Multi_label_classification.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "Aim of this task is to train CNN model being able to perform multi-label classification. Dataset contains images of hands imitating gestures used in Rock/Paper/Scissors game. To create final training set, two images and their labels are being selected randomly and contatenated. "
      ],
      "metadata": {
        "id": "7Q5yrA5wj5LP"
      }
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i-HAP41vQ7ja"
      },
      "source": [
        "# 1.Settings"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7QlN1syJRY90"
      },
      "source": [
        "###1.1 Loading data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vypLSgR5ZMCm",
        "outputId": "c53d28ba-d5dc-489a-8d74-2f7f0531d18e"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/gdrive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "_ELxpdC8ZW1j"
      },
      "outputs": [],
      "source": [
        "src_test = \"../rps_test\"\n",
        "src_train = \"../rps_train\""
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5bJyrZ9XRcod"
      },
      "source": [
        "###1.2 Connect to GPU"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lr0ske-9Ed_6",
        "outputId": "9a31d4bc-3997-495f-d8c2-88d5c5a781ec"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found GPU at: /device:GPU:0\n"
          ]
        }
      ],
      "source": [
        "%tensorflow_version 2.x\n",
        "import tensorflow as tf\n",
        "device_name = tf.test.gpu_device_name()\n",
        "if device_name != '/device:GPU:0':\n",
        "  raise SystemError('GPU device not found')\n",
        "print('Found GPU at: {}'.format(device_name))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qcQA_0ztRxq9"
      },
      "source": [
        "###1.3 Imports"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "id": "G_c2B70fCiIP"
      },
      "outputs": [],
      "source": [
        "import os, cv2\n",
        "import numpy as np\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "\n",
        "from sklearn.metrics import classification_report"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qX2mtTdkPe6z"
      },
      "source": [
        "# 2.Data processing"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1qkFyBi8Ps3u"
      },
      "source": [
        "### 2.1 Loading images and creating labels"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tEsifgxlUJD8"
      },
      "source": [
        "Both images and labels were converted into numpy arrays. Labels consists of ones and zeros to enable multi-label classification.\n",
        "\n",
        "Rock -> [ 1 0 0 ],\n",
        "\n",
        "Paper -> [ 0 1 0 ],\n",
        "\n",
        "Scissors -> [ 0 0 1 ]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "lmsp7HaVypx_"
      },
      "outputs": [],
      "source": [
        "def get_data(path):\n",
        "\n",
        "  arrays = []\n",
        "  labels = []\n",
        "  \n",
        "  for img in os.listdir(path):\n",
        "    image = cv2.imread(os.path.join(path, img))\n",
        "    rgb_image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
        "    arrays.append(rgb_image)\n",
        "\n",
        "    if \"rock\" in img:\n",
        "      labels.append(np.array([1,0,0]))\n",
        "    elif \"paper\" in img:\n",
        "      labels.append(np.array([0,1,0]))\n",
        "    elif \"scissor\" in img:\n",
        "      labels.append(np.array([0,0,1]))\n",
        "\n",
        "  return np.array(arrays),np.array(labels)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "JwM0mRq3yxLL"
      },
      "outputs": [],
      "source": [
        "train_im, train_lab = get_data(src_train)\n",
        "test_im, test_lab = get_data(src_test)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "53Z8K2HvVLj5"
      },
      "source": [
        "Sample image and label"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "9pc62hC91BdO",
        "outputId": "af87b8f6-788b-48d9-e3c9-1b2f0617b214"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0 0 1]\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.imshow(test_im[300], cmap=\"gray\")\n",
        "print(test_lab[300])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O5YzNLY1SAcH"
      },
      "source": [
        "### 2.2 Assembling dataset"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SC4cC9A4WCXl"
      },
      "source": [
        "Two images picked at random are being concatenated horizontally.\n",
        "Also their labels are getting merged. If selected images belong to different classes, their label gets multiclass shape like [ 0 1 1 ]. \n",
        "If both images belong to the same class, label remains unchanged --> [ 1 0 0 ]\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "oeR40uf50oV9"
      },
      "outputs": [],
      "source": [
        "def get_concatenated_set(image_set, label_set, n_samples=1000):\n",
        "  images = []\n",
        "  labels = []\n",
        "  \n",
        "  for _ in range (n_samples):\n",
        "    idx_1 = np.random.randint(len(image_set))\n",
        "    idx_2 = np.random.randint(len(image_set))\n",
        "  \n",
        "    image = np.concatenate([image_set[idx_1], image_set[idx_2]], axis=1)\n",
        "    images.append(image)\n",
        "\n",
        "    if (label_set[idx_1] == label_set[idx_2]).all():\n",
        "      labels.append(label_set[idx_1])\n",
        "    else:\n",
        "      label = label_set[idx_1] + label_set[idx_2]\n",
        "      labels.append(label)\n",
        "\n",
        "  return np.array(images), np.array(labels)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OPuJ9ZOub44D"
      },
      "source": [
        "Creating traing and test sets"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "id": "y_wltyxY5Amb"
      },
      "outputs": [],
      "source": [
        "X_train, y_train = get_concatenated_set(train_im, train_lab, n_samples=1000)\n",
        "X_test, y_test = get_concatenated_set(test_im, test_lab, n_samples=1000)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TTSPNkfHYTlq"
      },
      "source": [
        "Sample concatenated image and its label"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 236
        },
        "id": "OktXkFcH5KaQ",
        "outputId": "4066732f-9f90-4690-ea26-9d5c6046cfe8"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0 1 1]\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.imshow(X_test[5])\n",
        "print(y_test[5])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "W3laiZ3eQSua"
      },
      "source": [
        "# 3.Convolutional Neural Network "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f3XPjqmzSfnP"
      },
      "source": [
        "### 3.1 Model Architecture"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uOjiVXcz_Zm5",
        "outputId": "ac537283-2485-45b1-90a1-a32dc96336ce"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " zero_padding2d_1 (ZeroPaddi  (None, 304, 604, 3)      0         \n",
            " ng2D)                                                           \n",
            "                                                                 \n",
            " conv2d_5 (Conv2D)           (None, 300, 600, 32)      2432      \n",
            "                                                                 \n",
            " max_pooling2d_5 (MaxPooling  (None, 150, 300, 32)     0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " dropout_5 (Dropout)         (None, 150, 300, 32)      0         \n",
            "                                                                 \n",
            " conv2d_6 (Conv2D)           (None, 146, 296, 32)      25632     \n",
            "                                                                 \n",
            " max_pooling2d_6 (MaxPooling  (None, 73, 148, 32)      0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " dropout_6 (Dropout)         (None, 73, 148, 32)       0         \n",
            "                                                                 \n",
            " conv2d_7 (Conv2D)           (None, 73, 148, 64)       18496     \n",
            "                                                                 \n",
            " max_pooling2d_7 (MaxPooling  (None, 36, 74, 64)       0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " dropout_7 (Dropout)         (None, 36, 74, 64)        0         \n",
            "                                                                 \n",
            " conv2d_8 (Conv2D)           (None, 36, 74, 64)        36928     \n",
            "                                                                 \n",
            " max_pooling2d_8 (MaxPooling  (None, 18, 37, 64)       0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " dropout_8 (Dropout)         (None, 18, 37, 64)        0         \n",
            "                                                                 \n",
            " conv2d_9 (Conv2D)           (None, 18, 37, 128)       73856     \n",
            "                                                                 \n",
            " max_pooling2d_9 (MaxPooling  (None, 9, 18, 128)       0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " conv2d_10 (Conv2D)          (None, 9, 18, 256)        33024     \n",
            "                                                                 \n",
            " max_pooling2d_10 (MaxPoolin  (None, 4, 9, 256)        0         \n",
            " g2D)                                                            \n",
            "                                                                 \n",
            " dropout_9 (Dropout)         (None, 4, 9, 256)         0         \n",
            "                                                                 \n",
            " flatten_1 (Flatten)         (None, 9216)              0         \n",
            "                                                                 \n",
            " dropout_10 (Dropout)        (None, 9216)              0         \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 3)                 27651     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 218,019\n",
            "Trainable params: 218,019\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model = keras.models.Sequential()\n",
        "model.add(keras.layers.InputLayer(input_shape=[300,600,3]))\n",
        "model.add(keras.layers.ZeroPadding2D(padding=2))\n",
        "model.add(keras.layers.Conv2D(filters=32, kernel_size=(5,5), data_format='channels_last'))\n",
        "model.add(keras.layers.MaxPooling2D())\n",
        "model.add(keras.layers.Dropout(0.1))\n",
        "\n",
        "model.add(keras.layers.Conv2D(filters=32, kernel_size=(5,5), data_format='channels_last'))\n",
        "model.add(keras.layers.MaxPooling2D())\n",
        "model.add(keras.layers.Dropout(0.1))\n",
        "\n",
        "model.add(keras.layers.Conv2D(filters=64, kernel_size=(3,3), data_format='channels_last', padding=\"same\"))\n",
        "model.add(keras.layers.MaxPooling2D())\n",
        "model.add(keras.layers.Dropout(0.1))\n",
        "\n",
        "model.add(keras.layers.Conv2D(filters=64, kernel_size=(3,3), data_format='channels_last', padding=\"same\"))\n",
        "model.add(keras.layers.MaxPooling2D())\n",
        "model.add(keras.layers.Dropout(0.1))\n",
        "\n",
        "model.add(keras.layers.Conv2D(filters=128, kernel_size=(3,3), data_format='channels_last', padding=\"same\"))\n",
        "model.add(keras.layers.MaxPooling2D())\n",
        "\n",
        "model.add(keras.layers.Conv2D(filters=256, kernel_size=(1,1), data_format='channels_last', padding=\"same\"))\n",
        "model.add(keras.layers.MaxPooling2D())\n",
        "model.add(keras.layers.Dropout(0.1))\n",
        "\n",
        "model.add(keras.layers.Flatten())\n",
        "model.add(keras.layers.Dropout(0.2))\n",
        "model.add(keras.layers.Dense(3, activation=\"sigmoid\"))\n",
        "\n",
        "model.summary()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Skbg4ZR-S-09"
      },
      "source": [
        "### 3.2 Model compilation"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "start = 0.0001\n",
        "min = 0.00001\n",
        "max = 0.001\n",
        "rampup_epochs = 10\n",
        "sustain_epochs = 10\n",
        "decay = .8\n",
        "\n",
        "def lrfn(epoch):\n",
        "  if epoch < rampup_epochs:\n",
        "    return (max - start)/rampup_epochs * epoch + start\n",
        "  elif epoch < rampup_epochs + sustain_epochs:\n",
        "    return max\n",
        "  else:\n",
        "    return min\n",
        "    \n",
        "lr = keras.callbacks.LearningRateScheduler(lambda epoch: lrfn(epoch), verbose=True)"
      ],
      "metadata": {
        "id": "cEsg_1-JYGFO"
      },
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "id": "98NpkylZ8XOe"
      },
      "outputs": [],
      "source": [
        "model.compile(loss=\"binary_crossentropy\",\n",
        "             optimizer=Adam(),\n",
        "             metrics=[\"accuracy\"])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y9D0dO0ITGQ4"
      },
      "source": [
        "### 3.3 Trening"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kpQ9RIUf97Hh",
        "outputId": "da2aca40-9a9f-4138-835e-121417858e13"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Epoch 00001: LearningRateScheduler setting learning rate to 0.0001.\n",
            "Epoch 1/40\n",
            "27/27 [==============================] - 5s 184ms/step - loss: 3.8572 - accuracy: 0.3635 - val_loss: 6.8951 - val_accuracy: 0.5533 - lr: 1.0000e-04\n",
            "\n",
            "Epoch 00002: LearningRateScheduler setting learning rate to 0.00019.\n",
            "Epoch 2/40\n",
            "27/27 [==============================] - 5s 183ms/step - loss: 3.2609 - accuracy: 0.4106 - val_loss: 3.5367 - val_accuracy: 0.3933 - lr: 1.9000e-04\n",
            "\n",
            "Epoch 00003: LearningRateScheduler setting learning rate to 0.00028.\n",
            "Epoch 3/40\n",
            "27/27 [==============================] - 5s 183ms/step - loss: 2.8718 - accuracy: 0.4224 - val_loss: 4.1585 - val_accuracy: 0.4333 - lr: 2.8000e-04\n",
            "\n",
            "Epoch 00004: LearningRateScheduler setting learning rate to 0.00036999999999999994.\n",
            "Epoch 4/40\n",
            "27/27 [==============================] - 5s 195ms/step - loss: 2.3070 - accuracy: 0.4741 - val_loss: 2.8309 - val_accuracy: 0.3600 - lr: 3.7000e-04\n",
            "\n",
            "Epoch 00005: LearningRateScheduler setting learning rate to 0.00045999999999999996.\n",
            "Epoch 5/40\n",
            "27/27 [==============================] - 5s 183ms/step - loss: 1.8907 - accuracy: 0.4941 - val_loss: 0.8043 - val_accuracy: 0.6400 - lr: 4.6000e-04\n",
            "\n",
            "Epoch 00006: LearningRateScheduler setting learning rate to 0.00055.\n",
            "Epoch 6/40\n",
            "27/27 [==============================] - 5s 184ms/step - loss: 1.4790 - accuracy: 0.5600 - val_loss: 0.9661 - val_accuracy: 0.6133 - lr: 5.5000e-04\n",
            "\n",
            "Epoch 00007: LearningRateScheduler setting learning rate to 0.0006399999999999999.\n",
            "Epoch 7/40\n",
            "27/27 [==============================] - 5s 184ms/step - loss: 1.1765 - accuracy: 0.5976 - val_loss: 0.6284 - val_accuracy: 0.7867 - lr: 6.4000e-04\n",
            "\n",
            "Epoch 00008: LearningRateScheduler setting learning rate to 0.00073.\n",
            "Epoch 8/40\n",
            "27/27 [==============================] - 5s 195ms/step - loss: 0.7618 - accuracy: 0.6706 - val_loss: 0.4253 - val_accuracy: 0.7000 - lr: 7.3000e-04\n",
            "\n",
            "Epoch 00009: LearningRateScheduler setting learning rate to 0.00082.\n",
            "Epoch 9/40\n",
            "27/27 [==============================] - 5s 184ms/step - loss: 0.7028 - accuracy: 0.6529 - val_loss: 0.7958 - val_accuracy: 0.8067 - lr: 8.2000e-04\n",
            "\n",
            "Epoch 00010: LearningRateScheduler setting learning rate to 0.00091.\n",
            "Epoch 10/40\n",
            "27/27 [==============================] - 5s 183ms/step - loss: 0.6205 - accuracy: 0.6965 - val_loss: 0.3409 - val_accuracy: 0.7067 - lr: 9.1000e-04\n",
            "\n",
            "Epoch 00011: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 11/40\n",
            "27/27 [==============================] - 5s 184ms/step - loss: 0.5969 - accuracy: 0.7129 - val_loss: 0.4626 - val_accuracy: 0.6333 - lr: 0.0010\n",
            "\n",
            "Epoch 00012: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 12/40\n",
            "27/27 [==============================] - 5s 184ms/step - loss: 0.4358 - accuracy: 0.7235 - val_loss: 0.3088 - val_accuracy: 0.6000 - lr: 0.0010\n",
            "\n",
            "Epoch 00013: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 13/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.4613 - accuracy: 0.7271 - val_loss: 0.4554 - val_accuracy: 0.6733 - lr: 0.0010\n",
            "\n",
            "Epoch 00014: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 14/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.4828 - accuracy: 0.6941 - val_loss: 0.2362 - val_accuracy: 0.6067 - lr: 0.0010\n",
            "\n",
            "Epoch 00015: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 15/40\n",
            "27/27 [==============================] - 5s 187ms/step - loss: 0.3287 - accuracy: 0.6894 - val_loss: 0.1779 - val_accuracy: 0.6467 - lr: 0.0010\n",
            "\n",
            "Epoch 00016: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 16/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.2341 - accuracy: 0.7035 - val_loss: 0.2179 - val_accuracy: 0.6333 - lr: 0.0010\n",
            "\n",
            "Epoch 00017: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 17/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.1474 - accuracy: 0.7353 - val_loss: 0.1428 - val_accuracy: 0.6400 - lr: 0.0010\n",
            "\n",
            "Epoch 00018: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 18/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.1258 - accuracy: 0.6976 - val_loss: 0.3002 - val_accuracy: 0.6067 - lr: 0.0010\n",
            "\n",
            "Epoch 00019: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 19/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.1478 - accuracy: 0.6941 - val_loss: 0.1981 - val_accuracy: 0.6267 - lr: 0.0010\n",
            "\n",
            "Epoch 00020: LearningRateScheduler setting learning rate to 0.001.\n",
            "Epoch 20/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.1194 - accuracy: 0.7188 - val_loss: 0.1254 - val_accuracy: 0.7467 - lr: 0.0010\n",
            "\n",
            "Epoch 00021: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 21/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0599 - accuracy: 0.7459 - val_loss: 0.1259 - val_accuracy: 0.7400 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00022: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 22/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.0609 - accuracy: 0.7576 - val_loss: 0.1300 - val_accuracy: 0.7267 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00023: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 23/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.0746 - accuracy: 0.7400 - val_loss: 0.1290 - val_accuracy: 0.7200 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00024: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 24/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0556 - accuracy: 0.7365 - val_loss: 0.1272 - val_accuracy: 0.7200 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00025: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 25/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0469 - accuracy: 0.7388 - val_loss: 0.1285 - val_accuracy: 0.7200 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00026: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 26/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.0519 - accuracy: 0.7318 - val_loss: 0.1311 - val_accuracy: 0.7200 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00027: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 27/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0504 - accuracy: 0.7365 - val_loss: 0.1300 - val_accuracy: 0.7067 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00028: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 28/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0556 - accuracy: 0.7506 - val_loss: 0.1293 - val_accuracy: 0.7133 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00029: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 29/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0430 - accuracy: 0.7447 - val_loss: 0.1224 - val_accuracy: 0.7133 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00030: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 30/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0464 - accuracy: 0.7318 - val_loss: 0.1250 - val_accuracy: 0.7133 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00031: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 31/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0368 - accuracy: 0.7459 - val_loss: 0.1316 - val_accuracy: 0.7133 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00032: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 32/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0452 - accuracy: 0.7224 - val_loss: 0.1292 - val_accuracy: 0.7133 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00033: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 33/40\n",
            "27/27 [==============================] - 5s 186ms/step - loss: 0.0356 - accuracy: 0.7271 - val_loss: 0.1279 - val_accuracy: 0.7133 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00034: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 34/40\n",
            "27/27 [==============================] - 5s 185ms/step - loss: 0.0507 - accuracy: 0.7329 - val_loss: 0.1244 - val_accuracy: 0.7067 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00035: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 35/40\n",
            "27/27 [==============================] - 5s 187ms/step - loss: 0.0314 - accuracy: 0.7212 - val_loss: 0.1264 - val_accuracy: 0.7067 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00036: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 36/40\n",
            "27/27 [==============================] - 5s 188ms/step - loss: 0.0465 - accuracy: 0.7294 - val_loss: 0.1232 - val_accuracy: 0.7067 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00037: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 37/40\n",
            "27/27 [==============================] - 5s 188ms/step - loss: 0.0517 - accuracy: 0.7318 - val_loss: 0.1212 - val_accuracy: 0.7067 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00038: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 38/40\n",
            "27/27 [==============================] - 5s 187ms/step - loss: 0.0494 - accuracy: 0.7388 - val_loss: 0.1225 - val_accuracy: 0.7067 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00039: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 39/40\n",
            "27/27 [==============================] - 5s 188ms/step - loss: 0.0454 - accuracy: 0.7365 - val_loss: 0.1247 - val_accuracy: 0.7000 - lr: 1.0000e-05\n",
            "\n",
            "Epoch 00040: LearningRateScheduler setting learning rate to 1e-05.\n",
            "Epoch 40/40\n",
            "27/27 [==============================] - 5s 187ms/step - loss: 0.0285 - accuracy: 0.7341 - val_loss: 0.1266 - val_accuracy: 0.7067 - lr: 1.0000e-05\n"
          ]
        }
      ],
      "source": [
        "history = model.fit(X_train, y_train, epochs=40,\n",
        "                   batch_size=32, validation_split=0.15,\n",
        "                    shuffle=True, callbacks=[lr])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KPvPP4-BTNLk"
      },
      "source": [
        "### 3.4 Loss function"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "ZV3LnDTXMJg4",
        "outputId": "bb3dcff5-e789-4316-e839-9c0a9d7045bf"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HCr_ZATeTTk1"
      },
      "source": [
        "### 3.5 Model aacuracy"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "ZVWMpU0a-M5O",
        "outputId": "d6e4e336-bfb8-484f-c04a-4a88d9aa016e"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cENBXqgdTvsn"
      },
      "source": [
        "# 4.Results"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 25,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FRlnb7b4MB_8",
        "outputId": "5932237f-538f-4c39-e96a-24af5cc2ec8a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.87      0.99      0.92       536\n",
            "           1       0.94      0.76      0.84       556\n",
            "           2       0.81      0.88      0.84       571\n",
            "\n",
            "   micro avg       0.86      0.87      0.87      1663\n",
            "   macro avg       0.87      0.87      0.87      1663\n",
            "weighted avg       0.87      0.87      0.87      1663\n",
            " samples avg       0.89      0.89      0.87      1663\n",
            "\n"
          ]
        }
      ],
      "source": [
        "y_pred = model.predict(X_test)\n",
        "y_pred = np.rint(y_pred)\n",
        "\n",
        "print(classification_report(y_test, y_pred))"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "name": "Multi-label classification.ipynb",
      "provenance": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}