale = 1./255,\n",
        "                                   validation_split = 0.3,\n",
        "                                   rotation_range=30,      # Rotaciona aleatóriamente imagens em 30 graus\n",
        "                                   zoom_range=0.15,        # Zoom aleatório\n",
        "                                   width_shift_range=0.2,  # Shift de largura aleatório\n",
        "                                   height_shift_range=0.2, # Shift de altura aleatório\n",
        "                                   horizontal_flip=True)   # Flip horizontal aleatório\n",
        "\n",
        "train_generator = image_datagen.flow_from_dataframe(dataframe=train_data,\n",
        "                                                    directory=\"train\",\n",
        "                                                    x_col=\"image_name\",\n",
        "                                                    y_col=\"benign_malignant\",\n",
        "                                                    class_mode=\"binary\",\n",
        "                                                    target_size=(image_height,image_width),\n",
        "                                                    batch_size=batch_size,\n",
        "                                                    subset=\"training\",\n",
        "                                                    color_mode=\"rgb\")\n",
        "\n",
        "validation_generator = image_datagen.flow_from_dataframe(dataframe=train_data,\n",
        "                                                         directory=\"train\",\n",
        "                                                         x_col=\"image_name\",\n",
        "                                                         y_col=\"benign_malignant\",\n",
        "                                                         class_mode=\"binary\",\n",
        "                                                         target_size=(image_height,image_width),\n",
        "                                                         batch_size=batch_size,\n",
        "                                                         subset=\"validation\",\n",
        "                                                         color_mode=\"rgb\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 23189 validated image filenames belonging to 2 classes.\n",
            "Found 9937 validated image filenames belonging to 2 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "LaGb4uD8bF6b"
      },
      "source": [
        "Para esse teste usaremos uma outra arquitetura de rede famosa, a [ResNet](https://towardsdatascience.com/an-overview-of-resnet-and-its-variants-5281e2f56035). A ela, adicionamos uma camada densa com um neurônio que dará a probabilidade em estudo."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "QAUMeuOOeMyK",
        "colab": {},
        "outputId": "fd3cfe8e-dcc1-43f9-958d-5639961a1e64"
      },
      "source": [
        "# Criação do modelo\n",
        "\n",
        "from tensorflow.keras.applications.resnet50 import ResNet50\n",
        "# load model\n",
        "\n",
        "model = Sequential()\n",
        "model.add(ResNet50(input_shape=(image_height, image_width, 3), classes=2,include_top=False, weights=None))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(units=1, activation=\"sigmoid\"))\n",
        "\n",
        "\n",
        "# summarize the model\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "resnet50 (Functional)        (None, 7, 7, 2048)        23587712  \n",
            "_________________________________________________________________\n",
            "flatten_2 (Flatten)          (None, 100352)            0         \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 1)                 100353    \n",
            "=================================================================\n",
            "Total params: 23,688,065\n",
            "Trainable params: 23,634,945\n",
            "Non-trainable params: 53,120\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "OEzJyDM5FMX2",
        "colab": {}
      },
      "source": [
        "model.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = METRICS)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "3GMQu5LHEVg8"
      },
      "source": [
        "Calculando os pesos de cada classe."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "BxkcocFMfrO6",
        "colab": {},
        "outputId": "74e2dab3-798a-4571-ec88-2cc43e62b708"
      },
      "source": [
        "from sklearn.utils import class_weight\n",
        "\n",
        "class_weights = class_weight.compute_class_weight('balanced',\n",
        "                                                 np.unique(train_generator.classes),\n",
        "                                                 train_generator.classes)\n",
        "\n",
        "class_weights = dict(zip(np.unique(train_generator.classes), class_weights))\n",
        "print(class_weights)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "{0: 0.5089324905627249, 1: 28.487714987714988}\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w_kOi6qv6Y79",
        "colab_type": "code",
        "colab": {},
        "outputId": "c292af9b-ed2a-44c8-dd6e-4557890a77e0"
      },
      "source": [
        "STEP_SIZE_TRAIN = train_generator.n//train_generator.batch_size\n",
        "STEP_SIZE_VALID = validation_generator.n//validation_generator.batch_size\n",
        "\n",
        "model.fit(train_generator,\n",
        "          steps_per_epoch=STEP_SIZE_TRAIN,\n",
        "          validation_data=validation_generator,\n",
        "          validation_steps=STEP_SIZE_VALID,\n",
        "          epochs=50,\n",
        "          callbacks=early_stopping,\n",
        "          workers = 16,\n",
        "          class_weight=class_weights)\n",
        "\n",
        "model.save_weights('class_weights_balanced_weights.h5')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "1449/1449 [==============================] - 346s 239ms/step - loss: 20.1974 - tp: 337.0000 - fp: 17373.0000 - tn: 5393.0000 - fn: 70.0000 - accuracy: 0.2473 - precision: 0.0190 - recall: 0.8280 - auc: 0.5805 - val_loss: 936.9044 - val_tp: 72.0000 - val_fp: 4882.0000 - val_tn: 4877.0000 - val_fn: 105.0000 - val_accuracy: 0.4981 - val_precision: 0.0145 - val_recall: 0.4068 - val_auc: 0.4731\n",
            "Epoch 2/50\n",
            "1449/1449 [==============================] - 365s 252ms/step - loss: 13.3197 - tp: 355.0000 - fp: 19884.0000 - tn: 2883.0000 - fn: 51.0000 - accuracy: 0.1397 - precision: 0.0175 - recall: 0.8744 - auc: 0.5081 - val_loss: 1.1988 - val_tp: 176.0000 - val_fp: 9260.0000 - val_tn: 499.0000 - val_fn: 1.0000 - val_accuracy: 0.0679 - val_precision: 0.0187 - val_recall: 0.9944 - val_auc: 0.4980\n",
            "Epoch 3/50\n",
            "1449/1449 [==============================] - 352s 243ms/step - loss: 20.5082 - tp: 346.0000 - fp: 18201.0000 - tn: 4566.0000 - fn: 60.0000 - accuracy: 0.2120 - precision: 0.0187 - recall: 0.8522 - auc: 0.5454 - val_loss: 1.4856 - val_tp: 171.0000 - val_fp: 7678.0000 - val_tn: 2081.0000 - val_fn: 6.0000 - val_accuracy: 0.2267 - val_precision: 0.0218 - val_recall: 0.9661 - val_auc: 0.5743\n",
            "Epoch 4/50\n",
            "1449/1449 [==============================] - 358s 247ms/step - loss: 15.1218 - tp: 344.0000 - fp: 15503.0000 - tn: 7263.0000 - fn: 63.0000 - accuracy: 0.3283 - precision: 0.0217 - recall: 0.8452 - auc: 0.5632 - val_loss: 4.4677 - val_tp: 123.0000 - val_fp: 5118.0000 - val_tn: 4641.0000 - val_fn: 54.0000 - val_accuracy: 0.4795 - val_precision: 0.0235 - val_recall: 0.6949 - val_auc: 0.5806\n",
            "Epoch 5/50\n",
            "1449/1449 [==============================] - 361s 249ms/step - loss: 13.8487 - tp: 353.0000 - fp: 16033.0000 - tn: 6733.0000 - fn: 54.0000 - accuracy: 0.3058 - precision: 0.0215 - recall: 0.8673 - auc: 0.5530 - val_loss: 2.6770 - val_tp: 167.0000 - val_fp: 8544.0000 - val_tn: 1215.0000 - val_fn: 10.0000 - val_accuracy: 0.1391 - val_precision: 0.0192 - val_recall: 0.9435 - val_auc: 0.5082\n",
            "Epoch 6/50\n",
            "1449/1449 [==============================] - 368s 254ms/step - loss: 20.9947 - tp: 372.0000 - fp: 16953.0000 - tn: 5814.0000 - fn: 34.0000 - accuracy: 0.2669 - precision: 0.0215 - recall: 0.9163 - auc: 0.5587 - val_loss: 145.0804 - val_tp: 143.0000 - val_fp: 3872.0000 - val_tn: 5887.0000 - val_fn: 34.0000 - val_accuracy: 0.6069 - val_precision: 0.0356 - val_recall: 0.8079 - val_auc: 0.7056\n",
            "Epoch 7/50\n",
            "1449/1449 [==============================] - 366s 253ms/step - loss: 7.6334 - tp: 396.0000 - fp: 17276.0000 - tn: 5490.0000 - fn: 11.0000 - accuracy: 0.2540 - precision: 0.0224 - recall: 0.9730 - auc: 0.5986 - val_loss: 0.8317 - val_tp: 174.0000 - val_fp: 6884.0000 - val_tn: 2875.0000 - val_fn: 3.0000 - val_accuracy: 0.3069 - val_precision: 0.0247 - val_recall: 0.9831 - val_auc: 0.6406\n",
            "Epoch 8/50\n",
            "1449/1449 [==============================] - 377s 260ms/step - loss: 11.5160 - tp: 395.0000 - fp: 17882.0000 - tn: 4884.0000 - fn: 12.0000 - accuracy: 0.2278 - precision: 0.0216 - recall: 0.9705 - auc: 0.5663 - val_loss: 104.4596 - val_tp: 164.0000 - val_fp: 7384.0000 - val_tn: 2375.0000 - val_fn: 13.0000 - val_accuracy: 0.2555 - val_precision: 0.0217 - val_recall: 0.9266 - val_auc: 0.5653\n",
            "Epoch 9/50\n",
            "1449/1449 [==============================] - ETA: 0s - loss: 3.8404 - tp: 400.0000 - fp: 17879.0000 - tn: 4887.0000 - fn: 7.0000 - accuracy: 0.2282 - precision: 0.0219 - recall: 0.9828 - auc: 0.5825Restoring model weights from the end of the best epoch.\n",
            "1449/1449 [==============================] - 360s 248ms/step - loss: 3.8404 - tp: 400.0000 - fp: 17879.0000 - tn: 4887.0000 - fn: 7.0000 - accuracy: 0.2282 - precision: 0.0219 - recall: 0.9828 - auc: 0.5825 - val_loss: 0.9993 - val_tp: 175.0000 - val_fp: 7562.0000 - val_tn: 2197.0000 - val_fn: 2.0000 - val_accuracy: 0.2387 - val_precision: 0.0226 - val_recall: 0.9887 - val_auc: 0.6071\n",
            "Epoch 00009: early stopping\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f396a365048>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sm87TH6g6Y7_",
        "colab_type": "code",
        "colab": {},
        "outputId": "2b4ba477-f2d8-4ffa-d503-72eeba9b8e86"
      },
      "source": [
        "validation_predictions = model.predict(validation_generator,\n",
        "                                       verbose=1,\n",
        "                                       workers=16)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "622/622 [==============================] - 107s 173ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Fejr_71A6Y8B",
        "colab_type": "code",
        "colab": {},
        "outputId": "efb65bd9-131f-41f0-d528-d51502a80012"
      },
      "source": [
        "val_auc = roc_auc_score(y_true = validation_generator.classes, y_score=validation_predictions)\n",
        "print(val_auc)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.5150695795128276\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QDrPkpdx6Y8D",
        "colab_type": "code",
        "colab": {},
        "outputId": "5abf4e06-e66b-4c69-80e5-ac2799a40c9d"
      },
      "source": [
        "# Plota a matriz de confusão dos dados de validação\n",
        "plot_confusion_matrix(validation_generator.classes,\n",
        "                      (validation_predictions>0.5).astype(int),\n",
        "                      ['benign','malignant'], \"Matriz de confusão\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 396x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QwTHg7wac42a",
        "colab_type": "text"
      },
      "source": [
        "Vemos que o modelo da ResNet com augmentações e peso para as classes trouxe uma mudança significativa para o resultado. Alguns exemplos de tumor maligno acabaram sendo corretamente classificados, porém houve mais falsos positivos."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "y4cVKct8cEot",
        "colab": {},
        "outputId": "8f50489e-e380-41ed-934e-f14f6126963c"
      },
      "source": [
        "test_datagen = ImageDataGenerator(rescale = 1./255)\n",
        "test_generator = test_datagen.flow_from_dataframe(dataframe=test_data,\n",
        "                                                  directory=\"test\",\n",
        "                                                  x_col=\"image_name\",\n",
        "                                                  y_col=None,\n",
        "                                                  class_mode=None,\n",
        "                                                  target_size=(image_height,image_width),\n",
        "                                                  batch_size=batch_size)\n",
        "\n",
        "predictions = model.predict(test_generator,\n",
        "                            verbose=1,\n",
        "                            workers=16)\n",
        "\n",
        "\n",
        "submission = pd.DataFrame(test_data['image_name'].str.replace(r'.jpg', ''))\n",
        "submission['target'] = predictions\n",
        "\n",
        "submission.to_csv('class_weight_balanced_submission.csv',index=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 10982 validated image filenames.\n",
            "687/687 [==============================] - 105s 153ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i733VnASBHv3",
        "colab_type": "text"
      },
      "source": [
        "O score oficial dessa submissão foi 0.489."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "T-EB831a38ux"
      },
      "source": [
        "## Usando um modelo pré-treinado como extrator de features"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "BglT86QlAbqH"
      },
      "source": [
        "Uma estratégia comum nas aplicações de redes neurais ultimamente é a transferência de aprendizado ou \"[transfer learning](https://towardsdatascience.com/a-comprehensive-hands-on-guide-to-transfer-learning-with-real-world-applications-in-deep-learning-212bf3b2f27a)\". Nessa estratégia, um modelo treinado para uma finalidade é utilizado para tentar inferir exemplos de problemas diferentes. Em outras palavras, consiste em utilizar recursos aprendidos em um problema prévio e aproveitá-lo em um novo problema semelhante.\n",
        "\n",
        "A rede treinada já entrou em contato com várias imagens de várias classes e seus pesos já foram usados em alguma tarefa e trouxeram resultados satisfatórios. As camadas de convolução dessa rede já estão especialistas em encontrar certos tipos de features das imagens.\n",
        "\n",
        "Sendo assim, podemos utilizar tais pesos para treinar um dataset que, a principio, tem poucos dados para que seja feito um treinamento do zero ou, simplesmente, com o intuito de poupar tempo/custo de processamento.\n",
        "\n",
        "Para fazer a transferência de aprendizado, usa-se apenas as camadas convolucionais da rede já treinada e adiciona a esse modelo um classificador. \n",
        "\n",
        "A sua base de dados irá passar pelas camadas convolucionais da rede treinada e na prática ocorrerá a extração de features das suas imagens. O seu classificador irá usar essas features e tentar acertar a a classe correta.\n",
        "\n",
        "A partir dos diversos tipos de redes pré-treinadas, é possível verificar o desempenho alcançado por cada rede em comparativos relacionados às suas taxas de acurácia, como ilustrado abaixo. \n",
        "\n",
        "![texto alternativo](https://drive.google.com/uc?id=10wTfPtfhae6DGINgY-zTzTYhGP3qNM1w)\n",
        "[Fonte](https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qePvWkOzAlrm",
        "colab_type": "text"
      },
      "source": [
        "### Exemplos de Transfer Learning"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TVZ_NxI-J-Ws",
        "colab_type": "code",
        "colab": {},
        "outputId": "a1dd8294-1f13-4fe9-d64d-f1f751fdf6e8"
      },
      "source": [
        "#Importando bibliotecas\n",
        "import efficientnet.keras\n",
        "from efficientnet.keras import EfficientNetB3\n",
        "from tensorflow.keras.applications import InceptionResNetV2\n",
        "from tensorflow.keras.applications import ResNet50V2"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Rcu7u5uh8ZK3",
        "colab_type": "text"
      },
      "source": [
        "#### Leitura dos dados\n",
        "\n",
        "Realizado de forma similar às execuções anteriores."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ElwikWq_J-Wz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Carregamento das imagens\n",
        "\n",
        "train_data = pd.read_csv('train.csv')\n",
        "test_data = pd.read_csv('test.csv')\n",
        "\n",
        "train_data['image_name'] = train_data['image_name'].astype(str)+\".jpg\"\n",
        "test_data['image_name'] = test_data['image_name'].astype(str)+\".jpg\"\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1e7PqxvaCLnY",
        "colab_type": "text"
      },
      "source": [
        "####Seleção de dados\n",
        "\n",
        "Como estamos trabalhando com um dataset muito extenso, foi decidido a utilização de apenas uma parte do conjunto de dados para treinamento.\n",
        "\n",
        "Tal método é utilizado como uma estratégia diferente verificação do comportamento da rede perante um valor ainda mais limitado de dados de entrada.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Cd9sdGkUJ-W3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Selecionando apenas uma parte do conjunto de treinamento\n",
        "train_splited = train_data.iloc[:3000,:]\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Xr1lAxXoFPvn",
        "colab_type": "text"
      },
      "source": [
        "Utilizamos, novamente, as ferramentas [ImageDataGenerator](https://keras.io/api/preprocessing/image/) e [flow_from_dataframe](https://keras.io/api/preprocessing/image/) para que seja feito o carregamento das imagens em forma de pacotes a partir de dataframes."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FivniUCoJ-W6",
        "colab_type": "code",
        "colab": {},
        "outputId": "3b9f99e0-2dca-4cd6-9098-5e91d9c69f19"
      },
      "source": [
        "batch_size = 12   # Tamanho do pacote\n",
        "img_height = 224  # Altura da imagem\n",
        "img_width = 224   # Largura da imagem\n",
        "\n",
        "# Gera pacotes com dados das imagens e realiza augmentation\n",
        "image_datagenerated = ImageDataGenerator(rescale = 1./255,\n",
        "                                   validation_split = 0.3)\n",
        "\n",
        "# Cria pacotes a partir do dataframe e diretório das imagens\n",
        "\n",
        "train_flow = image_datagenerated.flow_from_dataframe(dataframe=train_splited,directory=\"train\",x_col=\"image_name\",\n",
        "                                                    y_col=\"benign_malignant\",class_mode=\"binary\", target_size=(img_height,img_width),\n",
        "                                                    batch_size=batch_size, subset=\"training\", color_mode=\"rgb\")\n",
        "\n",
        "validation_flow = image_datagenerated.flow_from_dataframe(dataframe=train_splited,directory=\"train\",x_col=\"image_name\",\n",
        "                                                         y_col=\"benign_malignant\",class_mode=\"binary\",target_size=(img_height,img_width),\n",
        "                                                         batch_size=batch_size,subset=\"validation\",color_mode=\"rgb\")\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 2100 validated image filenames belonging to 2 classes.\n",
            "Found 900 validated image filenames belonging to 2 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "LrHt4jAy5RYX",
        "colab_type": "code",
        "colab": {},
        "outputId": "596fa7f0-16cd-4173-a5f3-fd780001b9c8"
      },
      "source": [
        "test_datagenerated = ImageDataGenerator(rescale = 1./255)\n",
        "\n",
        "test_flow = test_datagenerated.flow_from_dataframe(dataframe=test_data,directory=\"test\",\n",
        "                                                  x_col=\"image_name\",\n",
        "                                                  y_col=None,\n",
        "                                                  class_mode=None,\n",
        "                                                  target_size=(img_height,img_width),\n",
        "                                                  batch_size=16)\n",
        "\n",
        "\n",
        "                                    "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 10982 validated image filenames.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9bgEE43EJ-XC",
        "colab_type": "text"
      },
      "source": [
        "#### ResNet50V2"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dh82tQhXKayJ",
        "colab_type": "text"
      },
      "source": [
        "Como primeira rede pré-treinada, utilizamos uma Rede Neural Residual ([ResNet](https://keras.io/api/applications/resnet/)).\n",
        "\n",
        "\n",
        "Trata-se de uma rede neural convolucional que apresenta um bom desempenho e efetividade ao trabalhar com um grande número de camadas.\n",
        "\n",
        "Basicamente, esse modelo de rede realiza um salto ou \"by-pass\" dos valores dos pesos entre camadas, fazendo com que camadas anteriores influenciem diretamente nas camadas futuras com a atribuição de valores residuais nas mesmas, como ilustrado na figura abaixo.\n",
        "\n",
        "\n",
        "\n",
        "![Ilustração explicativa de um salto da ResNet](https://drive.google.com/uc?id=1grKtq463vM2hSZf5DFNIaGfmOJQpFwWS)\n",
        "\n",
        "Esse tipo de arquitetura foi [desenvolvido](https://arxiv.org/abs/1512.03385) como o intuito de ser uma solução para o problema do “[gradiente de fuga](https://towardsdatascience.com/the-vanishing-gradient-problem-69bf08b15484)”. Em redes neurais baseadas nos conceitos de backpropagation, fundamentada na minimização dos pesos, tornou-se recorrente que, ao se realizar repetidas multiplicacões e passagens entre camadas, os pesos tornam-se extremamente pequenos, fazendo com que o desempenho fique saturado ou até degradado com cada camada adicional.\n",
        "\n",
        "Sendo assim, a utilização desses \"atalhos\" simplifica um rede considerada muito extensa, o que permite um aprendizado mais rápido. Então, quando a rede treina novamente, todas as camadas são expandidas e as partes \"residuais\" da rede exploram cada vez mais o espaço de recurso da imagem de origem.\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "-5yeX8_F5RYf",
        "colab_type": "code",
        "colab": {},
        "outputId": "003733f0-b4a6-445c-b61e-f9f94c67e0da"
      },
      "source": [
        "#ResNet50V2\n",
        "\n",
        "tf_ResNet50V2 = Sequential([ResNet50V2(input_shape=(img_height, img_width, 3),weights='imagenet',include_top=False),GlobalAveragePooling2D(),Dense(1, activation='sigmoid')])\n",
        "\n",
        "# Descricao\n",
        "tf_ResNet50V2.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/resnet/resnet50v2_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "94674944/94668760 [==============================] - 2s 0us/step\n",
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "resnet50v2 (Model)           (None, 7, 7, 2048)        23564800  \n",
            "_________________________________________________________________\n",
            "global_average_pooling2d (Gl (None, 2048)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 1)                 2049      \n",
            "=================================================================\n",
            "Total params: 23,566,849\n",
            "Trainable params: 23,521,409\n",
            "Non-trainable params: 45,440\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BtX0B4RwJ-XK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "tf_ResNet50V2.compile(optimizer = 'adam' , loss = 'categorical_crossentropy', metrics=[\"accuracy\"])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "5PQ34kHT5RYn",
        "colab_type": "code",
        "colab": {},
        "outputId": "e6d7cb45-aa58-4fce-902c-b8b64dbf3cf2"
      },
      "source": [
        "STEP_SIZE_TRAIN = train_flow.n//train_flow.batch_size\n",
        "STEP_SIZE_VALID = validation_flow.n//validation_flow.batch_size\n",
        "\n",
        "tf_ResNet50V2_history = tf_ResNet50V2.fit(train_flow, steps_per_epoch=STEP_SIZE_TRAIN,validation_data=validation_flow,validation_steps=STEP_SIZE_VALID,\n",
        "          epochs=2,callbacks=early_stopping,workers = 16)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/2\n",
            "175/175 [==============================] - 1133s 6s/step - loss: 1.5327e-09 - accuracy: 0.7576 - val_loss: 2.2517e-09 - val_accuracy: 0.7156\n",
            "Epoch 2/2\n",
            "175/175 [==============================] - 1119s 6s/step - loss: 1.5327e-09 - accuracy: 0.7652 - val_loss: 2.1193e-09 - val_accuracy: 0.7422\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "POy730w85RY-",
        "colab_type": "code",
        "colab": {},
        "outputId": "470de0c7-d4e3-462e-8d0e-d64b6602e86b"
      },
      "source": [
        "predictions_resnet = tf_ResNet50V2.predict(test_flow,verbose=1,workers=16,steps=687)\n",
        "\n",
        "submission = pd.DataFrame(test_data['image_name'].str.replace(r'.jpg', ''))\n",
        "submission['target'] = predictions_resnet\n",
        "\n",
        "print(\"criando csv\")\n",
        "\n",
        "submission.to_csv('resnet_model_submission.csv',index=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "687/687 [==============================] - 3240s 5s/step\n",
            "criando csv\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "NWtsavhi5RZB",
        "colab_type": "code",
        "colab": {},
        "outputId": "06c20cef-f7ea-41eb-931c-32e7a128eebf"
      },
      "source": [
        "#Análise do Score do modelo\n",
        "\n",
        "# Acurácia\n",
        "plt.plot(tf_ResNet50V2_history.history['accuracy'])\n",
        "plt.plot(tf_ResNet50V2_history.history['val_accuracy'])\n",
        "plt.title('Acuracia')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n",
        "# Loss\n",
        "plt.plot(tf_ResNet50V2_history.history['loss'])\n",
        "plt.plot(tf_ResNet50V2_history.history['val_loss'])\n",
        "plt.title('Loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ahWCiCyhAq8A",
        "colab_type": "text"
      },
      "source": [
        "Ao fim, o score obtido na submissão foi de 0.509.\n",
        "\n",
        "Percebe-se que, mesmo com o menor número de amostras para treinamento, a utilização de uma rede pré-treinada tornou possível alcançar um resultado de score similar ao obtido nas tentativas anteriores, com redes que fazem treinamentos a partir do zero."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aLaq3pRJJ-XX",
        "colab_type": "text"
      },
      "source": [
        "#### EfficientNetB3"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "U9JIxiXVbhMo",
        "colab_type": "text"
      },
      "source": [
        "Como próxima tentativa, foi utilizado a [EfficientNetB3](https://keras.io/api/applications/efficientnet/#efficientnetb3-function).\n",
        "\n",
        "\n",
        "Essa rede foi [desenvolvida](https://arxiv.org/abs/1905.11946) com a premissa de otimizar redes neurais convolucionais ao redimensionar a estrutura de um rede convencional.\n",
        "\n",
        "Em suma, sua estruturação foi feita baseada em:\n",
        "\n",
        "*   Para obter melhor precisão e eficiência, é feito uma análise com intuito de  equilibrar todas as dimensões de largura, profundidade e resolução da rede durante seu dimensionamento. \n",
        "\n",
        "*   A expansão de qualquer dimensão de largura, profundidade ou resolução da rede melhora a precisão, mas o ganho de precisão, geralmente, diminui para modelos maiores.\n",
        "\n",
        "\n",
        "![texto alternativo](https://drive.google.com/uc?id=1A_39ckWp8OBa0fcdXkH9EZ-qVZ77HiT6)\n",
        "[Fonte](https://ai.googleblog.com/2019/05/efficientnet-improving-accuracy-and.html)\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "NLEmJv7cXep5",
        "colab_type": "code",
        "colab": {},
        "outputId": "a7d1a6d9-0f45-4299-ec69-8d8c75db5bfd"
      },
      "source": [
        "# EfficientNetB3\n",
        "tf_EfficientNetB3 = Sequential([EfficientNetB3(input_shape=(img_height, img_width, 3),weights='imagenet',include_top=False),GlobalAveragePooling2D(),Dense(1, activation='sigmoid')])\n",
        "# Descrição do modelo\n",
        "tf_EfficientNetB3.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://github.com/Callidior/keras-applications/releases/download/efficientnet/efficientnet-b3_weights_tf_dim_ordering_tf_kernels_autoaugment_notop.h5\n",
            "44113920/44107200 [==============================] - 1s 0us/step\n",
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "efficientnet-b3 (Model)      (None, 7, 7, 1536)        10783528  \n",
            "_________________________________________________________________\n",
            "global_average_pooling2d (Gl (None, 1536)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 1)                 1537      \n",
            "=================================================================\n",
            "Total params: 10,785,065\n",
            "Trainable params: 10,697,769\n",
            "Non-trainable params: 87,296\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "QnEr8ZsaXeqO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "tf_EfficientNetB3.compile(optimizer = 'adam' , loss = 'categorical_crossentropy', metrics=[\"accuracy\"])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "GlxEzmaJXeqX",
        "colab_type": "code",
        "colab": {},
        "outputId": "c0ba6c5f-be72-4cce-ad0a-34c0983f3635"
      },
      "source": [
        "STEP_SIZE_TRAIN = train_flow.n//train_flow.batch_size\n",
        "STEP_SIZE_VALID = validation_flow.n//validation_flow.batch_size\n",
        "\n",
        "tf_EfficientNetB3_history = tf_EfficientNetB3.fit(train_flow,\n",
        "          steps_per_epoch=STEP_SIZE_TRAIN,\n",
        "          validation_data=validation_flow,\n",
        "          validation_steps=STEP_SIZE_VALID,\n",
        "          epochs=2,\n",
        "          callbacks=early_stopping,\n",
        "          workers = 16)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/2\n",
            "131/131 [==============================] - 1148s 9s/step - loss: 1.5356e-09 - accuracy: 0.4656 - val_loss: 2.2618e-09 - val_accuracy: 0.2344\n",
            "Epoch 2/2\n",
            " 72/131 [===============>..............] - ETA: 5:35 - loss: 1.8823e-09 - accuracy: 0.4430"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "QLu_Ovb6Xeqc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "predictions_effnet = tf_EfficientNetB3.predict(test_flow,\n",
        "                            verbose=1,\n",
        "                            workers=16,steps=687)\n",
        "\n",
        "\n",
        "submission = pd.DataFrame(test_data['image_name'].str.replace(r'.jpg', ''))\n",
        "submission['target'] = predictions_effnet\n",
        "\n",
        "print(\"criando csv\")\n",
        "\n",
        "submission.to_csv('effnet_model_submission.csv',index=False)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "uWaQpFzYYeLG",
        "colab_type": "code",
        "colab": {},
        "outputId": "88f1e7ea-59bb-4a05-80f7-0851c5901364"
      },
      "source": [
        "#Análise do Score do modelo\n",
        "\n",
        "# Acurácia\n",
        "plt.plot(tf_EfficientNetB3_history.history['accuracy'])\n",
        "plt.plot(tf_EfficientNetB3_history.history['val_accuracy'])\n",
        "plt.title('Acuracia')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n",
        "# Loss\n",
        "plt.plot(tf_EfficientNetB3_history.history['loss'])\n",
        "plt.plot(tf_EfficientNetB3_history.history['val_loss'])\n",
        "plt.title('Loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nO3deZhU1bX38e/qiWaeFZkE5xFRG0QR4xAVZ1HjFKMxJmgSjSavvmreJMY3N/fqHQwaB1REYtSgcY5DJCZOIAg0ggioDCo0IJPMdNPTun/sAqqbaqiGqj5dVb/P8/BQp+pU9TrS9q/3OXudbe6OiIhIfXlRFyAiIs2TAkJERBJSQIiISEIKCBERSUgBISIiCSkgREQkIQWESDNnZm+Y2VVR1yG5x9QHIRKY2TvAEUA3d98ccTkikdMIQgQwsz7AEMCBc1P82WZm+n9NMo6+aUWCK4FJwBhg6+kcM+tlZi+Y2QozW2Vm98ee/62ZPRm3Xx8zczMriG2/Y2a/N7MJwCZgHzO72szmmNl6M1tgZtfGF2Bm55nZdDNbZ2bzzWxo3Gf9MPZ4XzP7V6yWlWb2lJl1SO9/GslVCgiR4Ergqdif081sTzPLB14FvgL6AD2AsY34zO8Bw4G2sc9YDpwNtAOuBv5gZkcBmNlA4AngFqADcALwZYLPNOA/gO7AwUAv4LeNqEkkaQVRFyASNTM7HtgbeNbdV5rZfOBywoiiO3CLu1fHdh/fiI8e4+6z4rZfi3v8rpmNI5zWmgZcA4x293/EXl+c6APdfR4wL7a5wszuAe5oRE0iSdMIQiScUhrn7itj20/HnusFfBUXDo21KH7DzM4ws0lm9o2ZrQHOBLrEXu4FzN/ZB5rZHmY21swWm9k64Mm4zxBJKY0gJKeZWUvgYiDfzL6OPd2CcJpnGdDbzAoShMRGoFXcdrcEH791iqCZtQCeJ5zKetndq8zsJcIpIwhhsm8SJf9H7HP7ufsqMzsfuD+J94k0mkYQkuvOB2qAQ4D+sT8HA+/HXlsK3GVmrc2s2MwGx943HTjBzHqbWXvg9p18nSJC8KwAqs3sDOC0uNcfA642s1PMLM/MepjZQQk+py2wAVhjZj0I1yxE0kIBIbnuKuBxd1/o7l9v+UP4rfwy4BxgP2AhUAZcAhC7VvAM8DFQSriY3SB3Xw/8DHgWWE24xvFK3OuTiV24BtYC7xKui9R3J3BUbJ/XgBd26ahFkqBGORERSUgjCBERSUgBISIiCSkgREQkIQWEiIgklFV9EF26dPE+ffpEXYaISMYoLS1d6e5dE72WVQHRp08fpk6dGnUZIiIZw8y+aug1nWISEZGEFBAiIpKQAkJERBLKqmsQiVRVVVFWVkZFRUXUpaRVcXExPXv2pLCwMOpSRCRLZH1AlJWV0bZtW/r06YOZ7fwNGcjdWbVqFWVlZfTt2zfqckQkS2T9KaaKigo6d+6cteEAYGZ07tw560dJItK0sj4ggKwOhy1y4RhFpGll/SmmZCxbV4ERfsjmGeTlGXlbHsf+Dq9te85MP5RFJLspIIAV6zdT28jbnm8LlPhQ2T5I1q9bw8vPP8vVP7ouLnAShE9e3efOOftsnn76aTp06JCegxYR2QkFBHBYj/bUulPrjjvU1jq1Tt3nYo+3Pl9L7LX4fcPf1bW11NaGi8eLl61i9KOPcNYl36/zNWtqasjPz2+wprseeZpFG43Fm9ZuDRWLC5C8euFkBusqqhj1/gJaFuXTsjD2J/a4VVEBLYvyKN7yuDCf4sI8jYJEpEEKiJgtP3ABaPjndqP99ue/p+yrL7jy7BMpLCykdes2dNurGzOmz6B0xsdcfNGFLC5bREVFBcN/fD1XXn0NtQ5HHXYAb7w9ng0bNnLFRedTcsyxlE6ZxJ7duvPQn8ZSVFy8XXCtK6/m316b06j6QnjkUxwLky2PWxXVDZit+xXl0yr2fHzYbAui+NfyKczPictcIlkppwLizr/NYvaSdSn9zEO6t+OOcw5t8PW77rqLTz75hOnTp/POO+9w1lln8cknn2ydjvrEmMfp1KkT5eXlDBgwgCsvv4SunTuTb0a3di3ZkFfDFwvm8ddnx9K//+NcfPHFTH//Ta644ortvlb+2mJm/vY0yitrKK8KfzZV1lBRGf4ur6rZ/rWqGjZVVlNeWUt5VfXW11dvrGRJ3D7llTVsqqqhsQsQFuRZnfCID6IQLAW0LMyLe7x9ENUPoDphVpBPXp5GQSLpkFMB0RwMHDiwTq/Cfffdx4svvgjAokWLmDt3Lp07d67znr59+9K/f38Ajj76aL788suEn21mtC0upG1xeprl3J3N1bVbQ2RreMQeh+e3hE0N5ZXV9YIoLqAqa1izqarOZ5VX1VBZXdvouopjAdOqqCA8LsqnVWFB3ZBJeNqt7mhnu1NyhfkUF+VRlK9TcZKbciogdvSbflNp3br11sfvvPMOb731FhMnTqRVq1aceOKJCXsZWrRosfVxfn4+5eXlTVJrfWZGcWH4gdoxTV+jpta3jXTqjHaqE4ZM/N91RkmVNawrr2LZ2ortRku1jRwF5efZdqfb6oxodjLaSXT9J34k1LIwn3yNgqQZyqmAiELbtm1Zv359wtfWrl1Lx44dadWqFZ9++imTJk1q4uqan/w8o02LAtq0SM+3prtTWVNLRWUtm2Kn1OqPhBoKojqn5KpqqaisYdn6ijphtqmyhs27MAoqKshLeN2n/minZWEIm+1OyW136q5uSLUo0ChIGk8BkWadO3dm8ODBHHbYYbRs2ZI999xz62tDhw5l5MiR9OvXjwMPPJBBgwZFWGluMDNaFOTToiCf9qTnVFxtrVNRnexop+FTcuVVNWzYXM2K9ZvrjIw2VdVQ08hhUJ6xLTgamOVWXLiz0275sdNuiScxFGhCQtYxb+xVx2aspKTE6y8YNGfOHA4++OCIKmpauXSsua6qpjbBaGdb2NQ5JVeVeKLC1tcaGDE1VlF+HsWFebFTavUCZLvTbvnbBdbWkVDcabn4UZJGQelhZqXuXpLoNY0gRDJQYX4e7Vvm0b5l+kZBm6sThE2Dp91qtrt2FD8SWrWxMm7GXHiuqqbxv5xuFzLJjHa2BlFBgiCqG1Kall2XAkJEtpO3ZXpyUT6dWhel5WtU1dRunUJd3qjRTt1p2Zsqt5+WvWX/VEzLblUUPxIK07K3nJJLOFGhfu9QfJhl2LTstAaEmQ0F7iW0no1y97sa2G8AMAm4xN2fiz33c+CHgAMzgavdXbcrFckShfl5FObnNfm07LrhE8Jmu1NyTTgte9uU6u1DJuFoJ8Gpu9YtCtizXXHK/xumLSDMLB94ADgVKAOmmNkr7j47wX53A2/GPdcD+BlwiLuXm9mzwKXAmHTVKyLZpSmmZVfX1FKxJYS2hkd1gyGzXe9Q3Gm5tbsxLbtz6yJKf31qyo8vnSOIgcA8d18AYGZjgfOA2fX2uwF4HhiQoLaWZlYFtAKWpLFWEZFGK8jPo01+XiTTsuNHO+mSzoDoASyK2y4DjonfITZSGAacTFxAuPtiM/tvYCFQDoxz93GJvoiZDQeGA/Tu3TuV9YuIRKoppmXvSDov2Se6ElN/sDQCuNXd60SgmXUkjDb6At2B1ma2/c2HAHd/xN1L3L2ka9euKSg7tdasWcODDz64S+8dMWIEmzZtSnFFIiLJSWdAlAG94rZ7sv1pohJgrJl9CVwEPGhm5wPfBr5w9xXuXgW8AByXxlrTRgEhIpkqnaeYpgD7m1lfYDHhIvPl8Tu4+9a71pnZGOBVd3/JzI4BBplZK8IpplOAuh1wGeK2225j/vz59O/fn1NPPZU99tiDZ599ls2bNzNs2DDuvPNONm7cyMUXX0xZWRk1NTX8+te/ZtmyZSxZsoSTTjqJLl268Pbbb0d9KCKSY9IWEO5ebWbXE2Yn5QOj3X2WmV0Xe33kDt77oZk9B0wDqoGPgEd2u6g3boOvZ+72x9TR7XA4I+HsXaDu7b7HjRvHc889x+TJk3F3zj33XN577z1WrFhB9+7dee2114Bwj6b27dtzzz338Pbbb9OlS5fU1iwikoS09kG4++vA6/WeSxgM7v79ett3AHekrbgIjBs3jnHjxnHkkUcCsGHDBubOncuQIUO4+eabufXWWzn77LMZMmRIxJWKiORaJ/UOftNvCu7O7bffzrXXXrvda6Wlpbz++uvcfvvtnHbaafzmN7+JoEIRkW1045E0i7/d9+mnn87o0aPZsGEDAIsXL2b58uUsWbKEVq1accUVV3DzzTczbdq07d4rItLUcmsEEYH4232fccYZXH755Rx77LEAtGnThieffJJ58+Zxyy23kJeXR2FhIQ899BAAw4cP54wzzmCvvfbSRWoRaXK63XcWyaVjFZHU2NHtvnWKSUREElJAiIhIQjkRENl0Gq0huXCMItK0sj4giouLWbVqVVb/AHV3Vq1aRXFx6u8HLyK5K+tnMfXs2ZOysjJWrFgRdSlpVVxcTM+ePaMuQ0SySNYHRGFhIX379t35jiIiUkfWn2ISEZFdo4AQEZGEFBAiIpKQAkJERBJSQIiISEIKCBERSUgBISIiCSkgREQkIQWEiIgkpIAQEZGEFBAiIpKQAkJERBJSQIiISEIKCBERSUgBISIiCSkgREQkIQWEiIgkpIAQEZGEFBAiIpKQAkJERBJSQIiISEIKCBERSUgBISIiCSkgREQyWeUmWFyalo8uSMuniohIepWvhsmPwocjw/bPZ0Fhy5R+CQWEiEgmWbcUJt4PpWOgcgPsfzoc//OUhwMoIEREMsOq+TBhBMwYC7XVcNiFMPgm6HZY2r6kAkJEpDlbOgPevwdmvwz5RXDkFXDcz6BT37R/6bQGhJkNBe4F8oFR7n5XA/sNACYBl7j7c7HnOgCjgMMAB37g7hPTWa+ISLPgDl+Oh/F/gPn/hKK2MPhGGPQTaLtnk5WRtoAws3zgAeBUoAyYYmavuPvsBPvdDbxZ7yPuBf7u7heZWRHQKl21iog0C7W18PkbIRjKpkDrrnDKb6DkGmjZocnLSecIYiAwz90XAJjZWOA8YHa9/W4AngcGbHnCzNoBJwDfB3D3SqAyjbWKiESnpgo+eR7Gj4AVc6BDbzjzv8PppDRcfE5WOgOiB7AobrsMOCZ+BzPrAQwDTiYuIIB9gBXA42Z2BFAK3OjuG+t/ETMbDgwH6N27dyrrFxFJr8pN8NGT8MEfYe1C2OMQuOBROPQCyI/+EnE6G+UswXNeb3sEcKu719R7vgA4CnjI3Y8ENgK3Jfoi7v6Iu5e4e0nXrl13t2YRkfQrXw3v/ReMOBzeuAXa7QWXPQPXTYB+FzeLcID0jiDKgF5x2z2BJfX2KQHGmhlAF+BMM6smXLAuc/cPY/s9RwMBISKSMdZ/DRMfgKmPQ+V62O9UGPIL6H0sWKLfqaOVzoCYAuxvZn2BxcClwOXxO7j71nlaZjYGeNXdX4ptLzKzA939M+AUtr92ISKSGVbNhw/ug+lPhx6GQ4eFHoa9+kVd2Q6lLSDcvdrMrifMTsoHRrv7LDO7Lvb6yJ18xA3AU7EZTAuAq9NVq4hIWiz9OMxImv0S5BVA/+/CcTdA532jriwp5l7/skDmKikp8alTp0ZdhojkMnf46gMYfw/MewuK2kDJD+DYn0LbblFXtx0zK3X3kkSvNY8rISIima62Fua+GUYMiz6EVl3g5F/BgB9Cy45RV7dLFBAiIrujpjr0MEwYActnQ/vecMZ/hR6Goszu71VAiIjsiqryWA/DfbBmIXQ9CIY9HG6il18YdXUpoYAQEWmM8jUwZRRMegg2rYSeA2Do3XDAUMjLrjXYFBAiIslYvwwmPQhTR8PmdbDvKaGHYe/BzbKHIRUUECIiO/LNF+E00kdPQW0VHHJeWKBnryOiriztFBAiIol8/UmYkTTrhdDDcMRl4ZbbGdLDkAoKCBGReF9NDD0Mc8eFHoZjfwqDfhrul5RjFBAiIu4hEN6/BxZNglad4aRfwcDM7WFIBQWEiOSummqY9WI4lbR8FrTvBWf8Z6yHoXXU1UVOASEiuaeqAqY/CRPugzVfQZcD4fyRcPhFWdPDkAoKCBHJHRVrYcpjoYdh43LocTSc/u9w4JlZ18OQCkkFhJk9D4wG3nD32vSWJCKSYhuWh1CYMirWw3BymKraZ0jW9jCkQrIjiIcIt9u+z8z+Coxx90/TV5aISAqs/jIs5/nRk1C9GQ45NwRD9yOjriwjJBUQ7v4W8JaZtQcuA/5hZouAR4En3b0qjTWKiDTOslkwfkS4iZ7lwRGXhgV6uuwXdWUZJelrEGbWGbgC+B7wEfAUcDxwFXBiOooTEWmUhZPCjKTP/w6FrWHQj0MfQ7vuUVeWkZK9BvECcBDwZ+Acd18ae+kZM9MKPSISHfewMM/798DCD6BlJzjxlzDwR9CqU9TVZbRkRxD3u/u/Er3Q0EpEIiJpVVMdlvIcPwKWzYR2PWDoXXDUlephSJFkA+JgM5vm7msAzKwjcJm7P5i+0kREEqiqgBlPw4R7w0XoLgfAeQ/C4d+BgqKoq8sqyQbEj9z9gS0b7r7azH4EKCBEpGlUrAu32p70IGxYBt2PgtP+DQ48Sz0MaZJsQOSZmbm7A5hZPqCoFpH027ACPnwIJo+CzWthnxPhgkeg77fUw5BmyQbEm8CzZjYScOA64O9pq0pEZPVXsR6GP4cehoPPgeNvCt3P0iSSDYhbgWuBHwMGjANGpasoEclhy+eEqaozn4v1MFwCx90IXQ+IurKck2yjXC2hm/qh9JYjIjlr0eQQDJ+9DoWt4Jjr4NifQPueUVeWs5Ltg9gf+A/gEKB4y/Puvk+a6hKRXOAO8/4ZguGr8WHthW/dBsdcqx6GZiDZU0yPA3cAfwBOItyXSVeHRGTX1NbA7JdDMHz9MbTtHu6qetRV0KJN1NVJTLIB0dLd/xmbyfQV8Fsze58QGiIiyaneDDP+EnoYvlkAnfeDc++Hfpeoh6EZSjYgKswsD5hrZtcDi4E90leWiGSVzeth6uMw8QHY8DXs1R8ufgIOOhvy8qOuThqQbEDcBLQCfgb8jnCa6ap0FSUiWWLjSvhwJEx+JCzW0/cEGDYy9DKoh6HZ22lAxJriLnb3W4ANhOsPIiINW7Mo9DBMewKqy8NI4fhfQE/1MGSSnQaEu9eY2dHxndQiIgkt/xQmjICZfw3b/S6BwTdC1wOjrUt2SbKnmD4CXo6tJrdxy5Pu/kJaqhKRzFI2NcxI+vTV0MMw4Edw3PXqYchwyQZEJ2AVcHLccw4oIERylTvM/1cIhi/fh+IO8K1bYeC10Lpz1NVJCiTbSa3rDiIS1NbAnFdCMCydAW33gtN+D0dfBS3aRl2dpFCyndSPE0YMdbj7D1JekYg0T9WbYcbYWA/DfOi0L5xzX1jvuaBF1NVJGiR7iunVuMfFwDBgSerLEZFmZ/MGKB0DE++H9UuhWz/4zhg4+Fz1MGS5ZE8xPR+/bWZ/Ad5KS0Ui0jxsXAWTH4YPH4aKNdBnCJz3AOx7snoYckSyI4j69gd672wnMxsK3AvkA6Pc/a4G9hsATAIucffn4p7PB6YCi9397F2sVUQaY20ZfHA/TPsTVG0KK7Yd/3PoNSDqyqSJJXsNYj11r0F8TVgjYkfvyQceAE4FyoApZvaKu89OsN/dhEWJ6rsRmAO0S6ZOEdkNKz4PPQwfPxO2D/8ODL4J9jgo2rokMsmeYtqVqQkDgXnuvgDAzMYC5wGz6+13A/A8UOfXEzPrCZwF/B74xS58fRFJxuJSeP8e+PQ1KCiGkmtCD0OHnZ4kkCyX7AhiGPAvd18b2+4AnOjuL+3gbT2ARXHbZcAx9T63B+GC98nUCwhgBPB/gR2Gk5kNB4YD9O6tb2iRpLjDgnfCVNUv3oXi9nDCzWGRntZdoq5Omolkr0Hc4e4vbtlw9zVmdgewo4BIdBWr/lTZEcCtsdt5bHuj2dnAcncvNbMTd1SYuz8CPAJQUlKiW4GI7EhtLXz6txAMSz6CNt3g1N/B0d+HYp3JlbqSDYi8XXhvGdArbrsn20+NLQHGxsKhC3CmmVUTRhrnmtmZhGm17czsSXe/Isl6RSRedWW4tjDhXlg1FzrtA+fcC0dcph4GaVCyATHVzO4hXHR2wnWD0p28Zwqwv5n1JawfcSlwefwO7t53y2MzGwO8Gjtt9RJwe+z5E4GbFQ4iu2DzhjAb6YP7Yf0S6HY4XPQ4HHKeehhkp5INiBuAXwOx6Q2MA361oze4e3VscaE3CdNcR7v7LDO7Lvb6yF0rWUR2atM3YQ2GD0dC+WrY+3g494+w3ynqYZCkWTbdwbukpMSnTp0adRki0Vm7OKzaVjoGqjbCgWfGehgGRl2ZNFNmVuruJYleS3YW0z+A77j7mth2R2Csu5+eujJFZJetnBt6GGY8A14b62G4EfY8JOrKJIMle4qpy5ZwAHD31WamNalForbko9DDMOdv4WLz0d+H426AjntHXZlkgWQDotbMerv7QgAz60OCu7uKSBNwhy/eg/H3hF6GFu1hyC/gmB9Dm65RVydZJNmA+H/AeDN7N7Z9ArHmNBFpIrW18NlroYdhcSm02RO+fSeU/EA9DJIWyd5q4+9mVkIIhenAy0B5OgsTkZjqyrDG84QRsPJz6NgHzv4DHHE5FBZHXZ1ksWQvUv+QcOO8noSAGARMpO4SpCKSSpUbYdoToYdhXRnseThc+Bgccj7k7+qNmEWSl+x32Y2EeyVNcveTzOwg4M70lSWSwzZ9A5MfjfUwfAO9j4NzRsB+31YPgzSpZAOiwt0rzAwza+Hun5rZgWmtTCTXrFuyrYehcgMcMDT0MPQeFHVlkqOSDYiy2B1cXwL+YWar0ZKjIqmxch58cC9M/0voYTjsQjj+Jtjz0KgrkxyX7EXqYbGHvzWzt4H2wN/TVpVILlgyPcxImv0y5BfB0VfFehj6RF2ZCLALS466+7s730tEEnKHL8eHHob5/4IW7cJppEE/hjbqPZXmRVMhRJpCbS18/kboel48FVp3hVPugAHXhMV6RJohBYRIOtVUwcznQg/Dik+hw95w1v9A/+9CYcuoqxPZIQWESDpUboKP/gwf/BHWLoI9DoULRsGhw9TDIBlD36kiqVS+GiaPgg8fgk2roNegMGLY/zT1MEjGUUCIpMK6pTDpAZj6eOhh2P/0cPF572OjrkxklykgRHbHqvlhnecZf4Haajj0ghAM3Q6LujKR3aaAENkVS2fA+BEw+yXIK4Qjr4Djfgad+u78vSIZQgEhkix3+GpCaG6b9xYUtQ2hMOgn0HbPqKsTSTkFhMjO1NbC3DdDD0PZ5FgPw2+g5Bpo2SHq6kTSRgEh0pCaKvjkhTBiWDEHOvSGM/87nE5SD4PkAAWESH1V5fDRkzDhPli7ELoeDMMegcMugPzCqKsTaTIKCJEtytfAlFEw6SHYtBJ6DoQz/zNMWc3Li7o6kSangBBZvyz0MEwZDZXrYb9TYz0Mx6m5TXKaAkJy1zcLwmmk6U9DbVW4Dcbgm2CvflFXJtIsKCAk93w9M1x4nvUi5BVA/8vDdNXO+0ZdmUizooCQ3PHVByEY5o6DojZw7PVw7E+hbbeoKxNplhQQkt3c4fM3QzAsmgStOsPJv4IBP4SWHaOuTqRZU0BIdqqphlkvhNthLJ8F7XvBGf8VehiKWkVdnUhGUEBIdtnSw/DBfbBmIXQ9CIY9DIddqB4GkUZSQEh2qFgLUx4LPQwbl0PPATD0bjhgqHoYRHaRAkIy24blMOnBEA6b18G+p8CQX8Deg9XDILKbFBCSmVZ/GXoYPnoSairh0PNDD0P3/lFXJpI1FBCSWZbNCjOSPnkB8vLhiMtg8I3qYRBJAwWEZIaFk8Lttue+Geth+AkM+im02yvqykSylgJCmi93mPsPGH8PLJwYehhO+hUMuAZadYq6OpGsp4CQ5qemOizlOf4PsOwTaNczzEg66ntQ1Drq6kRyRlrn/5nZUDP7zMzmmdltO9hvgJnVmNlFse1eZva2mc0xs1lmdmM665RmoqoizEa6/2h4/pqwYM/5D8GN02HQdQoHkSaWthGEmeUDDwCnAmXAFDN7xd1nJ9jvbuDNuKergf/j7tPMrC1Qamb/qP9eyRIV62DqYzDxwdDD0ONoOO33cOCZ6mEQiVA6TzENBOa5+wIAMxsLnAfU/yF/A/A8MGDLE+6+FFgae7zezOYAPRK8VzLZhhXw4UMweRRsXgv7nARDHoM+Q9TDINIMpDMgegCL4rbLgGPidzCzHsAw4GTiAqLePn2AI4EPG3h9ODAcoHfv3rtZsjSJ1V/BB3+Ej/4M1ZvhkHPDAj3dj4y6MhGJk86ASPQroNfbHgHc6u41luA3RjNrQxhd3OTu6xJ9EXd/BHgEoKSkpP7nS3OybDZMGAEznwPLgyMuDT0MXfaPujIRSSCdAVEG9Irb7gksqbdPCTA2Fg5dgDPNrNrdXzKzQkI4POXuL6SxTkm3RZNDD8Pnb0Bhaxj047AOQ7vuUVcmIjuQzoCYAuxvZn2BxcClwOXxO7h73y2PzWwM8GosHAx4DJjj7veksUZJF3eY98/Qw/DVBGjZCU78JQz8kXoYRDJE2gLC3avN7HrC7KR8YLS7zzKz62Kvj9zB2wcD3wNmmtn02HO/dPfX01WvpEhtzbYehq9nQrseMPQuOOpKTVMVyTDmnj2n7UtKSnzq1KlRl5GbqjfD9Kdhwr2w+gvockC4ed7h34GCoqirE5EGmFmpu5ckek2d1LJ7Nq+HqaNDD8OGr8NMpFP/DAedrR4GkQyngJBds3FlWJxnyqNhsZ59ToQLHoa+31IPg0iWUEBI46xZGHoYpv0Zqivg4HPg+JtC97OIZBUFhCRn+ZxwfWHmX8N2v1gPQ9cDoq1LRNJGASE7tmhKmJH02WtQ2AoGDg89DO17Rl2ZiKSZAkK25w7z/xWC4cv3obgDfOs2OOZa9TCI5BAFhGxTWwNzXgnBsHQGtO0Op/87HHUVtGgTdXUi0sQUEBJ6GGaMDdcYvpkPnfeDc++HfhdDQYuoqxORiCggctnm9VA6BiY+AOuXwl794eInYj0M+VFXJyIRU0DkotulXfUAAAmiSURBVI2r4MORMPkRqFgDfU8IK7ftc6J6GERkKwVELlmzCCbeD6V/guryMFI4/hfQUz0MIrI9BUQuWPEZjB8BM58N2/0uifUwHBhtXSLSrCkgsllZabjd9qevQUExDPghHHs9dOi18/eKSM5TQGQbd1jwdpiq+sV7oYfhhFvgmOugdeeoqxORDKKAyBa1NTDnb7EehunQdi847d/g6O9Di7ZRVyciGUgBkemqK+HjZ8Jaz6vmQad94Zz7wnrP6mEQkd2ggMhUmzfAtD/BB/fD+iXQrR98ZwwcfK56GEQkJRQQmWbTN/DhwzD5YShfDX2GwHn3w74nq4dBRFJKAZEp1i6O9TCMgapNcOBZcPzPodeAqCsTkSylgGjuVs4NPQwfPwNeG+6PNPgm2OOgqCsTkSyngGiuFk8LM5Lm/C30MJT8AI67Hjr0jroyEckRCojmxB2+eBfevyf8XdweTrg51sPQJerqRCTHKCCag9pa+PTVMGJYMg3adINTfxd6GIrbRV2diOQoBUSUqivD/ZHGj4BVc6FjXzh7BBxxGRQWR12diOQ4BUQUKjfCtCdCD8O6Muh2OFw0Gg45Xz0MItJsKCCa0qZvYPKjYS2G8m9g78Fwzr2w3ynqYRCRZkcB0RTWLQmrtk19HKo2wgFnhB6G3sdEXZmISIMUEOm0cl64R9KMsaGH4fCLQg/DnodEXZmIyE4pINJhyUdhRtLsV8IN847+Phx3A3TcO+rKRESSpoBIFXf48v3Qw7DgbWjRHob8Ao75MbTpGnV1IiKNpoDYXbW18NnrYeW2xaXQeg/49p2h81k9DCKSwRQQu6qmCmb+NfQwrPwMOvaBs/8AR1yuHgYRyQoKiMaq3BTrYfhj6GHY8zC48LHQw5Cv/5wikj30Ey1Z5ath8ij48CHYtAp6HxdGDPufqh4GEclKCoidWbcUJsV6GCo3wAFDYz0Mg6KuTEQkrRQQDVk1HybcCzP+ArU1cNiFMPhG6HZY1JWJiDQJBUR9S2fEehhehrxCOOpKOPZ66NQ36spERJpUXjo/3MyGmtlnZjbPzG7bwX4DzKzGzC5q7HtTwh2+HA9/vgAePgHm/TOMFn7+CZz1PwoHEclJaRtBmFk+8ABwKlAGTDGzV9x9doL97gbebOx7U6JiHTx5AZRNgdZd4ZQ7YMA1YbEeEZEcls5TTAOBee6+AMDMxgLnAfV/yN8APA8M2IX37r7idmEdhiMuhf7fhcKWKf8SIiKZKJ0B0QNYFLddBtS5famZ9QCGASdTNyB2+t64zxgODAfo3XsX12u+8NFde5+ISBZL5zWIRM0BXm97BHCru9fswnvDk+6PuHuJu5d07ap7HomIpEo6RxBlQK+47Z7Aknr7lABjLTSadQHONLPqJN8rIiJplM6AmALsb2Z9gcXApcDl8Tu4+9bpQWY2BnjV3V8ys4KdvVdERNIrbQHh7tVmdj1hdlI+MNrdZ5nZdbHXRzb2vemqVUREtmfuCU/tZ6SSkhKfOnVq1GWIiGQMMyt195JEr6W1UU5ERDKXAkJERBJSQIiISEJZdQ3CzFYAX+3i27sAK1NYTibQMWe/XDte0DE31t7unrCJLKsCYneY2dSGLtRkKx1z9su14wUdcyrpFJOIiCSkgBARkYQUENs8EnUBEdAxZ79cO17QMaeMrkGIiEhCGkGIiEhCCggREUkopwJiZ+tcW3Bf7PWPzeyoKOpMpSSO+buxY/3YzD4wsyOiqDOVdmct9EyVzDGb2YlmNt3MZpnZu01dY6ol8b3d3sz+ZmYzYsd8dRR1poqZjTaz5Wb2SQOvp/7nl7vnxB/CXWHnA/sARcAM4JB6+5wJvEFYsGgQ8GHUdTfBMR8HdIw9PiMXjjluv38BrwMXRV13E/w7dyAs2ds7tr1H1HU3wTH/Erg79rgr8A1QFHXtu3HMJwBHAZ808HrKf37l0ghi6zrX7l4JbFnnOt55wBMeTAI6mNleTV1oCu30mN39A3dfHducRFicKZMl8+8M29ZCX96UxaVJMsd8OfCCuy8EcPdMP+5kjtmBthZWJGtDCIjqpi0zddz9PcIxNCTlP79yKSASrXPdYxf2ySSNPZ5rCL+BZLKdHnPcWugNrkmSYZL5dz4A6Ghm75hZqZld2WTVpUcyx3w/cDBhNcqZwI3uXts05UUi5T+/0rmiXHOTzDrXSa+FnSGSPh4zO4kQEMentaL0a9Ra6LHlbjNdMsdcABwNnAK0BCaa2SR3/zzdxaVJMsd8OjAdOBnYF/iHmb3v7uvSXVxEUv7zK5cCIpl1rrNtLeykjsfM+gGjgDPcfVUT1ZYuu7wWuru/1DQlplyy39sr3X0jsNHM3gOOADI1IJI55quBuzycoJ9nZl8ABwGTm6bEJpfyn1+5dIpp6xrZZlZEWOf6lXr7vAJcGZsNMAhY6+5Lm7rQFNrpMZtZb+AF4HsZ/NtkvJ0es7v3dfc+7t4HeA74SQaHAyT3vf0yMMTMCsysFXAMMKeJ60ylZI55IWHEhJntCRwILGjSKptWyn9+5cwIwpNbI/t1wkyAecAmwm8gGSvJY/4N0Bl4MPYbdbVn8J0wkzzmrJLMMbv7HDP7O/AxUAuMcveE0yUzQZL/zr8DxpjZTMLpl1vdPWNvA25mfwFOBLqYWRlwB1AI6fv5pVttiIhIQrl0iklERBpBASEiIgkpIEREJCEFhIiIJKSAEBGRhBQQIs1A7E6rr0Zdh0g8BYSIiCSkgBBpBDO7wswmx9ZVeNjM8s1sg5n9j5lNM7N/mlnX2L79zWxS7N78L5pZx9jz+5nZW7F1CqaZ2b6xj29jZs+Z2adm9pRlyY2iJHMpIESSZGYHA5cAg929P1ADfBdoDUxz96OAdwkdrgBPELp3+xHuJrrl+aeAB9z9CMJ6HFtuh3AkcBNwCGGdg8FpPyiRHciZW22IpMAphDuiTon9ct+SsJ5ELfBMbJ8ngRfMrD3Qwd23rNz2J+CvZtYW6OHuLwK4ewVA7PMmu3tZbHs60AcYn/7DEklMASGSPAP+5O6313nS7Nf19tvR/Wt2dNpoc9zjGvT/p0RMp5hEkvdP4CIz2wPAzDqZ2d6E/4+2rGt9OTDe3dcCq81sSOz57wHvxtYiKDOz82Of0SJ2d1WRZke/oYgkyd1nm9mvgHFmlgdUAT8FNgKHmlkpsJZwnQLgKmBkLAAWsO3umt8DHjaz/x/7jO804WGIJE13cxXZTWa2wd3bRF2HSKrpFJOIiCSkEYSIiCSkEYSIiCSkgBARkYQUECIikpACQkREElJAiIhIQv8Lm6ZsQRwxEgEAAAAASUVORK5CYII=\n"
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "N6r1CHk5YqbO",
        "colab_type": "text"
      },
      "source": [
        "Ao fim, obteve-se um score de 0.482 na submissão.\n",
        "\n",
        "Com esse resultados, percebemos que, novamente, uma rede com menor número de amostras consegue alcançar resultados similares à redes que utilizariam quantidades muito maiores de amostras. Mas, devido a sua estrutra e, possivelmente,  quais dados utilizados para realizar esse treinamento prévio, os resultados podem variar nos primeiros testes. Deve-se levar em consideração a pequena quantidade de épocas utilizada."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PzTVLqkVBaJr",
        "colab_type": "text"
      },
      "source": [
        "#### EfficientNetB0"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8jTheVuWdtPJ",
        "colab_type": "text"
      },
      "source": [
        "Mais um teste foi feito usando a técnica de transfer learning, dessa vez aplicando o treinamento a todas as imagens e usando como modelo a [EfficienteNetB0](https://keras.io/api/applications/efficientnet/)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "xopTkqJL4ptP",
        "colab": {}
      },
      "source": [
        "# Leitura dos dataframes dos arquivos csv\n",
        "train_data = pd.read_csv('train.csv')\n",
        "test_data = pd.read_csv('test.csv')\n",
        "\n",
        "# Adicionando a extensão aos nomes das imagens\n",
        "train_data['image_name'] = train_data['image_name'].astype(str)+\".jpg\"\n",
        "test_data['image_name'] = test_data['image_name'].astype(str)+\".jpg\"\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "4-WhnjVU6Oye",
        "colab": {},
        "outputId": "3bd353cb-ef13-4661-8867-0528673bbd05"
      },
      "source": [
        "# Carregamento das imagens\n",
        "train_data = pd.read_csv('train.csv')\n",
        "test_data = pd.read_csv('test.csv')\n",
        "\n",
        "train_data['image_name'] = train_data['image_name'].astype(str)+\".jpg\"\n",
        "test_data['image_name'] = test_data['image_name'].astype(str)+\".jpg\"\n",
        "\n",
        "\n",
        "batch_size = 32       # Tamanho do pacote\n",
        "image_height = 224    # Altura da imagem\n",
        "image_width = 224     # Largura da imagem\n",
        "\n",
        "# Realiza as augmentações nas imagens\n",
        "image_datagen = ImageDataGenerator(rescale = 1./255,\n",
        "                                   rotation_range=30,      \n",
        "                                   zoom_range=0.15,        \n",
        "                                   width_shift_range=0.2,  \n",
        "                                   height_shift_range=0.2, \n",
        "                                   horizontal_flip=True)   \n",
        "\n",
        "# Nesse treinamento, todas as imagens serão usadas como dataset de treinamento\n",
        "train_generator = image_datagen.flow_from_dataframe(dataframe=train_data,\n",
        "                                                    directory=\"train\",\n",
        "                                                    x_col=\"image_name\",\n",
        "                                                    y_col=\"benign_malignant\",\n",
        "                                                    class_mode=\"binary\",\n",
        "                                                    target_size=(image_height,image_width),\n",
        "                                                    batch_size=batch_size, subset=\"training\",\n",
        "                                                    color_mode=\"rgb\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 33126 validated image filenames belonging to 2 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8qWlYNsv6Y8Z",
        "colab_type": "code",
        "colab": {},
        "outputId": "90cd4ac4-59c9-443e-851c-36b6a5fdcda0"
      },
      "source": [
        "import efficientnet.tfkeras as efn\n",
        "# load model\n",
        "\n",
        "model = Sequential([\n",
        "        efn.EfficientNetB0(\n",
        "            input_shape=(image_height, image_width, 3),\n",
        "            weights='imagenet',\n",
        "            include_top=False\n",
        "        ),\n",
        "        GlobalAveragePooling2D(),\n",
        "        Dense(1, activation='sigmoid')\n",
        "    ])\n",
        "\n",
        "\n",
        "\n",
        "# summarize the model\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "efficientnet-b0 (Functional) (None, 7, 7, 1280)        4049564   \n",
            "_________________________________________________________________\n",
            "global_average_pooling2d_2 ( (None, 1280)              0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 1)                 1281      \n",
            "=================================================================\n",
            "Total params: 4,050,845\n",
            "Trainable params: 4,008,829\n",
            "Non-trainable params: 42,016\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GJXVc4OD6Y8c",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = METRICS)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aLXMeJtE6Y8e",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "early_stopping = EarlyStopping(\n",
        "    monitor='auc', \n",
        "    verbose=1,\n",
        "    patience=3,\n",
        "    mode='max',\n",
        "    restore_best_weights=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2xTbiBiE6Y8g",
        "colab_type": "code",
        "colab": {},
        "outputId": "f0cf1305-744c-4444-8ed3-224d11dd38fa"
      },
      "source": [
        "STEP_SIZE_TRAIN = train_generator.n//train_generator.batch_size\n",
        "\n",
        "model.fit(train_generator,\n",
        "          steps_per_epoch=STEP_SIZE_TRAIN,\n",
        "          epochs=50,\n",
        "          callbacks=early_stopping,\n",
        "          workers = 16)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "1035/1035 [==============================] - 350s 339ms/step - loss: 0.0829 - tp: 1.0000 - fp: 28.0000 - tn: 32483.0000 - fn: 582.0000 - accuracy: 0.9816 - precision: 0.0345 - recall: 0.0017 - auc: 0.767416s - loss: 0.0835 - tp: 1.0000 - fp: 28.0000 - tn: 30969.0000 - fn:\n",
            "Epoch 2/50\n",
            "1035/1035 [==============================] - 355s 343ms/step - loss: 0.0780 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 32511.0000 - fn: 583.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.7996\n",
            "Epoch 3/50\n",
            "1035/1035 [==============================] - 352s 340ms/step - loss: 0.0755 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 32510.0000 - fn: 584.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8257\n",
            "Epoch 4/50\n",
            "1035/1035 [==============================] - 350s 338ms/step - loss: 0.0747 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 32511.0000 - fn: 583.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8315\n",
            "Epoch 5/50\n",
            "1035/1035 [==============================] - 358s 346ms/step - loss: 0.0743 - tp: 0.0000e+00 - fp: 1.0000 - tn: 32509.0000 - fn: 584.0000 - accuracy: 0.9823 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8396\n",
            "Epoch 6/50\n",
            "1035/1035 [==============================] - 345s 333ms/step - loss: 0.0733 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 32510.0000 - fn: 584.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8424\n",
            "Epoch 7/50\n",
            "1035/1035 [==============================] - 355s 343ms/step - loss: 0.0724 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 32510.0000 - fn: 584.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8500\n",
            "Epoch 8/50\n",
            "1035/1035 [==============================] - 358s 346ms/step - loss: 0.0719 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 32510.0000 - fn: 584.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8492\n",
            "Epoch 9/50\n",
            "1035/1035 [==============================] - 365s 352ms/step - loss: 0.0708 - tp: 0.0000e+00 - fp: 0.0000e+00 - tn: 32511.0000 - fn: 583.0000 - accuracy: 0.9824 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8562\n",
            "Epoch 10/50\n",
            "1035/1035 [==============================] - 353s 341ms/step - loss: 0.0693 - tp: 0.0000e+00 - fp: 2.0000 - tn: 32508.0000 - fn: 584.0000 - accuracy: 0.9823 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8703\n",
            "Epoch 11/50\n",
            "1035/1035 [==============================] - 364s 352ms/step - loss: 0.0698 - tp: 1.0000 - fp: 2.0000 - tn: 32508.0000 - fn: 583.0000 - accuracy: 0.9823 - precision: 0.3333 - recall: 0.0017 - auc: 0.8641\n",
            "Epoch 12/50\n",
            "1035/1035 [==============================] - 355s 343ms/step - loss: 0.0678 - tp: 1.0000 - fp: 0.0000e+00 - tn: 32511.0000 - fn: 582.0000 - accuracy: 0.9824 - precision: 1.0000 - recall: 0.0017 - auc: 0.8791\n",
            "Epoch 13/50\n",
            "1035/1035 [==============================] - 348s 336ms/step - loss: 0.0679 - tp: 1.0000 - fp: 3.0000 - tn: 32507.0000 - fn: 583.0000 - accuracy: 0.9823 - precision: 0.2500 - recall: 0.0017 - auc: 0.8763\n",
            "Epoch 14/50\n",
            "1035/1035 [==============================] - 364s 352ms/step - loss: 0.0675 - tp: 0.0000e+00 - fp: 4.0000 - tn: 32507.0000 - fn: 583.0000 - accuracy: 0.9823 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8769\n",
            "Epoch 15/50\n",
            "1035/1035 [==============================] - 364s 352ms/step - loss: 0.0660 - tp: 2.0000 - fp: 2.0000 - tn: 32508.0000 - fn: 582.0000 - accuracy: 0.9824 - precision: 0.5000 - recall: 0.0034 - auc: 0.8877\n",
            "Epoch 16/50\n",
            "1035/1035 [==============================] - 350s 339ms/step - loss: 0.0669 - tp: 3.0000 - fp: 3.0000 - tn: 32507.0000 - fn: 581.0000 - accuracy: 0.9824 - precision: 0.5000 - recall: 0.0051 - auc: 0.8800\n",
            "Epoch 17/50\n",
            "1035/1035 [==============================] - 362s 350ms/step - loss: 0.0663 - tp: 0.0000e+00 - fp: 5.0000 - tn: 32505.0000 - fn: 584.0000 - accuracy: 0.9822 - precision: 0.0000e+00 - recall: 0.0000e+00 - auc: 0.8848\n",
            "Epoch 18/50\n",
            "1035/1035 [==============================] - 362s 349ms/step - loss: 0.0648 - tp: 2.0000 - fp: 3.0000 - tn: 32507.0000 - fn: 582.0000 - accuracy: 0.9823 - precision: 0.4000 - recall: 0.0034 - auc: 0.8951\n",
            "Epoch 19/50\n",
            "1035/1035 [==============================] - 357s 345ms/step - loss: 0.0633 - tp: 8.0000 - fp: 6.0000 - tn: 32505.0000 - fn: 575.0000 - accuracy: 0.9824 - precision: 0.5714 - recall: 0.0137 - auc: 0.9004\n",
            "Epoch 20/50\n",
            "1035/1035 [==============================] - 367s 355ms/step - loss: 0.0641 - tp: 15.0000 - fp: 14.0000 - tn: 32496.0000 - fn: 569.0000 - accuracy: 0.9824 - precision: 0.5172 - recall: 0.0257 - auc: 0.8941\n",
            "Epoch 21/50\n",
            "1035/1035 [==============================] - 365s 353ms/step - loss: 0.0636 - tp: 14.0000 - fp: 11.0000 - tn: 32500.0000 - fn: 569.0000 - accuracy: 0.9825 - precision: 0.5600 - recall: 0.0240 - auc: 0.8994\n",
            "Epoch 22/50\n",
            "1035/1035 [==============================] - 353s 341ms/step - loss: 0.0623 - tp: 25.0000 - fp: 15.0000 - tn: 32495.0000 - fn: 559.0000 - accuracy: 0.9827 - precision: 0.6250 - recall: 0.0428 - auc: 0.9009\n",
            "Epoch 23/50\n",
            "1035/1035 [==============================] - 363s 351ms/step - loss: 0.0621 - tp: 17.0000 - fp: 15.0000 - tn: 32495.0000 - fn: 567.0000 - accuracy: 0.9824 - precision: 0.5312 - recall: 0.0291 - auc: 0.9053s - loss: 0.0622 - tp: 17.0000 - fp: 15.0000 - tn: 32399.0000 - fn: 567.0000 - accuracy: 0.9824 - precision: 0.5312 - recall: 0.0291 - auc: 0.905 - ETA: 0s - loss: 0.0622 - tp: 17.0000 - fp: 15.0000 - tn: 32431.0000 - fn: 567.0000 - accuracy: 0.9824 - precision: 0.5312 - recall: 0.0291 - auc: 0.9\n",
            "Epoch 24/50\n",
            "1035/1035 [==============================] - 371s 359ms/step - loss: 0.0604 - tp: 24.0000 - fp: 20.0000 - tn: 32491.0000 - fn: 559.0000 - accuracy: 0.9825 - precision: 0.5455 - recall: 0.0412 - auc: 0.9090\n",
            "Epoch 25/50\n",
            "1035/1035 [==============================] - 360s 348ms/step - loss: 0.0607 - tp: 40.0000 - fp: 25.0000 - tn: 32485.0000 - fn: 544.0000 - accuracy: 0.9828 - precision: 0.6154 - recall: 0.0685 - auc: 0.9046\n",
            "Epoch 26/50\n",
            "1035/1035 [==============================] - 367s 354ms/step - loss: 0.0590 - tp: 36.0000 - fp: 27.0000 - tn: 32484.0000 - fn: 547.0000 - accuracy: 0.9827 - precision: 0.5714 - recall: 0.0617 - auc: 0.9132\n",
            "Epoch 27/50\n",
            "1035/1035 [==============================] - 373s 360ms/step - loss: 0.0598 - tp: 31.0000 - fp: 27.0000 - tn: 32483.0000 - fn: 553.0000 - accuracy: 0.9825 - precision: 0.5345 - recall: 0.0531 - auc: 0.9152\n",
            "Epoch 28/50\n",
            "1035/1035 [==============================] - 351s 339ms/step - loss: 0.0581 - tp: 41.0000 - fp: 29.0000 - tn: 32481.0000 - fn: 543.0000 - accuracy: 0.9827 - precision: 0.5857 - recall: 0.0702 - auc: 0.9216\n",
            "Epoch 29/50\n",
            "1035/1035 [==============================] - 359s 347ms/step - loss: 0.0578 - tp: 38.0000 - fp: 23.0000 - tn: 32487.0000 - fn: 546.0000 - accuracy: 0.9828 - precision: 0.6230 - recall: 0.0651 - auc: 0.9209\n",
            "Epoch 30/50\n",
            "1035/1035 [==============================] - 364s 352ms/step - loss: 0.0571 - tp: 54.0000 - fp: 30.0000 - tn: 32480.0000 - fn: 530.0000 - accuracy: 0.9831 - precision: 0.6429 - recall: 0.0925 - auc: 0.9203\n",
            "Epoch 31/50\n",
            "1035/1035 [==============================] - 363s 350ms/step - loss: 0.0563 - tp: 50.0000 - fp: 32.0000 - tn: 32478.0000 - fn: 534.0000 - accuracy: 0.9829 - precision: 0.6098 - recall: 0.0856 - auc: 0.9306\n",
            "Epoch 32/50\n",
            "1035/1035 [==============================] - 371s 358ms/step - loss: 0.0559 - tp: 62.0000 - fp: 39.0000 - tn: 32471.0000 - fn: 522.0000 - accuracy: 0.9830 - precision: 0.6139 - recall: 0.1062 - auc: 0.9283\n",
            "Epoch 33/50\n",
            "1035/1035 [==============================] - 380s 367ms/step - loss: 0.0554 - tp: 70.0000 - fp: 41.0000 - tn: 32471.0000 - fn: 512.0000 - accuracy: 0.9833 - precision: 0.6306 - recall: 0.1203 - auc: 0.93053s - loss: 0.0552 - tp: 68.0000 - fp: 40.0000 - tn: 31312.0000 - fn: 490.0000 - accuracy: 0.9834 - \n",
            "Epoch 34/50\n",
            "1035/1035 [==============================] - ETA: 0s - loss: 0.0542 - tp: 99.0000 - fp: 56.0000 - tn: 32454.0000 - fn: 485.0000 - accuracy: 0.9837 - precision: 0.6387 - recall: 0.1695 - auc: 0.9304Restoring model weights from the end of the best epoch.\n",
            "1035/1035 [==============================] - 347s 335ms/step - loss: 0.0542 - tp: 99.0000 - fp: 56.0000 - tn: 32454.0000 - fn: 485.0000 - accuracy: 0.9837 - precision: 0.6387 - recall: 0.1695 - auc: 0.9304\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 00034: early stopping\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f47a5dd3358>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qgaHBNxc6Y8j",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.save_weights('tranfer_learning_weights.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LaSWq22NePVS",
        "colab_type": "text"
      },
      "source": [
        "Como vemos pelas métricas de treinamento, esse modelo obteve um AUC score máximo de 0.9306 nos dados de treinamento. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yu9IR5-H6Y8l",
        "colab_type": "code",
        "colab": {},
        "outputId": "bc362a8b-f256-46a2-bd0f-406c4aba4d4a"
      },
      "source": [
        "test_datagen = ImageDataGenerator(rescale = 1./255)\n",
        "test_generator = test_datagen.flow_from_dataframe(dataframe=test_data,\n",
        "                                                  directory=\"test\",\n",
        "                                                  x_col=\"image_name\",\n",
        "                                                  y_col=None,\n",
        "                                                  class_mode=None,\n",
        "                                                  target_size=(image_height,image_width),\n",
        "                                                  batch_size=batch_size)\n",
        "\n",
        "predictions = model.predict(test_generator,\n",
        "                            verbose=1,\n",
        "                            workers=16)\n",
        "\n",
        "\n",
        "submission = pd.DataFrame(test_data['image_name'].str.replace(r'.jpg', ''))\n",
        "submission['target'] = predictions\n",
        "\n",
        "submission.to_csv('effnet3_model_submission.csv',index=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 10982 validated image filenames.\n",
            "344/344 [==============================] - 103s 298ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qIBLmQMcBZSd",
        "colab_type": "text"
      },
      "source": [
        "O score oficial dessa submissão foi 0.523, o melhor conseguido pelo grupo. Infelizmente, o resultado nos dados de treinamento não se repetiu nos dados de teste, isso pode indicar que houve overfitting do modelo.\n",
        "\n",
        "Alternativamente aos métodos utilizados nas duas últimas tentativas, a utilização do conjunto completo de amostras e maior número de épocas não gerou grandes ganhos em desempenho e score, indicamento, novamente, um overfitting."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "QLZBTNHiTt4V"
      },
      "source": [
        "## Rede neural multi-input\n",
        "\n",
        "Sabemos que essa base de dados, além das imagens, possui um dataframe com informações sobre o paciente referente a cada imagem de melanoma. Assim, é natural imaginar um método que utilize os dois tipos de informação para a classificação. Dessa forma, foi criada uma rede neural que recebe duas entradas, imagem e dataframe, para gerar a classificação."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Ap2yhU1sqiLl",
        "colab": {}
      },
      "source": [
        "# Imports\n",
        "from pathlib import Path\n",
        "\n",
        "from tensorflow.data import Dataset\n",
        "from tensorflow.keras import utils\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.layers import Input\n",
        "from tensorflow.keras.layers import concatenate\n",
        "\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator, load_img, img_to_array"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "4pN1sNkusUWr"
      },
      "source": [
        "Para que a rede neural receba duas entradas diferentes, essas entradas precisam estar em um formato compatível entre si. Assim, foi criada a classe MixedDataGenerator para gerar a entrada mista da rede neural.\n",
        "\n",
        "Essa classe permite que os dados sejam carregados na memória em pequenos pacotes e os pacotes são carregados iterativamente durante os processos de treinamento, validação e predição. Dessa forma, não é necessário carregar os dados todos de uma única vez, gerando menos impacto na memória RAM do computador."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "fNtsb_bdr4aD",
        "colab": {}
      },
      "source": [
        "class MixedDataGenerator(utils.Sequence):\n",
        "\n",
        "    def __init__(self, values: pd.DataFrame,\n",
        "                images: pd.Series, directory: str, labels: pd.Series=None,\n",
        "                target_size: tuple=(32,32), batch_size: int=32, shuffle: bool=True):\n",
        "        'Inicialização'\n",
        "\n",
        "        self.values = values\n",
        "        self.labels = labels\n",
        "        self.images = images\n",
        "        self.directory = Path(directory)\n",
        "        self.target_size = target_size\n",
        "        self.batch_size = batch_size\n",
        "        self.shuffle = shuffle\n",
        "        self.num_files = images.count()\n",
        "        self.columns = self.values.columns\n",
        "        self.indices = None\n",
        "        self.values_gen = ImageDataGenerator()\n",
        "        self.labels_indices = None\n",
        "        if labels is not None:\n",
        "            self.labels_indices = list(self.labels.unique())\n",
        "\n",
        "        self.on_epoch_end()\n",
        "\n",
        "    \n",
        "    def __len__(self):\n",
        "        'Indica a quantidades de pacotes por épocas'\n",
        "\n",
        "        return (self.num_files + self.batch_size - 1) // self.batch_size\n",
        "\n",
        "    \n",
        "    def __getitem__(self, index):\n",
        "        'Gera um pacote de dados'\n",
        "\n",
        "        # Gera os índices do pacote\n",
        "        if index >= len(self):\n",
        "            raise ValueError('Asked to retrieve element {index}, '\n",
        "                             'but the Sequence '\n",
        "                             'has length {length}'.format(index=index,\n",
        "                                                          length=len(self)))\n",
        "        indices = self.indices[index * self.batch_size : (index + 1) * self.batch_size]\n",
        "        x_images = [self.__process_image(image) for image in self.images.loc[indices]]\n",
        "        images = np.array(x_images)\n",
        "        x_values = self.values.loc[indices]\n",
        "        values = np.array(x_values)\n",
        "        if self.labels is not None:\n",
        "            y_labels = [self.labels_indices.index(label) for label in self.labels.loc[indices]]\n",
        "            labels = np.eye(len(self.labels_indices))[y_labels]\n",
        "        # Gera os dados\n",
        "        if self.labels is not None:\n",
        "            return [images, values], labels\n",
        "        else:\n",
        "            return [images, values]\n",
        "\n",
        "\n",
        "    def on_epoch_end(self):\n",
        "        'Atualiza os índices após cada época'\n",
        "\n",
        "        self.indices = np.arange(len(self.values))\n",
        "        if self.shuffle:\n",
        "            np.random.shuffle(self.indices)\n",
        "\n",
        "    \n",
        "    def __process_image(self, image_name):\n",
        "        'Lê a imagem da memória'\n",
        "\n",
        "        image = load_img(Path(self.directory, image_name), target_size=self.target_size)\n",
        "        image = self.values_gen.apply_transform(image, dict(rescale=1./255))    # normaliza valores dos pixels da imagem entre 0-1\n",
        "        array = img_to_array(image)\n",
        "        image.close()\n",
        "\n",
        "        return array"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "ITjAqNbv1Xb4"
      },
      "source": [
        "Alguns exemplos do dataframe possuem dados faltantes nas colunas 'sex', 'age_approx' e 'anatom_site_general_challenge'. Assim, foi feita uma função para tratar esses dados. \n",
        "\n",
        "Como os exemplos com dados faltante nas colunas 'sex' e 'age_approx' eram poucos, e estes só existiam no conjunto de treinamento, foram simplesmente removidos. \n",
        "\n",
        "Com relação ao atributo 'anatom_site_general_challenge', foi criada uma nova categoria chamada 'unknown' (desconhecido), e os dados faltantes foram substituídos por essa categoria."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "KG3V84f11W2f",
        "colab": {}
      },
      "source": [
        "def treatNan(data):\n",
        "    # remove nan na coluna 'sex'\n",
        "    data = data[data.sex == data.sex]\n",
        "    # remove nan na coluna 'age_approx'\n",
        "    data = data[data.age_approx == data.age_approx]\n",
        "\n",
        "    # substitui nan na coluna 'anatom_site_general_challenge' pelo atributo 'unknown'\n",
        "    data.anatom_site_general_challenge.fillna(value='unknown', inplace=True)\n",
        "\n",
        "    return data"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "JxMxeMIM27hV"
      },
      "source": [
        "### Leitura dos dados"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "A-os8Ckq3Cks",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "outputId": "d147a599-33a1-47c8-ae9e-946afc86c5c1"
      },
      "source": [
        "# Leitura dos dataframes dos arquivos csv\n",
        "train_data = pd.read_csv('train.csv')\n",
        "test_data = pd.read_csv('test.csv')\n",
        "\n",
        "# Adicionando a extensão aos nomes das imagens\n",
        "train_data['image_name'] = train_data['image_name'].astype(str)+\".jpg\"\n",
        "test_data['image_name'] = test_data['image_name'].astype(str)+\".jpg\"\n",
        "\n",
        "train_data = treatNan(train_data)\n",
        "test_data = treatNan(test_data)\n",
        "\n",
        "train_data.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>image_name</th>\n",
              "      <th>patient_id</th>\n",
              "      <th>sex</th>\n",
              "      <th>age_approx</th>\n",
              "      <th>anatom_site_general_challenge</th>\n",
              "      <th>diagnosis</th>\n",
              "      <th>benign_malignant</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>ISIC_2637011.jpg</td>\n",
              "      <td>IP_7279968</td>\n",
              "      <td>male</td>\n",
              "      <td>45.0</td>\n",
              "      <td>head/neck</td>\n",
              "      <td>unknown</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>ISIC_0015719.jpg</td>\n",
              "      <td>IP_3075186</td>\n",
              "      <td>female</td>\n",
              "      <td>45.0</td>\n",
              "      <td>upper extremity</td>\n",
              "      <td>unknown</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>ISIC_0052212.jpg</td>\n",
              "      <td>IP_2842074</td>\n",
              "      <td>female</td>\n",
              "      <td>50.0</td>\n",
              "      <td>lower extremity</td>\n",
              "      <td>nevus</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>ISIC_0068279.jpg</td>\n",
              "      <td>IP_6890425</td>\n",
              "      <td>female</td>\n",
              "      <td>45.0</td>\n",
              "      <td>head/neck</td>\n",
              "      <td>unknown</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>ISIC_0074268.jpg</td>\n",
              "      <td>IP_8723313</td>\n",
              "      <td>female</td>\n",
              "      <td>55.0</td>\n",
              "      <td>upper extremity</td>\n",
              "      <td>unknown</td>\n",
              "      <td>benign</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         image_name  patient_id     sex  age_approx  \\\n",
              "0  ISIC_2637011.jpg  IP_7279968    male        45.0   \n",
              "1  ISIC_0015719.jpg  IP_3075186  female        45.0   \n",
              "2  ISIC_0052212.jpg  IP_2842074  female        50.0   \n",
              "3  ISIC_0068279.jpg  IP_6890425  female        45.0   \n",
              "4  ISIC_0074268.jpg  IP_8723313  female        55.0   \n",
              "\n",
              "  anatom_site_general_challenge diagnosis benign_malignant  target  \n",
              "0                     head/neck   unknown           benign       0  \n",
              "1               upper extremity   unknown           benign       0  \n",
              "2               lower extremity     nevus           benign       0  \n",
              "3                     head/neck   unknown           benign       0  \n",
              "4               upper extremity   unknown           benign       0  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "8moVpLJm7W-A",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "outputId": "2edead6a-3f9d-4d9a-ebb6-f5cf63816801"
      },
      "source": [
        "test_data.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>image_name</th>\n",
              "      <th>patient_id</th>\n",
              "      <th>sex</th>\n",
              "      <th>age_approx</th>\n",
              "      <th>anatom_site_general_challenge</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>ISIC_0052060.jpg</td>\n",
              "      <td>IP_3579794</td>\n",
              "      <td>male</td>\n",
              "      <td>70.0</td>\n",
              "      <td>unknown</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>ISIC_0052349.jpg</td>\n",
              "      <td>IP_7782715</td>\n",
              "      <td>male</td>\n",
              "      <td>40.0</td>\n",
              "      <td>lower extremity</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>ISIC_0058510.jpg</td>\n",
              "      <td>IP_7960270</td>\n",
              "      <td>female</td>\n",
              "      <td>55.0</td>\n",
              "      <td>torso</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>ISIC_0073313.jpg</td>\n",
              "      <td>IP_6375035</td>\n",
              "      <td>female</td>\n",
              "      <td>50.0</td>\n",
              "      <td>torso</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>ISIC_0073502.jpg</td>\n",
              "      <td>IP_0589375</td>\n",
              "      <td>female</td>\n",
              "      <td>45.0</td>\n",
              "      <td>lower extremity</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         image_name  patient_id     sex  age_approx  \\\n",
              "0  ISIC_0052060.jpg  IP_3579794    male        70.0   \n",
              "1  ISIC_0052349.jpg  IP_7782715    male        40.0   \n",
              "2  ISIC_0058510.jpg  IP_7960270  female        55.0   \n",
              "3  ISIC_0073313.jpg  IP_6375035  female        50.0   \n",
              "4  ISIC_0073502.jpg  IP_0589375  female        45.0   \n",
              "\n",
              "  anatom_site_general_challenge  \n",
              "0                       unknown  \n",
              "1               lower extremity  \n",
              "2                         torso  \n",
              "3                         torso  \n",
              "4               lower extremity  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "BFIwd_J63sgc"
      },
      "source": [
        "Os dados provenientes do arquivo 'train.csv' foram separados em treinamento e validação de forma a manter a proporção de amostras das classes 'benign' e 'malignant' nos dois conjuntos."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "UcaTrPhT3xXS",
        "colab": {}
      },
      "source": [
        "# Separa os dados em treinamento e validação mantendo a proporção entre classes 'benign' e 'malignant'\n",
        "train_data_ben = train_data[train_data.benign_malignant == 'benign']\n",
        "train_data_mal = train_data[train_data.benign_malignant == 'malignant']\n",
        "\n",
        "valid_data_ben = train_data_ben.sample(frac=0.3, random_state=1)\n",
        "valid_data_mal = train_data_mal.sample(frac=0.3, random_state=1)\n",
        "\n",
        "valid_data = valid_data_ben.append(valid_data_mal)\n",
        "\n",
        "train_data.drop(index=valid_data.index)\n",
        "\n",
        "valid_data = valid_data.reindex(np.random.permutation(valid_data.index))\n",
        "valid_data.reset_index(drop=True, inplace=True)\n",
        "train_data.reset_index(drop=True, inplace=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "6IFMHPRx6VTX"
      },
      "source": [
        "As colunas de rótulos e atributos foram separadas. Foram selecioandos apenas os atributos que geram informação para o modelo, ou seja, excluímos as colunas 'image_name' e 'patient_id', e que também constam nos dados de teste. Assim, apenas as colunas 'sex', 'age_approx' e 'anatom_site_general_challenge' foram selecionadas."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "YXxnynne4kd4",
        "colab": {}
      },
      "source": [
        "# Rótulos dos dados\n",
        "y_train = train_data.benign_malignant\n",
        "y_valid = valid_data.benign_malignant\n",
        "\n",
        "# Dados com atributos para o treinamento\n",
        "X_train = train_data[['sex', 'age_approx', 'anatom_site_general_challenge']]\n",
        "X_valid = valid_data[['sex', 'age_approx', 'anatom_site_general_challenge']]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "nOjhav1I-0Wo"
      },
      "source": [
        "Essa base de dados possui dados categóricos, contudo a rede neural não aceita esse tipo de dado. Assim, usamos o método [get_dummies()](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Series.str.get_dummies.html) para transformar os dados categórico em numéricos."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "_jKOJvDh_qcA",
        "colab": {}
      },
      "source": [
        "train_sex = X_train['sex'].str.get_dummies()\n",
        "train_anatom = X_train['anatom_site_general_challenge'].str.get_dummies()\n",
        "\n",
        "valid_sex = X_valid['sex'].str.get_dummies()\n",
        "valid_anatom = X_valid['anatom_site_general_challenge'].str.get_dummies()\n",
        "\n",
        "X_train = X_train.join([train_sex, train_anatom])\n",
        "X_valid = X_valid.join([valid_sex, valid_anatom])\n",
        "\n",
        "X_train.drop(columns=['sex', 'anatom_site_general_challenge'], inplace=True)\n",
        "X_valid.drop(columns=['sex', 'anatom_site_general_challenge'], inplace=True)\n",
        "\n",
        "train_columns = X_train.columns\n",
        "valid_columns = X_valid.columns\n",
        "\n",
        "for column in valid_columns:\n",
        "    if column not in train_columns:\n",
        "        X_train[column] = 0\n",
        "\n",
        "for column in train_columns:\n",
        "    if column not in valid_columns:\n",
        "        X_valid[column] = 0"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "vxlBDD-e_5ai"
      },
      "source": [
        "Finalmente, as imagens de treinamento e validação forma lidas e os dados mistos forma criados."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Fj4ZrdRtADs_",
        "colab": {}
      },
      "source": [
        "gen_train = MixedDataGenerator(values=X_train,\n",
        "                            images=train_data['image_name'],\n",
        "                            directory='train',\n",
        "                            labels=y_train,\n",
        "                            target_size=(32, 32),\n",
        "                            batch_size=32)\n",
        "gen_valid = MixedDataGenerator(values=X_valid,\n",
        "                            images=valid_data['image_name'],\n",
        "                            directory='train',\n",
        "                            labels=y_valid,\n",
        "                            target_size=(32, 32),\n",
        "                            batch_size=32)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "6vZycFrvAJHd"
      },
      "source": [
        "### Criação da rede neural multi-input\n",
        "\n",
        "A rede neural multi-input consiste na concatenação de duas, ou mais, redes neurais para gerar uma única saída. Nesse caso, a nossa rede neural é uma concatenação de um Multi Layer Perceptron (MLP) e uma Rede Convolucional (CNN) como é mostrado na figura a seguir.\n",
        "\n",
        "\n",
        "<img src=\"https://drive.google.com/uc?id=122h0V2nbJCg-UUfUNfXGFSl5aUmwHZqL\" alt=\"drawing\" width=\"400\"/>\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "FR41sjFxnPOD"
      },
      "source": [
        "Para criar a rede, primeiramente foram criados modelos de MLP e CNN usando as classes [Sequential()](https://keras.io/api/models/sequential/#sequential-class) e [Model()](https://keras.io/api/models/model/#model-class) do Keras, respectivamente. É importante ressaltar que os modelos só serão compilados após a concatenação."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "29UmK2zAo1cF",
        "colab": {}
      },
      "source": [
        "def createMLP(dimension):\n",
        "    model = Sequential()\n",
        "    model.add(Dense(units=8, input_dim=dimension, activation=\"relu\"))\n",
        "    model.add(Dense(units=4, activation=\"relu\"))\n",
        "\n",
        "    return model\n",
        "\n",
        "\n",
        "def createCNN(shape):\n",
        "    inputImage = Input(shape=shape)\n",
        "    x = inputImage\n",
        "    x = Conv2D(filters=32, kernel_size=(7,7), activation='relu')(x)\n",
        "    x = MaxPooling2D(pool_size = (2, 2))(x)\n",
        "    x = Conv2D(filters=20, kernel_size=(3, 3), activation = 'relu')(x)\n",
        "    x = MaxPooling2D(pool_size = (2, 2))(x)\n",
        "\n",
        "    x = Flatten()(x)\n",
        "    x = Dense(units = 10, activation = 'relu')(x)\n",
        "    x = Dense(units = 4, activation = 'relu')(x)\n",
        "\n",
        "    model = Model(inputImage, x)\n",
        "\n",
        "    return model\n",
        "\n",
        "\n",
        "#Criação do modelo multi-Input\n",
        "mlp = createMLP(dimension=len(X_train.columns))\n",
        "cnn = createCNN((32,32,3))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "VEm8qi3bqoJi"
      },
      "source": [
        "Em seguida, foi criado um modelo concatenando as saídas do MLP e da CNN. Foram adicionadas uma camada oculta e, por fim, uma de saída. E, finalmente, para compilar o nosso modelo, escolhemos o otimizador [Adam](https://machinelearningmastery.com/adam-optimization-algorithm-for-deep-learning/) e a função de loss [binary_crossentropy](https://keras.io/api/losses/probabilistic_losses/#binarycrossentropy-class)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "XlgNjrwDq4gt",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 660
        },
        "outputId": "79d63e00-c203-426e-f953-b61901d4364b"
      },
      "source": [
        "combinedInput = concatenate([cnn.output, mlp.output])\n",
        "\n",
        "x = Dense(units=4, activation=\"relu\")(combinedInput)\n",
        "x = Dense(units=2, activation=\"sigmoid\")(x)\n",
        "\n",
        "model = Model(inputs=[cnn.input, mlp.input], outputs=x)\n",
        "\n",
        "model.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"functional_3\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            [(None, 32, 32, 3)]  0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv2d (Conv2D)                 (None, 26, 26, 32)   4736        input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D)    (None, 13, 13, 32)   0           conv2d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_1 (Conv2D)               (None, 11, 11, 20)   5780        max_pooling2d[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2D)  (None, 5, 5, 20)     0           conv2d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "flatten (Flatten)               (None, 500)          0           max_pooling2d_1[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "dense_input (InputLayer)        [(None, 10)]         0                                            \n",
            "__________________________________________________________________________________________________\n",
            "dense_2 (Dense)                 (None, 10)           5010        flatten[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense (Dense)                   (None, 8)            88          dense_input[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense_3 (Dense)                 (None, 4)            44          dense_2[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 4)            36          dense[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "concatenate (Concatenate)       (None, 8)            0           dense_3[0][0]                    \n",
            "                                                                 dense_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense_4 (Dense)                 (None, 4)            36          concatenate[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense_5 (Dense)                 (None, 2)            10          dense_4[0][0]                    \n",
            "==================================================================================================\n",
            "Total params: 15,740\n",
            "Trainable params: 15,740\n",
            "Non-trainable params: 0\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "iNib-LjasAxw"
      },
      "source": [
        "Para fazer o treinamento, usamos o método fit com as imagens de treinamento e validação. O parâmetro workers faz paralelização do processamento quando múltiplos cores de processamento estão disponíveis."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "siwF1z7XsG78",
        "colab": {},
        "outputId": "85c6aa61-4089-4be8-8815-0ded9f0283f6"
      },
      "source": [
        "# Treinamento\n",
        "\n",
        "model.fit(gen_train,\n",
        "          epochs=4,\n",
        "          validation_data=gen_valid,\n",
        "          workers=16)\n",
        "model.save_weights('cnn_melanoma.h5')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/4\n",
            "1034/1034 [==============================] - 396s 383ms/step - loss: 0.1125 - accuracy: 0.9777 - val_loss: 0.0910 - val_accuracy: 0.9825\n",
            "Epoch 2/4\n",
            "1034/1034 [==============================] - 396s 383ms/step - loss: 0.0870 - accuracy: 0.9822 - val_loss: 0.0811 - val_accuracy: 0.9824\n",
            "Epoch 3/4\n",
            "1034/1034 [==============================] - 398s 385ms/step - loss: 0.0841 - accuracy: 0.9824 - val_loss: 0.0794 - val_accuracy: 0.9824\n",
            "Epoch 4/4\n",
            "1034/1034 [==============================] - 396s 383ms/step - loss: 0.0834 - accuracy: 0.9824 - val_loss: 0.0868 - val_accuracy: 0.9824\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "Wn5bfEgGsc4a"
      },
      "source": [
        "Antes de fazer a inferência dos dados de teste, vamos analisar o resultado nos dados de validação."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "SJYhf8_Uslwa",
        "colab": {},
        "outputId": "d90251c5-8ce3-41e2-b25b-1a33ab8f1d38"
      },
      "source": [
        "# Validação\n",
        "\n",
        "validation_predictions = model.predict(gen_valid,\n",
        "                                       verbose=1,\n",
        "                                       workers=16)\n",
        "validation_predictions = (validation_predictions > 0.5).astype(int)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "310/310 [==============================] - 89s 289ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "S-N00olatSKJ"
      },
      "source": [
        "Para obter alguns dados de avaliação, utilizamos a função [classification_report](https://scikit-learn.org/stable/modules/generated/sklearn.metrics.classification_report.html?highlight=classification_report#sklearn.metrics.classification_report) da biblioteca ScikitLearn."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "75TOsYxitrq_",
        "colab": {},
        "outputId": "aa92aba0-9deb-4b15-dc37-aa7d2f762014"
      },
      "source": [
        "# Faz o classification report do modelo aplicado nos dados de validação\n",
        "from sklearn.metrics import classification_report\n",
        "report = classification_report(pd.get_dummies(y_valid),\n",
        "                               validation_predictions)\n",
        "print(report)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.98      1.00      0.99      9742\n",
            "           1       0.00      0.00      0.00       175\n",
            "\n",
            "   micro avg       0.98      0.98      0.98      9917\n",
            "   macro avg       0.49      0.50      0.50      9917\n",
            "weighted avg       0.97      0.98      0.97      9917\n",
            " samples avg       0.98      0.98      0.98      9917\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "woXYesN-s0S4"
      },
      "source": [
        "Para podermos visualizar o resultado, contruímos a matriz de confusão para os dados de validação"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Oj4ELpant-0e",
        "colab": {},
        "outputId": "0d720973-613b-4b00-a1a5-7ef68ec517ce"
      },
      "source": [
        "# Mostra a matriz de confusão dos dados de validação\n",
        "plot_confusion_matrix(pd.get_dummies(y_valid).values.argmax(axis=1),\n",
        "                      validation_predictions.argmax(axis=1),['benign', 'malignant'], \"Matriz de confusão\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 396x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "6mLyj-jD_N4W"
      },
      "source": [
        "É possível observar que o modelo não classificou nenhum exemplo como 'malignant'. Isso provavelmente se deve ao tamanho bastante reduzido das imagens, o uso da acurácia como métrica do modelo em uma base desbalanceada e, principalmente, ao enorme desbalanceamento da base de dados em si."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "fHih9gY-_2WU"
      },
      "source": [
        "Para fazer a submissão, inferimos os dados de teste."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "zdVDRLlSzl92",
        "colab": {},
        "outputId": "f6dab559-2b0b-4399-9e8f-8665ad054ee3"
      },
      "source": [
        "# Teste\n",
        "\n",
        "X_test = test_data.drop(columns=['image_name', 'patient_id'])\n",
        "\n",
        "test_sex = X_test['sex'].str.get_dummies()\n",
        "test_anatom = X_test['anatom_site_general_challenge'].str.get_dummies()\n",
        "\n",
        "X_test = X_test.join([test_sex, test_anatom])\n",
        "\n",
        "X_test.drop(columns=['sex', 'anatom_site_general_challenge'], inplace=True)\n",
        "\n",
        "test_columns = X_test.columns\n",
        "\n",
        "for column in train_columns:\n",
        "    if column not in test_columns:\n",
        "        X_test[column] = 0\n",
        "\n",
        "\n",
        "gen_test = MixedDataGenerator(values=X_test,\n",
        "                            images=test_data['image_name'],\n",
        "                            directory='test',\n",
        "                            target_size=(32, 32),\n",
        "                            batch_size=32)\n",
        "\n",
        "\n",
        "predictions = model.predict(gen_test,\n",
        "                            verbose=1,\n",
        "                            workers=16)\n",
        "\n",
        "\n",
        "# predictions é um vetor em que cada elemento é um vetor de dois valores, já que temos duas classes (benigno e maligno).\n",
        "# Pega-se então a probabilidade do tumor ser maligno\n",
        "submission = pd.DataFrame(test_data['image_name'].str.replace(r'.jpg', ''))\n",
        "submission['target']=predictions[:,1]\n",
        "\n",
        "# Cria arquivo de submissão\n",
        "submission.to_csv('multi_input_submission.csv',index=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "344/344 [==============================] - 100s 291ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "b2emHw0H-GtF",
        "colab_type": "text"
      },
      "source": [
        "O score oficial dessa submissão foi 0.499.\n",
        "\n",
        "Esse resultado não é satisfatório, mas já era esperado, tendo em vista que que o modelo não classificou nehum dos dados de validação como pertencentes à classe 'malignant'. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "BsX-aA1hAmvg"
      },
      "source": [
        "### Segunda submissão\n",
        "\n",
        "Para uma nova submissão, algumas mudanças foram feitas a fim de tentar melhorar a classificação do modelo. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "PV7G4-hLODF7"
      },
      "source": [
        "Primeiramente, as imagens foram carregadas com uma resolução de 128x128 pixels"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "GBPmbQDwNg38",
        "colab": {}
      },
      "source": [
        "# Leitura dos dataframes dos arquivos csv\n",
        "train_data = pd.read_csv('train.csv')\n",
        "test_data = pd.read_csv('test.csv')\n",
        "\n",
        "# Adicionando a extensão aos nomes das imagens\n",
        "train_data['image_name'] = train_data['image_name'].astype(str)+\".jpg\"\n",
        "test_data['image_name'] = test_data['image_name'].astype(str)+\".jpg\"\n",
        "\n",
        "train_data = treatNan(train_data)\n",
        "test_data = treatNan(test_data)\n",
        "\n",
        "\n",
        "# Separa os dados em treinamento e validação mantendo a proporção entre classes 'benign' e 'malignant'\n",
        "train_data_ben = train_data[train_data.benign_malignant == 'benign']\n",
        "train_data_mal = train_data[train_data.benign_malignant == 'malignant']\n",
        "\n",
        "valid_data_ben = train_data_ben.sample(frac=0.3, random_state=1)\n",
        "valid_data_mal = train_data_mal.sample(frac=0.3, random_state=1)\n",
        "\n",
        "valid_data = valid_data_ben.append(valid_data_mal)\n",
        "\n",
        "train_data.drop(index=valid_data.index)\n",
        "\n",
        "valid_data = valid_data.reindex(np.random.permutation(valid_data.index))\n",
        "valid_data.reset_index(drop=True, inplace=True)\n",
        "train_data.reset_index(drop=True, inplace=True)\n",
        "\n",
        "\n",
        "# Rótulos dos dados\n",
        "y_train = train_data.benign_malignant\n",
        "y_valid = valid_data.benign_malignant\n",
        "\n",
        "# Dados com atributos para o treinamento\n",
        "X_train = train_data[['sex', 'age_approx', 'anatom_site_general_challenge']]\n",
        "X_valid = valid_data[['sex', 'age_approx', 'anatom_site_general_challenge']]\n",
        "\n",
        "\n",
        "train_sex = X_train['sex'].str.get_dummies()\n",
        "train_anatom = X_train['anatom_site_general_challenge'].str.get_dummies()\n",
        "\n",
        "valid_sex = X_valid['sex'].str.get_dummies()\n",
        "valid_anatom = X_valid['anatom_site_general_challenge'].str.get_dummies()\n",
        "\n",
        "X_train = X_train.join([train_sex, train_anatom])\n",
        "X_valid = X_valid.join([valid_sex, valid_anatom])\n",
        "\n",
        "X_train.drop(columns=['sex', 'anatom_site_general_challenge'], inplace=True)\n",
        "X_valid.drop(columns=['sex', 'anatom_site_general_challenge'], inplace=True)\n",
        "\n",
        "train_columns = X_train.columns\n",
        "valid_columns = X_valid.columns\n",
        "\n",
        "for column in valid_columns:\n",
        "    if column not in train_columns:\n",
        "        X_train[column] = 0\n",
        "\n",
        "for column in train_columns:\n",
        "    if column not in valid_columns:\n",
        "        X_valid[column] = 0\n",
        "\n",
        "gen_train = MixedDataGenerator(values=X_train,\n",
        "                            images=train_data['image_name'],\n",
        "                            directory='train',\n",
        "                            labels=y_train,\n",
        "                            target_size=(128,128),\n",
        "                            batch_size=32)\n",
        "gen_valid = MixedDataGenerator(values=X_valid,\n",
        "                            images=valid_data['image_name'],\n",
        "                            directory='train',\n",
        "                            labels=y_valid,\n",
        "                            target_size=(128,128),\n",
        "                            batch_size=32)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "AQP2RlRVOxSQ"
      },
      "source": [
        "Nesse modelo, foi usada a área abaixo da curva ROC como métrica de avaliação em vez da acurácia."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "dt1B3BA3OzUH",
        "colab": {}
      },
      "source": [
        "#Criação do modelo multi-Input\n",
        "\n",
        "mlp = createMLP(dimension=len(X_train.columns))\n",
        "cnn = createCNN((128,128,3))\n",
        "\n",
        "combinedInput = concatenate([cnn.output, mlp.output])\n",
        "\n",
        "x = Dense(units=4, activation=\"relu\")(combinedInput)\n",
        "x = Dense(units=2, activation=\"sigmoid\")(x)\n",
        "\n",
        "model = Model(inputs=[cnn.input, mlp.input], outputs=x)\n",
        "\n",
        "model.compile(optimizer = 'adam', loss = 'categorical_crossentropy', metrics = [AUC()])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "Rh5d_xfZPF0M"
      },
      "source": [
        "Além disso, como uma tentativa de compensar o desbalanceamento da base, foram definidos pesos para cada classe, utilizando a função [compute_class_weight()](https://scikit-learn.org/stable/modules/generated/sklearn.utils.class_weight.compute_class_weight.html?highlight=class_weight#sklearn.utils.class_weight.compute_class_weight) do ScikitLearn."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "S2tVnVtNQpk0",
        "colab": {}
      },
      "source": [
        "from sklearn.utils import class_weight\n",
        "\n",
        "class_weights = class_weight.compute_class_weight('balanced',\n",
        "                                                 np.unique(y_train),\n",
        "                                                 y_train)\n",
        "\n",
        "class_weights = dict(zip([0,1], class_weights))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5mFar7ED6Y9Z",
        "colab_type": "code",
        "colab": {},
        "outputId": "d91f95e5-baa3-4cf9-bae7-d7422ee86558"
      },
      "source": [
        "# Treinamento\n",
        "model.fit(gen_train,\n",
        "          epochs=4,\n",
        "          validation_data=gen_valid,\n",
        "          class_weight=class_weights,\n",
        "          workers=16)\n",
        "model.save_weights('cnn_melanoma.h5')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/4\n",
            "1034/1034 [==============================] - 394s 381ms/step - loss: 0.6933 - auc: 0.4124 - val_loss: 0.6920 - val_auc: 0.5000\n",
            "Epoch 2/4\n",
            "1034/1034 [==============================] - 391s 378ms/step - loss: 0.6933 - auc: 0.3715 - val_loss: 0.6914 - val_auc: 0.5000\n",
            "Epoch 3/4\n",
            "1034/1034 [==============================] - 390s 378ms/step - loss: 0.6933 - auc: 0.5553 - val_loss: 0.6912 - val_auc: 0.5000\n",
            "Epoch 4/4\n",
            "1034/1034 [==============================] - 392s 380ms/step - loss: 0.6933 - auc: 0.8064 - val_loss: 0.6922 - val_auc: 0.5000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I2cOSS3l6Y9c",
        "colab_type": "code",
        "colab": {},
        "outputId": "2ce3c69e-c09b-4c66-9fcf-8f9e8bbef0b6"
      },
      "source": [
        "# Validação\n",
        "\n",
        "validation_predictions = model.predict(gen_valid,\n",
        "                                       verbose=1,\n",
        "                                       workers=16)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "310/310 [==============================] - 89s 288ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "STgdRBtEQvny",
        "colab": {},
        "outputId": "443b660c-255e-40fb-c1b9-8caf0de33dc1"
      },
      "source": [
        "validation_predictions = (validation_predictions > 0.5).astype(int)\n",
        "\n",
        "\n",
        "# class_labels = list(validation_generator.class_indices.keys())\n",
        "# validation_true = validation_generator.classes\n",
        "\n",
        "# Faz o classification report do modelo aplicado nos dados de validação\n",
        "from sklearn.metrics import classification_report\n",
        "report = classification_report(pd.get_dummies(y_valid),\n",
        "                               validation_predictions)\n",
        "print(report)\n",
        "\n",
        "# Plota a matriz de confusão dos dados de validação\n",
        "plot_confusion_matrix(pd.get_dummies(y_valid).values.argmax(axis=1),\n",
        "                      validation_predictions.argmax(axis=1),['benign', 'malignant'], \"Matriz de confusão\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.98      1.00      0.99      9742\n",
            "           1       0.00      0.00      0.00       175\n",
            "\n",
            "   micro avg       0.98      0.98      0.98      9917\n",
            "   macro avg       0.49      0.50      0.50      9917\n",
            "weighted avg       0.97      0.98      0.97      9917\n",
            " samples avg       0.98      0.98      0.98      9917\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 396x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AT0w8w3B6Y9f",
        "colab_type": "code",
        "colab": {},
        "outputId": "fd22d4b7-e7f6-40ec-92af-5cd07075f331"
      },
      "source": [
        "# Teste\n",
        "\n",
        "X_test = test_data.drop(columns=['image_name', 'patient_id'])\n",
        "\n",
        "test_sex = X_test['sex'].str.get_dummies()\n",
        "test_anatom = X_test['anatom_site_general_challenge'].str.get_dummies()\n",
        "\n",
        "X_test = X_test.join([test_sex, test_anatom])\n",
        "\n",
        "X_test.drop(columns=['sex', 'anatom_site_general_challenge'], inplace=True)\n",
        "\n",
        "test_columns = X_test.columns\n",
        "\n",
        "for column in train_columns:\n",
        "    if column not in test_columns:\n",
        "        X_test[column] = 0\n",
        "\n",
        "\n",
        "gen_test = MixedDataGenerator(values=X_test,\n",
        "                            images=test_data['image_name'],\n",
        "                            directory='test',\n",
        "                            target_size=(128, 128),\n",
        "                            batch_size=32)\n",
        "\n",
        "\n",
        "predictions = model.predict(gen_test,\n",
        "                            verbose=1,\n",
        "                            workers=16)\n",
        "\n",
        "\n",
        "# predictions é um vetor em que cada elemento é um vetor de dois valores, já que temos duas classes (benigno e maligno).\n",
        "# Pega-se então a probabilidade do tumor ser maligno\n",
        "submission = pd.DataFrame(test_data['image_name'].str.replace(r'.jpg', ''))\n",
        "submission['target']=predictions[:,1]\n",
        "\n",
        "# Cria arquivo de submissão\n",
        "submission.to_csv('multi_input_submission2.csv',index=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "344/344 [==============================] - 101s 295ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IurCUUiQ-l8F",
        "colab_type": "text"
      },
      "source": [
        "O score oficial dessa submissão foi 0.500."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ffkLUbCU-pJH",
        "colab_type": "text"
      },
      "source": [
        "Após a submissão dos resultados, o score obtido foi ligeiramente maior. Contudo, a diferença não foi significativa, o que nos faz perceber que o uso de pesos para as classes ainda não foi suficiente para compensar o grande desbalanceamento da base de dados.\n",
        "\n",
        "Além disso, comparando o modelo multi-input com a CNN, observamos que utilizar as características dos pacientes contidas no dataframe não gerou um ganho de informação suficiente a ponto de melhorar significativamente a capacidade de classificação do modelo para essa base de dados."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9r0cr6AchAgK",
        "colab_type": "text"
      },
      "source": [
        "## Conclusão\n",
        "\n",
        "Os resultados obtidos na competição foram:\n",
        "\n",
        "* Modelo Simples - 0.506\n",
        "\n",
        "* Modelo VGG - 0.500\n",
        "\n",
        "* Modelo Resnet50, augmentações e peso de classe - 0.489\n",
        "\n",
        "* Modelo com transfer learning ResNet50 - 0.509\n",
        "\n",
        "* Modelo com transfer learning EfficientNetB3 - 0.482\n",
        "\n",
        "* Modelo com transfer learning EfficientNetB0 - 0.523\n",
        "\n",
        "* Modelo com rede multi-input 1 - 0.499\n",
        "\n",
        "* Modelo com rede multi-input 2 - 0.500\n",
        "\n",
        "\n",
        "A melhor submissão alcançou um scor de 0.523 e a colocação 1337 em 29/06/2020\n",
        "\n",
        "[Link para notebook no kaggle](https://www.kaggle.com/nataliaborges/compara-o-entre-modelos-usando-keras)\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TiWNGFTFi4KS",
        "colab_type": "text"
      },
      "source": [
        "## Trabalhos futuros\n",
        "\n",
        "Por fim, pensando em alcançar resultados mais precisos e mantendo como modelo redes neurais artificiais, seria necessário um trabalho mais profundo em pré-processamento do banco de dados e, também, é possível que seja abordado outra parte do dataset disponibilizado.\n",
        "\n",
        "Além do dataset abordado até o momento, formado por imagens do tipo JPEG para treinamento e  teste, a competição também disponibiliza o mesmo dataset no formato DICOM (Digital Imaging and COmmunications in Medicine).\n",
        "\n",
        "A partir de nossas pesquisas e comparações entre diferentes formatos de imagem, tal formato pode ser dito como promissor para a realização de um estudo e aplicação de modelos, mas mantendo em mente à necessidade de um poder de processamento mais robusto.\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bpd-QnyzQ3Q_",
        "colab_type": "text"
      },
      "source": [
        "### DICOM\n",
        "\n",
        "DICOM é o formato padronizado para uso em aplicações destinadas à medicina. Basicamente, o formato DICOM consegue abrigar, além das informações pixel a pixel para representação da imagem em si, dados que descrevem a imagem (metadata).\n",
        "\n",
        "O que torna tal formato interessante para aplicações de Machine Learning é o fato de representar nativamente as imagens no espaço de cor YCrCb, que, em nível computacional, consegue agregar um detalhamento maior para o que está sendo ilustrado.\n",
        "\n",
        "Fazendo uma comparação direta, temos que a perda de detalhamento devido a compressão de dados que ocorre no formato JPEG possibilita que o formato DICOM seja considerado interessante para tais aplicações.\n",
        "Em python, há bibliotecas, como pydicom, que possibilitam uma fácil manipulação desse formato.\n",
        "\n",
        "Fontes: [1](http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.452.4373&rep=rep1&type=pdf) [2](https://en.wikipedia.org/wiki/DICOM)\n",
        "\n"
      ]
    }
  ]
}