modelo_least_loss.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4",
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "7xXSKt4mbBIL"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import cv2\n",
        "import os\n",
        "import zipfile\n",
        "from PIL import Image\n",
        "import tensorflow as tf"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pD-8crTwB-Z0",
        "outputId": "c1a6cf3b-7037-4fc5-f4f6-7aebf8bb2235"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "pip install requests"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KhaCvu2h3nCb",
        "outputId": "69784d83-a3ef-40ea-bb73-e37899ef4b39"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: requests in /usr/local/lib/python3.10/dist-packages (2.31.0)\n",
            "Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.10/dist-packages (from requests) (3.3.2)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests) (3.4)\n",
            "Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.10/dist-packages (from requests) (2.0.7)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests) (2023.7.22)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Definir o endereço para baixar arquivo\n",
        "!wget -P /content/drive/MyDrive/ELT579/Problema4 https://data.lhncbc.nlm.nih.gov/public/Malaria/cell_images.zip\n"
      ],
      "metadata": {
        "id": "9zuSL3Asb2wh",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b5a2cb6c-5b39-4805-b0dc-2e357dad1264"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2023-11-30 19:49:40--  https://data.lhncbc.nlm.nih.gov/public/Malaria/cell_images.zip\n",
            "Resolving data.lhncbc.nlm.nih.gov (data.lhncbc.nlm.nih.gov)... 108.138.94.6, 108.138.94.36, 108.138.94.100, ...\n",
            "Connecting to data.lhncbc.nlm.nih.gov (data.lhncbc.nlm.nih.gov)|108.138.94.6|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 353452851 (337M) [application/zip]\n",
            "Saving to: ‘/content/drive/MyDrive/ELT579/Problema4/cell_images.zip’\n",
            "\n",
            "cell_images.zip     100%[===================>] 337.08M  41.4MB/s    in 7.0s    \n",
            "\n",
            "2023-11-30 19:49:48 (48.1 MB/s) - ‘/content/drive/MyDrive/ELT579/Problema4/cell_images.zip’ saved [353452851/353452851]\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Extrair conteúdo dos arquivos comprimidos\n",
        "local_zip = '/content/drive/MyDrive/ELT579/Problema4/cell_images.zip'\n",
        "zip_ref = zipfile.ZipFile(local_zip, 'r')\n",
        "zip_ref.extractall('/content/drive/MyDrive/ELT579/Problema4')\n",
        "zip_ref.close()"
      ],
      "metadata": {
        "id": "Rxh8sYBx50Ve"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "pasta = '/content/drive/MyDrive/ELT579/Problema4/cell_images/'\n",
        "\n",
        "positivas = os.listdir(pasta + 'Parasitized/')\n",
        "\n",
        "negativas = os.listdir(pasta + 'Uninfected/')"
      ],
      "metadata": {
        "id": "xzVO5Nz79fTs"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "dataset = list()\n",
        "\n",
        "label = list()\n",
        "\n",
        "SIZE = 64\n",
        "\n",
        "for i, image_name in enumerate(positivas):\n",
        "  if(image_name.split('.')[1] == 'png'):\n",
        "    img = cv2.imread(pasta + 'Parasitized/' + image_name)\n",
        "    img = Image.fromarray(img, 'RGB')\n",
        "    img = img.resize((SIZE, SIZE))\n",
        "    dataset.append(np.array(img))\n",
        "    label.append(1)\n",
        "\n",
        "for i, image_name in enumerate(negativas):\n",
        "  if(image_name.split('.')[1] == 'png'):\n",
        "    img = cv2.imread(pasta + 'Uninfected/' + image_name)\n",
        "    img = Image.fromarray(img, 'RGB')\n",
        "    img = img.resize((SIZE, SIZE))\n",
        "    dataset.append(np.array(img))\n",
        "    label.append(0)"
      ],
      "metadata": {
        "id": "m_6LEadddCbM"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "print('label = ', label[-2])\n",
        "\n",
        "print('imagem')\n",
        "plt.imshow(dataset[-2])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 484
        },
        "id": "FloEBnQSxT3K",
        "outputId": "ec45ea8d-b2a0-4de1-a496-c547b1c802af"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "label =  0\n",
            "imagem\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7df80638a890>"
            ]
          },
          "metadata": {},
          "execution_count": 6
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "X = np.array(dataset)\n",
        "\n",
        "X = X/255.0\n",
        "\n",
        "y = tf.keras.utils.to_categorical(np.array(label))\n",
        "\n"
      ],
      "metadata": {
        "id": "caSO7Pvd1p85"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(np.shape(X))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VXwTRUhP2CJc",
        "outputId": "59d9b89c-b55a-4790-b58c-912c5c50894b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(27558, 64, 64, 3)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(np.shape(y))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Tk6H83UJyk9u",
        "outputId": "d4749dba-298a-4e85-c32e-3d0ed618ad00"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(27558, 2)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(y)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kFGhAe2417ON",
        "outputId": "8569f436-a0cf-47cd-9c4d-fbf6bea5423c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[0. 1.]\n",
            " [0. 1.]\n",
            " [0. 1.]\n",
            " ...\n",
            " [1. 0.]\n",
            " [1. 0.]\n",
            " [1. 0.]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.20, random_state = 0)\n",
        "X_train, X_val, y_train, y_val  = train_test_split(X_train, y_train, test_size=0.20, random_state=0)"
      ],
      "metadata": {
        "id": "4GtqxTJH3m9Z"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(np.shape(X_train))\n",
        "\n",
        "print(np.shape(y_train))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hqc5VuBHfCfv",
        "outputId": "ebb3cfc1-07ae-438a-e2dc-86a2d684a37b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(17636, 64, 64, 3)\n",
            "(17636, 2)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(np.shape(X_test))\n",
        "\n",
        "print(np.shape(y_test))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mZtJsyIsyIs3",
        "outputId": "81ff0ffc-a6a1-4945-aa68-851206921ff8"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(5512, 64, 64, 3)\n",
            "(5512, 2)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(np.shape(X_val))\n",
        "\n",
        "print(np.shape(y_val))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3DGdFu3lyJpA",
        "outputId": "d3e891b3-a265-4d88-d8d6-c0b056725211"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(4410, 64, 64, 3)\n",
            "(4410, 2)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import gc\n",
        "\n",
        "gc.collect()"
      ],
      "metadata": {
        "id": "lqskf9_-tdAJ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "96670b60-be0a-4835-b4be-2be28742708f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0"
            ]
          },
          "metadata": {},
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "gc.enable()"
      ],
      "metadata": {
        "id": "_oQfRNE3Hl1r"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def train_model(X_train, y_train, conv_nodes, dense_nodes, dropout_prob, lr, batch_size, epochs):\n",
        "  nn_model = tf.keras.Sequential([\n",
        "      tf.keras.layers.Conv2D(conv_nodes, kernel_size = (3,3), activation = 'relu', padding = 'same',  input_shape=(64,64,3)),\n",
        "      tf.keras.layers.MaxPooling2D(pool_size = (2,2)),\n",
        "      tf.keras.layers.Dropout(dropout_prob),\n",
        "      tf.keras.layers.Conv2D(conv_nodes, kernel_size = (3,3), activation = 'relu', padding = 'same'),\n",
        "      tf.keras.layers.MaxPooling2D(pool_size = (2,2)),\n",
        "      tf.keras.layers.Dropout(dropout_prob),\n",
        "      tf.keras.layers.Conv2D(conv_nodes, kernel_size = (3,3), activation = 'relu', padding = 'same'),\n",
        "      tf.keras.layers.MaxPooling2D(pool_size = (2,2)),\n",
        "      tf.keras.layers.Dropout(dropout_prob),\n",
        "      tf.keras.layers.Flatten(),\n",
        "      tf.keras.layers.Dense(dense_nodes, activation = 'relu'),\n",
        "      tf.keras.layers.Dropout(rate = 0.2),\n",
        "      tf.keras.layers.Dense(dense_nodes, activation = 'relu'),\n",
        "      tf.keras.layers.Dropout(rate = 0.2),\n",
        "      tf.keras.layers.Dense(2, activation= 'sigmoid'),\n",
        "])\n",
        "\n",
        "  nn_model.compile(optimizer=tf.keras.optimizers.Adam(lr), loss='binary_crossentropy',\n",
        "                  metrics=['accuracy'])\n",
        "  history = nn_model.fit(\n",
        "    X_train, y_train, epochs=epochs, batch_size=batch_size, validation_split=0.2, verbose=0\n",
        "  )\n",
        "\n",
        "  return nn_model, history"
      ],
      "metadata": {
        "id": "kYKih4sF8-sI"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "least_val_loss = float('inf')\n",
        "least_loss_model = None\n",
        "epochs=100\n",
        "for conv_nodes in [32, 64, 128]:\n",
        "  for dense_nodes in [256, 512]:\n",
        "    for dropout_prob in [0, 0.2]:\n",
        "     for lr in [0.01, 0.001, 0.005]:\n",
        "        for batch_size in [32, 64, 128]:\n",
        "            print(f\"conv nodes {conv_nodes}, dense nodes {dense_nodes}, fc_nodes dropout {dropout_prob}, lr {lr}, batch size {batch_size}\")\n",
        "            model, history = train_model(X_train, y_train, conv_nodes, dense_nodes, dropout_prob, lr, batch_size, epochs)\n",
        "            #plot_history(history)\n",
        "            val_loss = model.evaluate(X_val, y_val)[0]\n",
        "            if val_loss < least_val_loss:\n",
        "              least_val_loss = val_loss\n",
        "              least_loss_model = model"
      ],
      "metadata": {
        "id": "xdIReNHW-cwM",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a814b904-b038-4124-9480-6ca4ffc61557"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2884 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1077 - accuracy: 0.9708\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2756 - accuracy: 0.9720\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2884 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.3083 - accuracy: 0.9775\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2305 - accuracy: 0.9720\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1059 - accuracy: 0.9783\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1459 - accuracy: 0.9787\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.0718 - accuracy: 0.9733\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0.2, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2884 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0.2, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0.2, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.0866 - accuracy: 0.9745\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0.2, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0.2, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.0845 - accuracy: 0.9720\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0.2, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.0966 - accuracy: 0.9683\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0.2, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1101 - accuracy: 0.9754\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0.2, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1094 - accuracy: 0.9820\n",
            "conv nodes 32, dense nodes 256, fc_nodes dropout 0.2, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.0955 - accuracy: 0.9816\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2887 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2887 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2884 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2926 - accuracy: 0.9783\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2230 - accuracy: 0.9687\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2424 - accuracy: 0.9770\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1743 - accuracy: 0.9820\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1564 - accuracy: 0.9787\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0.2, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0.2, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2883 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0.2, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.0963 - accuracy: 0.9733\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0.2, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0.2, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.2903 - accuracy: 0.9161\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0.2, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.0688 - accuracy: 0.9779\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0.2, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1415 - accuracy: 0.9795\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0.2, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1486 - accuracy: 0.9741\n",
            "conv nodes 32, dense nodes 512, fc_nodes dropout 0.2, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 3ms/step - loss: 0.1180 - accuracy: 0.9820\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2883 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2883 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1675 - accuracy: 0.9741\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.3328 - accuracy: 0.9758\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2226 - accuracy: 0.9775\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1880 - accuracy: 0.9800\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1718 - accuracy: 0.9708\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2121 - accuracy: 0.9779\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0.2, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0.2, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2884 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0.2, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0.2, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0.2, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.0738 - accuracy: 0.9749\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0.2, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.0911 - accuracy: 0.9779\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0.2, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1207 - accuracy: 0.9762\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0.2, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1268 - accuracy: 0.9816\n",
            "conv nodes 64, dense nodes 256, fc_nodes dropout 0.2, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1400 - accuracy: 0.9791\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2887 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.4551 - accuracy: 0.9144\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.3033 - accuracy: 0.9729\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2381 - accuracy: 0.9775\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2015 - accuracy: 0.9770\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1972 - accuracy: 0.9779\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0.2, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0.2, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2884 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0.2, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2883 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0.2, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0.2, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.0865 - accuracy: 0.9662\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0.2, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.0647 - accuracy: 0.9795\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0.2, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1216 - accuracy: 0.9820\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0.2, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1550 - accuracy: 0.9783\n",
            "conv nodes 64, dense nodes 512, fc_nodes dropout 0.2, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 4ms/step - loss: 0.1112 - accuracy: 0.9800\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 6ms/step - loss: 0.2883 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1404 - accuracy: 0.9666\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2132 - accuracy: 0.9766\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1765 - accuracy: 0.9754\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.3146 - accuracy: 0.9795\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1334 - accuracy: 0.9779\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2657 - accuracy: 0.9775\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0.2, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2884 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0.2, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2888 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0.2, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.0747 - accuracy: 0.9737\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0.2, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2887 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0.2, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0.2, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.0882 - accuracy: 0.9758\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0.2, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1607 - accuracy: 0.9820\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0.2, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1084 - accuracy: 0.9812\n",
            "conv nodes 128, dense nodes 256, fc_nodes dropout 0.2, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1437 - accuracy: 0.9808\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2883 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2885 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2884 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1595 - accuracy: 0.9729\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2587 - accuracy: 0.9800\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2640 - accuracy: 0.9800\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1359 - accuracy: 0.9791\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0.2, lr 0.01, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2883 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0.2, lr 0.01, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0.2, lr 0.01, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2894 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0.2, lr 0.005, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2884 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0.2, lr 0.005, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.2886 - accuracy: 0.9161\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0.2, lr 0.005, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.0909 - accuracy: 0.9699\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0.2, lr 0.001, batch size 32\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1511 - accuracy: 0.9687\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0.2, lr 0.001, batch size 64\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1534 - accuracy: 0.9812\n",
            "conv nodes 128, dense nodes 512, fc_nodes dropout 0.2, lr 0.001, batch size 128\n",
            "75/75 [==============================] - 0s 5ms/step - loss: 0.1529 - accuracy: 0.9795\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model.save('malaria_least_loss_model.h5')"
      ],
      "metadata": {
        "id": "u3bCJ6dNkNty"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from tensorflow.keras.models import load_model\n",
        "\n",
        "model = load_model('malaria_least_loss_model.h5')"
      ],
      "metadata": {
        "id": "2BIdmQfFkRm6"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "y_pred = model.predict(X_test)\n",
        "y_pred = y_pred.astype(int).reshape(-1,)\n",
        "y_test = y_test.astype(int).reshape(-1,)"
      ],
      "metadata": {
        "id": "Q0QK10WtuNnM",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9ca141c2-3e26-4d9e-9fa0-d848d4fa1e2d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "94/94 [==============================] - 1s 5ms/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import classification_report\n",
        "print(classification_report(y_test, y_pred))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EAJLlPpF7uTm",
        "outputId": "20178a96-d129-43a8-8fef-59b232d2976b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.89      1.00      0.94      2993\n",
            "           1       1.00      0.88      0.94      2993\n",
            "\n",
            "    accuracy                           0.94      5986\n",
            "   macro avg       0.94      0.94      0.94      5986\n",
            "weighted avg       0.94      0.94      0.94      5986\n",
            "\n"
          ]
        }
      ]
    }
  ]
}