2. CNN from scratch.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"2. CNN from scratch.ipynb","provenance":[],"collapsed_sections":["e72dqDQqdrSN","cOl-TkG_dzP8","5HUo3t9Oex7z","q_GNavPeguxx","mwIeet0C4YP2","zRQbZde9FYMQ"],"toc_visible":true},"kernelspec":{"name":"python3","display_name":"Python 3"},"language_info":{"name":"python"},"widgets":{"application/vnd.jupyter.widget-state+json":{"de026197934345d8b005dfd727df4a9c":{"model_module":"@jupyter-widgets/controls","model_name":"HBoxModel","model_module_version":"1.5.0","state":{"_view_name":"HBoxView","_dom_classes":[],"_model_name":"HBoxModel","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.5.0","box_style":"","layout":"IPY_MODEL_f29a3a9dc3374f78b19fc06f49dd6c1e","_model_module":"@jupyter-widgets/controls","children":["IPY_MODEL_53d1738c7db94516a101771272b65319","IPY_MODEL_817faed4398b408e98ed9a7e2fba34d2","IPY_MODEL_84bc22506c0c427888dc08ae9a186f56"]}},"f29a3a9dc3374f78b19fc06f49dd6c1e":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","model_module_version":"1.2.0","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"53d1738c7db94516a101771272b65319":{"model_module":"@jupyter-widgets/controls","model_name":"HTMLModel","model_module_version":"1.5.0","state":{"_view_name":"HTMLView","style":"IPY_MODEL_f72ddc3cd38a4be9b4953f1ceeefb080","_dom_classes":[],"description":"","_model_name":"HTMLModel","placeholder":"​","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":"100%","_view_count":null,"_view_module_version":"1.5.0","description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_2d1ddae54e8940089ba04a31b72a62a2"}},"817faed4398b408e98ed9a7e2fba34d2":{"model_module":"@jupyter-widgets/controls","model_name":"FloatProgressModel","model_module_version":"1.5.0","state":{"_view_name":"ProgressView","style":"IPY_MODEL_491579824071429cb02f0b2328642669","_dom_classes":[],"description":"","_model_name":"FloatProgressModel","bar_style":"success","max":3198,"_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":3198,"_view_count":null,"_view_module_version":"1.5.0","orientation":"horizontal","min":0,"description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_9e0575c87dd04d2d93c011a0f7378729"}},"84bc22506c0c427888dc08ae9a186f56":{"model_module":"@jupyter-widgets/controls","model_name":"HTMLModel","model_module_version":"1.5.0","state":{"_view_name":"HTMLView","style":"IPY_MODEL_101d3173b4c34e2ba821cbc193826960","_dom_classes":[],"description":"","_model_name":"HTMLModel","placeholder":"​","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":" 3198/3198 [00:00&lt;00:00, 5915.96it/s]","_view_count":null,"_view_module_version":"1.5.0","description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_524efdf6e1264b68ac139f1714a44876"}},"f72ddc3cd38a4be9b4953f1ceeefb080":{"model_module":"@jupyter-widgets/controls","model_name":"DescriptionStyleModel","model_module_version":"1.5.0","state":{"_view_name":"StyleView","_model_name":"DescriptionStyleModel","description_width":"","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","_model_module":"@jupyter-widgets/controls"}},"2d1ddae54e8940089ba04a31b72a62a2":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","model_module_version":"1.2.0","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"491579824071429cb02f0b2328642669":{"model_module":"@jupyter-widgets/controls","model_name":"ProgressStyleModel","model_module_version":"1.5.0","state":{"_view_name":"StyleView","_model_name":"ProgressStyleModel","description_width":"","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","bar_color":null,"_model_module":"@jupyter-widgets/controls"}},"9e0575c87dd04d2d93c011a0f7378729":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","model_module_version":"1.2.0","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"101d3173b4c34e2ba821cbc193826960":{"model_module":"@jupyter-widgets/controls","model_name":"DescriptionStyleModel","model_module_version":"1.5.0","state":{"_view_name":"StyleView","_model_name":"DescriptionStyleModel","description_width":"","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","_model_module":"@jupyter-widgets/controls"}},"524efdf6e1264b68ac139f1714a44876":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","model_module_version":"1.2.0","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"cdddbf7ae6a24654874591dab870821c":{"model_module":"@jupyter-widgets/controls","model_name":"HBoxModel","model_module_version":"1.5.0","state":{"_view_name":"HBoxView","_dom_classes":[],"_model_name":"HBoxModel","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.5.0","box_style":"","layout":"IPY_MODEL_68410e2358a94c51a93067508ad1428d","_model_module":"@jupyter-widgets/controls","children":["IPY_MODEL_889afabe18af4e3691c6664e30dfa99e","IPY_MODEL_e1d661c1c8884e8f98d92cbac7d8fb12","IPY_MODEL_76b3d5a439554a1094a3d7b73585292c"]}},"68410e2358a94c51a93067508ad1428d":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","model_module_version":"1.2.0","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"889afabe18af4e3691c6664e30dfa99e":{"model_module":"@jupyter-widgets/controls","model_name":"HTMLModel","model_module_version":"1.5.0","state":{"_view_name":"HTMLView","style":"IPY_MODEL_9607820cb56e46b1bfc582ef4153d840","_dom_classes":[],"description":"","_model_name":"HTMLModel","placeholder":"​","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":"100%","_view_count":null,"_view_module_version":"1.5.0","description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_075429fc77b44f9c9310e6170726d0d7"}},"e1d661c1c8884e8f98d92cbac7d8fb12":{"model_module":"@jupyter-widgets/controls","model_name":"FloatProgressModel","model_module_version":"1.5.0","state":{"_view_name":"ProgressView","style":"IPY_MODEL_28c2df7a971f4f0f820f07d0267eb7f9","_dom_classes":[],"description":"","_model_name":"FloatProgressModel","bar_style":"success","max":3198,"_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":3198,"_view_count":null,"_view_module_version":"1.5.0","orientation":"horizontal","min":0,"description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_4c53b520e7714d59a49027c4704199f4"}},"76b3d5a439554a1094a3d7b73585292c":{"model_module":"@jupyter-widgets/controls","model_name":"HTMLModel","model_module_version":"1.5.0","state":{"_view_name":"HTMLView","style":"IPY_MODEL_47663ad77f284abca4cb2ae1b19c6af1","_dom_classes":[],"description":"","_model_name":"HTMLModel","placeholder":"​","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":" 3198/3198 [00:00&lt;00:00, 7606.77it/s]","_view_count":null,"_view_module_version":"1.5.0","description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_f76658d49a5c4d5ab4053a8d44dd1cb4"}},"9607820cb56e46b1bfc582ef4153d840":{"model_module":"@jupyter-widgets/controls","model_name":"DescriptionStyleModel","model_module_version":"1.5.0","state":{"_view_name":"StyleView","_model_name":"DescriptionStyleModel","description_width":"","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","_model_module":"@jupyter-widgets/controls"}},"075429fc77b44f9c9310e6170726d0d7":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","model_module_version":"1.2.0","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"28c2df7a971f4f0f820f07d0267eb7f9":{"model_module":"@jupyter-widgets/controls","model_name":"ProgressStyleModel","model_module_version":"1.5.0","state":{"_view_name":"StyleView","_model_name":"ProgressStyleModel","description_width":"","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","bar_color":null,"_model_module":"@jupyter-widgets/controls"}},"4c53b520e7714d59a49027c4704199f4":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","model_module_version":"1.2.0","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"47663ad77f284abca4cb2ae1b19c6af1":{"model_module":"@jupyter-widgets/controls","model_name":"DescriptionStyleModel","model_module_version":"1.5.0","state":{"_view_name":"StyleView","_model_name":"DescriptionStyleModel","description_width":"","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","_model_module":"@jupyter-widgets/controls"}},"f76658d49a5c4d5ab4053a8d44dd1cb4":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","model_module_version":"1.2.0","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}}}},"accelerator":"GPU"},"cells":[{"cell_type":"markdown","source":["# Import libraries"],"metadata":{"id":"e72dqDQqdrSN"}},{"cell_type":"code","execution_count":1,"metadata":{"id":"ZjeQZaSZEYnl","executionInfo":{"status":"ok","timestamp":1642021637279,"user_tz":-60,"elapsed":4049,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}}},"outputs":[],"source":["from google.colab import drive\n","from google.colab.patches import cv2_imshow\n","import numpy as np\n","import cv2\n","import os\n","from tqdm.auto import tqdm\n","import tensorflow as tf\n","from tensorflow import keras\n","from tensorflow.keras import layers\n","from tensorflow.keras import regularizers\n","from tensorflow.keras import optimizers\n","from tensorflow.keras.utils import image_dataset_from_directory\n","import matplotlib.pyplot as plt\n","import shutil\n","import random as rn\n","from sklearn import metrics\n","from sklearn.metrics import confusion_matrix"]},{"cell_type":"markdown","source":["# Load dataset from Google Drive\n","The dataset is stored in Google Drive in a zip file, we unzip it locally for better performance."],"metadata":{"id":"cOl-TkG_dzP8"}},{"cell_type":"code","source":["drive.mount('/content/drive', force_remount=True)\n","!unzip -q '/content/drive/My Drive/Baldi_Cioffo/SplittedDataset.zip' -d '/content'"],"metadata":{"id":"07Mbdhrnd4Tr","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1642021734931,"user_tz":-60,"elapsed":95576,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"380d16ae-755e-496b-a18f-1877abfea102"},"execution_count":2,"outputs":[{"output_type":"stream","name":"stdout","text":["Mounted at /content/drive\n"]}]},{"cell_type":"markdown","source":["# Set parameters"],"metadata":{"id":"5HUo3t9Oex7z"}},{"cell_type":"code","source":["DATASET_PATH = \"/content/SplittedDataset\"\n","TRAINING_PATH = DATASET_PATH + '/training_set'\n","TRAINING_ALL_PATH = TRAINING_PATH + '/all'\n","TRAINING_HEM_PATH = TRAINING_PATH + '/hem'\n","VALIDATION_PATH = DATASET_PATH + '/validation_set'\n","TEST_PATH = DATASET_PATH + '/test_set'\n","\n","LOCAL_MODELS_FOLDER = \"/content/models\"\n","if not os.path.exists(LOCAL_MODELS_FOLDER):\n","  os.makedirs(LOCAL_MODELS_FOLDER)\n","GLOBAL_MODELS_FOLDER = \"/content/drive/My Drive/Baldi_Cioffo/models/from_scratch\"\n","if not os.path.exists(GLOBAL_MODELS_FOLDER):\n","  os.makedirs(GLOBAL_MODELS_FOLDER)\n","\n","# Default values to load the dataset\n","BATCH_SIZE = 32\n","IMAGE_HEIGHT = 300\n","IMAGE_WIDTH = 300"],"metadata":{"id":"afWiK7ODe0Ae","executionInfo":{"status":"ok","timestamp":1642021738359,"user_tz":-60,"elapsed":24,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}}},"execution_count":3,"outputs":[]},{"cell_type":"markdown","source":["# Utility functions"],"metadata":{"id":"q_GNavPeguxx"}},{"cell_type":"code","source":["def set_seed ():\n","  ''' \n","  set_seed is used to obtain reproducible results using keras during the development phase\n","  '''\n","  seed = 24\n","  # The below is necessary for reproducible results of certain Python hash-based operations.\n","  os.environ[\"PYTHONHASHSEED\"]=\"0\"\n","  # The below is necessary for starting Numpy generated random numbers in a well-defined initial state.\n","  np.random.seed(seed)\n","  # The below is necessary for starting core Python generated random numbers in a well-defined state.\n","  rn.seed(seed)\n","  # The below tf.random.set_seed will make random number generation in TensorFlow have a well-defined initial state.\n","  tf.random.set_seed(seed)\n","\n","def show_training_and_validation_performance(history):\n","  '''\n","  show_training_and_validation_performance is used to plot the performances during the training phase\n","  :param history: object in which are recorded all the events\n","  '''\n","  acc = history.history['accuracy']\n","  val_acc = history.history['val_accuracy']\n","  loss = history.history['loss']\n","  val_loss = history.history['val_loss']\n","\n","  epochs = range(len(acc))\n","\n","  plt.plot(epochs, acc, 'bo', label='Training accuracy')\n","  plt.plot(epochs, val_acc, 'b', label='Validation accuracy')\n","  plt.title('Training and validation accuracy')\n","  plt.legend()\n","\n","  plt.figure()\n","\n","  plt.plot(epochs, loss, 'bo', label='Training loss')\n","  plt.plot(epochs, val_loss, 'b', label='Validation loss')\n","  plt.title('Training and validation loss')\n","  plt.legend()\n","\n","  plt.show()\n","\n","def load_training_set (image_height, image_width, batch_size):\n","  '''\n","  load_training_set loads the training set from the \"training_set\" folder. The images are resized to height x width\n","  :param image_height: standard height of the images\n","  :param image_width: standard width of the images\n","  :param batch_size: size of each batch\n","  :return: the training set\n","  '''\n","  return image_dataset_from_directory(\n","    TRAINING_PATH,\n","    image_size = (image_height, image_width), # 300x300\n","    batch_size = batch_size,\n","    class_names = ['hem', 'all']) # explicit ordering of the labels, 0 for hem, 1 for all \n","\n","def load_validation_set (image_height, image_width, batch_size):\n","  '''\n","  load_validation_set loads the validation set from the \"validation_set\" folder. The images are resized to height x width\n","  :param image_height: standard height of the images\n","  :param image_width: standard width of the images\n","  :param batch_size: size of each batch\n","  :return: the validation set\n","  '''\n","  return image_dataset_from_directory(\n","    VALIDATION_PATH,\n","    image_size = (image_height, image_width),\n","    batch_size = batch_size,\n","    class_names = ['hem', 'all']) # explicit ordering of the labels, 0 for hem, 1 for all \n","\n","def load_test_set (image_height, image_width, batch_size):\n","  '''\n","  load_test_set loads the test set from the \"test_set\" folder. The images are resized to height x width\n","  :param image_height: standard height of the images\n","  :param image_width: standard width of the images\n","  :param batch_size: size of each batch\n","  :return: the test set\n","  '''\n","  return image_dataset_from_directory(\n","    TEST_PATH,\n","    image_size = (image_height, image_width),\n","    batch_size = batch_size,\n","    shuffle=False,\n","    class_names = ['hem', 'all']) # explicit ordering of the labels, 0 for hem, 1 for all\n","\n","def compile_model (model, optimizer='adam', learning_rate = 0.001):\n","  '''\n","  compile_model is used to compile the current model\n","  :param model: model to compile\n","  :param optimizer: optimizer to be used\n","  :param learning_rate: learning rate parameter for the optimizer\n","  '''\n","  if optimizer == 'adam':\n","    model.compile(loss=\"binary_crossentropy\",\n","      optimizer=optimizers.Adam(learning_rate=learning_rate),\n","      metrics=[\"accuracy\"])\n","  elif optimizer == 'rmsprop':\n","    model.compile(loss=\"binary_crossentropy\",\n","                  optimizer = optimizers.RMSprop(learning_rate=learning_rate),\n","                  metrics=['accuracy'])\n","  model.summary()\n","\n","def run_model (model, model_name, epochs = 20, patience=5, monitor='val_loss'):\n","  '''\n","  run_model is used to run the current mode\n","  :param model: model to run\n","  :param model_name: name given to save the model\n","  :param epochs: how many epochs to do\n","  :param patience: patience value for Early Stopping\n","  :param monitor: what to monitor for Early Stopping and Model Checkpoint\n","  '''\n","  # local save path for the models\n","  save_path = LOCAL_MODELS_FOLDER + '/' + model_name + '.h5'\n","  callbacks_list = [\n","                  keras.callbacks.EarlyStopping(monitor=monitor, patience=patience),\n","                  keras.callbacks.ModelCheckpoint(\n","                      filepath = save_path,\n","                      monitor=monitor,\n","                      verbose=1,\n","                      save_best_only=True)\n","                  ]\n","  history = model.fit(train_dataset,\n","                    epochs=epochs,\n","                    validation_data=validation_dataset,\n","                    callbacks=callbacks_list)\n","  # save on Drive only the best model\n","  shutil.copy(save_path, GLOBAL_MODELS_FOLDER + '/' + model_name + '.h5')\n","  show_training_and_validation_performance(history)\n","\n","def evaluate_model (model):\n","  '''\n","  evaluate_model is used to plot some statistics about the performance on the test set\n","  :param model: model to consider\n","  '''\n","  y_score = model.predict(test_dataset)\n","  y_pred = np.rint(y_score) # to have 0 or 1\n","  y_true = tf.concat([labels_batch for data_batch, labels_batch in test_dataset], axis = 0)\n","  print(\"Classification report: \")\n","  print(metrics.classification_report(y_true,y_pred,digits = 4))\n","  metrics.ConfusionMatrixDisplay.from_predictions(y_true, y_pred)\n","\n","  # ROC curve\n","  fpr,tpr,th = metrics.roc_curve(y_true,y_score)\n","  roc_auc = metrics.roc_auc_score(y_true,y_score)\n","\n","  plt.figure()\n","  plt.plot(fpr, tpr, label='ROC curve (area = %0.2f)' % roc_auc)\n","  plt.plot([0, 1], [0, 1], color='navy', linestyle='--')\n","  plt.xlim([0.0, 1.0])\n","  plt.ylim([0.0, 1.0])\n","  plt.xlabel('False Positive Rate')\n","  plt.ylabel('True Positive Rate')\n","  plt.title('ROC curve')\n","  plt.legend(loc=\"lower right\")\n","  plt.show()"],"metadata":{"id":"fSX3ug4Cgwww","executionInfo":{"status":"ok","timestamp":1642021738715,"user_tz":-60,"elapsed":374,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}}},"execution_count":4,"outputs":[]},{"cell_type":"markdown","source":["# Analyze shapes of the images in the training set\n","After the preprocessing phase, we have images of different shapes, we need to find the best size to feed into the neural network. To do this, we decided to analyze the shapes of our training pictures."],"metadata":{"id":"mwIeet0C4YP2"}},{"cell_type":"code","source":["shapes = {} \n","\n","total_entries = []\n","entries = os.listdir(TRAINING_ALL_PATH)\n","total_entries += entries\n","for entry in tqdm(entries):\n","  img = cv2.imread(TRAINING_ALL_PATH + '/' + entry)\n","  if img.shape <= (300, 300):\n","    shapes[(300,300)] = shapes.get((300,300), 0) + 1\n","  if img.shape <= (299, 299): # like Inception and Xception\n","    shapes[(299,299)] = shapes.get((299,299), 0) + 1\n","  if img.shape <= (224, 224): # like ResNet, DenseNet, VGG, EfficientNet\n","    shapes[(224,224)] = shapes.get((224,224), 0) + 1\n","entries = os.listdir(TRAINING_HEM_PATH)\n","total_entries += entries\n","for entry in tqdm(entries):\n","  img = cv2.imread(TRAINING_HEM_PATH + '/' + entry)\n","  if img.shape <= (300, 300):\n","    shapes[(300,300)] = shapes.get((300,300), 0) + 1\n","  if img.shape <= (299, 299):\n","    shapes[(299,299)] = shapes.get((299,299), 0) + 1\n","  if img.shape <= (224, 224):\n","    shapes[(224,224)] = shapes.get((224,224), 0) + 1"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":81,"referenced_widgets":["de026197934345d8b005dfd727df4a9c","f29a3a9dc3374f78b19fc06f49dd6c1e","53d1738c7db94516a101771272b65319","817faed4398b408e98ed9a7e2fba34d2","84bc22506c0c427888dc08ae9a186f56","f72ddc3cd38a4be9b4953f1ceeefb080","2d1ddae54e8940089ba04a31b72a62a2","491579824071429cb02f0b2328642669","9e0575c87dd04d2d93c011a0f7378729","101d3173b4c34e2ba821cbc193826960","524efdf6e1264b68ac139f1714a44876","cdddbf7ae6a24654874591dab870821c","68410e2358a94c51a93067508ad1428d","889afabe18af4e3691c6664e30dfa99e","e1d661c1c8884e8f98d92cbac7d8fb12","76b3d5a439554a1094a3d7b73585292c","9607820cb56e46b1bfc582ef4153d840","075429fc77b44f9c9310e6170726d0d7","28c2df7a971f4f0f820f07d0267eb7f9","4c53b520e7714d59a49027c4704199f4","47663ad77f284abca4cb2ae1b19c6af1","f76658d49a5c4d5ab4053a8d44dd1cb4"]},"id":"sRwGzwo_4e4t","executionInfo":{"status":"ok","timestamp":1640181706216,"user_tz":-60,"elapsed":1313,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"375022ba-b18d-484b-a3c9-cd58a577a256"},"execution_count":null,"outputs":[{"output_type":"display_data","data":{"application/vnd.jupyter.widget-view+json":{"model_id":"de026197934345d8b005dfd727df4a9c","version_minor":0,"version_major":2},"text/plain":["  0%|          | 0/3198 [00:00<?, ?it/s]"]},"metadata":{}},{"output_type":"display_data","data":{"application/vnd.jupyter.widget-view+json":{"model_id":"cdddbf7ae6a24654874591dab870821c","version_minor":0,"version_major":2},"text/plain":["  0%|          | 0/3198 [00:00<?, ?it/s]"]},"metadata":{}}]},{"cell_type":"code","source":["print(\"Number of training images: \", len(total_entries))\n","print(\"Common shapes: \", shapes)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"GEOw086Z85qD","executionInfo":{"status":"ok","timestamp":1640182224011,"user_tz":-60,"elapsed":298,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"c2b9e012-554f-4afa-bf45-5efe319387c5"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Number of training images:  6396\n","Common shapes:  {(300, 300): 6190, (299, 299): 6180, (224, 224): 3856}\n"]}]},{"cell_type":"markdown","source":["Out of 6396 images in the training set, 6190 are at most 300x300. We decided that 300x300 is a good shape for the input layer of the neural network."],"metadata":{"id":"jQs64Si5Bika"}},{"cell_type":"markdown","source":["# CNN Implementation\n","In the next sections we are going to perform various experiments. Many architectures have been tried, but only the bests have been reported. All reported models follow the same convolutional scheme, with increasing max pooling. The differences can be seen in the final levels of the networks, where the classification on the features extracted from the images actually takes place."],"metadata":{"id":"6OmTEyJ2Iy2A"}},{"cell_type":"markdown","source":["## Load dataset"],"metadata":{"id":"ODZjrkokWip5"}},{"cell_type":"code","source":["set_seed()\n","\n","train_dataset = load_training_set(IMAGE_HEIGHT, IMAGE_WIDTH, BATCH_SIZE) # with default values\n","validation_dataset = load_validation_set(IMAGE_HEIGHT, IMAGE_WIDTH, BATCH_SIZE) # with default values\n","test_dataset = load_test_set(IMAGE_HEIGHT, IMAGE_WIDTH, BATCH_SIZE)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"QJL_cyN-I3Qs","executionInfo":{"status":"ok","timestamp":1642021741767,"user_tz":-60,"elapsed":3057,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"de933365-345b-4a39-b184-cb79f0b95999"},"execution_count":5,"outputs":[{"output_type":"stream","name":"stdout","text":["Found 6396 files belonging to 2 classes.\n","Found 2132 files belonging to 2 classes.\n","Found 2133 files belonging to 2 classes.\n"]}]},{"cell_type":"markdown","source":["## Experiment 1: one dense layer with 256 neurons and one dropout\n","We try a simple model with a single dense layer of 256 neurons and dropout to fight the high overfitting."],"metadata":{"id":"zRQbZde9FYMQ"}},{"cell_type":"code","source":["inputs = keras.Input(shape=(IMAGE_HEIGHT, IMAGE_WIDTH, 3))\n","x = layers.Rescaling(1./255)(inputs)\n","x = layers.Conv2D(filters=32, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=64, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=128, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(3,3))(x)\n","x = layers.Conv2D(filters=256, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(5,5))(x)\n","x = layers.Flatten()(x)\n","x = layers.Dense(256, activation='relu') (x)\n","x = layers.Dropout(0.5)(x)\n","outputs = layers.Dense(1, activation='sigmoid')(x)\n","model = keras.Model(inputs=inputs, outputs=outputs)\n","\n","compile_model(model)"],"metadata":{"id":"hqnuPM5WFcOX","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1641667644918,"user_tz":-60,"elapsed":246,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"d2d590ab-f634-4db3-975e-6d7dc4fb5013"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Model: \"model_1\"\n","_________________________________________________________________\n"," Layer (type)                Output Shape              Param #   \n","=================================================================\n"," input_2 (InputLayer)        [(None, 300, 300, 3)]     0         \n","                                                                 \n"," rescaling_1 (Rescaling)     (None, 300, 300, 3)       0         \n","                                                                 \n"," conv2d_4 (Conv2D)           (None, 300, 300, 32)      896       \n","                                                                 \n"," max_pooling2d_4 (MaxPooling  (None, 150, 150, 32)     0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_5 (Conv2D)           (None, 150, 150, 64)      18496     \n","                                                                 \n"," max_pooling2d_5 (MaxPooling  (None, 75, 75, 64)       0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_6 (Conv2D)           (None, 75, 75, 128)       73856     \n","                                                                 \n"," max_pooling2d_6 (MaxPooling  (None, 25, 25, 128)      0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_7 (Conv2D)           (None, 25, 25, 256)       295168    \n","                                                                 \n"," max_pooling2d_7 (MaxPooling  (None, 5, 5, 256)        0         \n"," 2D)                                                             \n","                                                                 \n"," flatten_1 (Flatten)         (None, 6400)              0         \n","                                                                 \n"," dense_2 (Dense)             (None, 256)               1638656   \n","                                                                 \n"," dropout_1 (Dropout)         (None, 256)               0         \n","                                                                 \n"," dense_3 (Dense)             (None, 1)                 257       \n","                                                                 \n","=================================================================\n","Total params: 2,027,329\n","Trainable params: 2,027,329\n","Non-trainable params: 0\n","_________________________________________________________________\n"]}]},{"cell_type":"code","source":["run_model(model, \"OneDense256_1dropout\", epochs=30)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"tyYJjiL-34xW","executionInfo":{"status":"ok","timestamp":1641666928763,"user_tz":-60,"elapsed":1656999,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"779602a8-681a-4351-ab9d-c9306d0e6d00"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Epoch 1/30\n","200/200 [==============================] - ETA: 0s - loss: 0.6938 - accuracy: 0.5086\n","Epoch 00001: val_loss improved from inf to 0.68463, saving model to /content/models/prova.h5\n","200/200 [==============================] - 65s 256ms/step - loss: 0.6938 - accuracy: 0.5086 - val_loss: 0.6846 - val_accuracy: 0.6628\n","Epoch 2/30\n","200/200 [==============================] - ETA: 0s - loss: 0.6742 - accuracy: 0.5766\n","Epoch 00002: val_loss improved from 0.68463 to 0.64231, saving model to /content/models/prova.h5\n","200/200 [==============================] - 48s 236ms/step - loss: 0.6742 - accuracy: 0.5766 - val_loss: 0.6423 - val_accuracy: 0.6501\n","Epoch 3/30\n","200/200 [==============================] - ETA: 0s - loss: 0.6048 - accuracy: 0.6750\n","Epoch 00003: val_loss improved from 0.64231 to 0.58783, saving model to /content/models/prova.h5\n","200/200 [==============================] - 46s 230ms/step - loss: 0.6048 - accuracy: 0.6750 - val_loss: 0.5878 - val_accuracy: 0.7106\n","Epoch 4/30\n","200/200 [==============================] - ETA: 0s - loss: 0.5593 - accuracy: 0.7104\n","Epoch 00004: val_loss improved from 0.58783 to 0.50308, saving model to /content/models/prova.h5\n","200/200 [==============================] - 47s 235ms/step - loss: 0.5593 - accuracy: 0.7104 - val_loss: 0.5031 - val_accuracy: 0.7659\n","Epoch 5/30\n","200/200 [==============================] - ETA: 0s - loss: 0.5396 - accuracy: 0.7281\n","Epoch 00005: val_loss improved from 0.50308 to 0.47627, saving model to /content/models/prova.h5\n","200/200 [==============================] - 47s 232ms/step - loss: 0.5396 - accuracy: 0.7281 - val_loss: 0.4763 - val_accuracy: 0.7824\n","Epoch 6/30\n","200/200 [==============================] - ETA: 0s - loss: 0.5105 - accuracy: 0.7491\n","Epoch 00006: val_loss did not improve from 0.47627\n","200/200 [==============================] - 46s 229ms/step - loss: 0.5105 - accuracy: 0.7491 - val_loss: 0.5563 - val_accuracy: 0.7223\n","Epoch 7/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4724 - accuracy: 0.7753\n","Epoch 00007: val_loss did not improve from 0.47627\n","200/200 [==============================] - 47s 232ms/step - loss: 0.4724 - accuracy: 0.7753 - val_loss: 0.6881 - val_accuracy: 0.5971\n","Epoch 8/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4442 - accuracy: 0.7952\n","Epoch 00008: val_loss improved from 0.47627 to 0.43868, saving model to /content/models/prova.h5\n","200/200 [==============================] - 47s 233ms/step - loss: 0.4442 - accuracy: 0.7952 - val_loss: 0.4387 - val_accuracy: 0.7969\n","Epoch 9/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4077 - accuracy: 0.8172\n","Epoch 00009: val_loss did not improve from 0.43868\n","200/200 [==============================] - 46s 229ms/step - loss: 0.4077 - accuracy: 0.8172 - val_loss: 0.4609 - val_accuracy: 0.7983\n","Epoch 10/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3816 - accuracy: 0.8279\n","Epoch 00010: val_loss improved from 0.43868 to 0.39978, saving model to /content/models/prova.h5\n","200/200 [==============================] - 46s 229ms/step - loss: 0.3816 - accuracy: 0.8279 - val_loss: 0.3998 - val_accuracy: 0.8302\n","Epoch 11/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3579 - accuracy: 0.8407\n","Epoch 00011: val_loss did not improve from 0.39978\n","200/200 [==============================] - 47s 232ms/step - loss: 0.3579 - accuracy: 0.8407 - val_loss: 0.4070 - val_accuracy: 0.8410\n","Epoch 12/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3298 - accuracy: 0.8604\n","Epoch 00012: val_loss did not improve from 0.39978\n","200/200 [==============================] - 48s 236ms/step - loss: 0.3298 - accuracy: 0.8604 - val_loss: 0.4134 - val_accuracy: 0.8213\n","Epoch 13/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3205 - accuracy: 0.8588\n","Epoch 00013: val_loss improved from 0.39978 to 0.34448, saving model to /content/models/prova.h5\n","200/200 [==============================] - 48s 238ms/step - loss: 0.3205 - accuracy: 0.8588 - val_loss: 0.3445 - val_accuracy: 0.8583\n","Epoch 14/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2806 - accuracy: 0.8770\n","Epoch 00014: val_loss did not improve from 0.34448\n","200/200 [==============================] - 45s 226ms/step - loss: 0.2806 - accuracy: 0.8770 - val_loss: 0.4393 - val_accuracy: 0.8157\n","Epoch 15/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2772 - accuracy: 0.8793\n","Epoch 00015: val_loss improved from 0.34448 to 0.34271, saving model to /content/models/prova.h5\n","200/200 [==============================] - 46s 227ms/step - loss: 0.2772 - accuracy: 0.8793 - val_loss: 0.3427 - val_accuracy: 0.8626\n","Epoch 16/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2488 - accuracy: 0.8918\n","Epoch 00016: val_loss did not improve from 0.34271\n","200/200 [==============================] - 46s 226ms/step - loss: 0.2488 - accuracy: 0.8918 - val_loss: 0.3773 - val_accuracy: 0.8447\n","Epoch 17/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2427 - accuracy: 0.8976\n","Epoch 00017: val_loss did not improve from 0.34271\n","200/200 [==============================] - 47s 231ms/step - loss: 0.2427 - accuracy: 0.8976 - val_loss: 0.3783 - val_accuracy: 0.8485\n","Epoch 18/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2192 - accuracy: 0.9088\n","Epoch 00018: val_loss improved from 0.34271 to 0.34034, saving model to /content/models/prova.h5\n","200/200 [==============================] - 47s 234ms/step - loss: 0.2192 - accuracy: 0.9088 - val_loss: 0.3403 - val_accuracy: 0.8757\n","Epoch 19/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2110 - accuracy: 0.9120\n","Epoch 00019: val_loss did not improve from 0.34034\n","200/200 [==============================] - 46s 228ms/step - loss: 0.2110 - accuracy: 0.9120 - val_loss: 0.3979 - val_accuracy: 0.8743\n","Epoch 20/30\n","200/200 [==============================] - ETA: 0s - loss: 0.1848 - accuracy: 0.9237\n","Epoch 00020: val_loss did not improve from 0.34034\n","200/200 [==============================] - 46s 226ms/step - loss: 0.1848 - accuracy: 0.9237 - val_loss: 0.3757 - val_accuracy: 0.8720\n","Epoch 21/30\n","200/200 [==============================] - ETA: 0s - loss: 0.1676 - accuracy: 0.9290\n","Epoch 00021: val_loss did not improve from 0.34034\n","200/200 [==============================] - 46s 226ms/step - loss: 0.1676 - accuracy: 0.9290 - val_loss: 0.4138 - val_accuracy: 0.8729\n","Epoch 22/30\n","200/200 [==============================] - ETA: 0s - loss: 0.1570 - accuracy: 0.9361\n","Epoch 00022: val_loss did not improve from 0.34034\n","200/200 [==============================] - 46s 227ms/step - loss: 0.1570 - accuracy: 0.9361 - val_loss: 0.4187 - val_accuracy: 0.8429\n","Epoch 23/30\n","200/200 [==============================] - ETA: 0s - loss: 0.1398 - accuracy: 0.9434\n","Epoch 00023: val_loss did not improve from 0.34034\n","200/200 [==============================] - 46s 227ms/step - loss: 0.1398 - accuracy: 0.9434 - val_loss: 0.4557 - val_accuracy: 0.8466\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"}},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":["evaluate_model(tf.keras.models.load_model(GLOBAL_MODELS_FOLDER + '/OneDense256_1dropout.h5'))"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":741},"id":"kQ230evdBKtP","executionInfo":{"status":"ok","timestamp":1641667695714,"user_tz":-60,"elapsed":12275,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"61eec40a-d0cc-4624-aaab-9f70a7a3ce3f"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Classification report: \n","              precision    recall  f1-score   support\n","\n","           0     0.7994    0.8132    0.8063       696\n","           1     0.9088    0.9012    0.9050      1437\n","\n","    accuracy                         0.8725      2133\n","   macro avg     0.8541    0.8572    0.8556      2133\n","weighted avg     0.8731    0.8725    0.8728      2133\n","\n"]},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 2 Axes>"]},"metadata":{"needs_background":"light"}},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAYoAAAEWCAYAAAB42tAoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3hU1dbA4d+iGTqCDelNSIDQIgiI9CJVLkiTKggWxGvDfuUqH9de0CDSxIaIdBQBURRQ6YSWSIcQimCkhZqyvj9mEmJIGSBTs97nmYc5M3vOrBySWbP3PnsdUVWMMcaYjOTydgDGGGN8myUKY4wxmbJEYYwxJlOWKIwxxmTKEoUxxphMWaIwxhiTKUsUxhhjMmWJwgQcEdknIudEJE5EjojIVBEplKZNIxH5SUROi8hJEVkgIiFp2hQRkfdEJNq5r93O7Rs8+xMZ412WKEyg6qSqhYDaQB3gueQnRKQhsASYB9wKVAA2Ab+KSEVnm3zAj0B1oB1QBGgIxAL13RW0iORx176NuVqWKExAU9UjwGIcCSPZG8Bnqvq+qp5W1b9V9UVgFTDK2aY/UBboqqqRqpqkqkdV9VVVXZjee4lIdRH5QUT+FpE/ReR55+NTRWR0qnbNRCQm1fY+EXlGRDYDZ5z3Z6bZ9/siMtZ5v6iITBaRwyJyUERGi0juazxUxmTIEoUJaCJSGrgb2OXcLgA0Ar5Jp/kMoLXzfitgkarGufg+hYGlwCIcvZTKOHokruoNdACKAdOB9s594kwCPYBpzrZTgQTne9QB2gBDruC9jLkilihMoJorIqeBA8BR4GXn48Vx/N4fTuc1h4Hk+YcSGbTJSEfgiKq+rarnnT2V1Vfw+rGqekBVz6nqfmAD0NX5XAvgrKquEpGbgfbAv1X1jKoeBd4Fel3BexlzRSxRmEB1j6oWBpoB1biUAI4DSUDJdF5TEvjLeT82gzYZKQPsvqpIHQ6k2Z6Go5cB0IdLvYlyQF7gsIicEJETwMfATdfw3sZkyhKFCWiq+guOoZq3nNtngN+Be9Np3oNLw0VLgbYiUtDFtzoAVMzguTNAgVTbt6QXaprtb4BmzqGzrlxKFAeAC8ANqlrMeSuiqtVdjNOYK2aJwuQE7wGtRaSWc/tZYICIjBCRwiJyvXOyuSHwX2ebz3F8KM8SkWoikktESojI8yLSPp33+BYoKSL/FpHrnPtt4HwuAsecQ3ERuQX4d1YBq+ox4GfgE2CvqkY5Hz+M44ytt52n7+YSkUoi0vQqjosxLrFEYQKe80P3M+A/zu2VQFvgXzjmIfbjmBS+U1V3OttcwDGh/QfwA3AKWINjCOuyuQdVPY1jIrwTcATYCTR3Pv05jtNv9+H4kP/axdCnOWOYlubx/kA+IBLHUNpMrmyYzJgrInbhImOMMZmxHoUxxphMuS1RiMgUETkqIlszeF5EZKyI7BKRzSJS112xGGOMuXru7FFMxVH6ICN3A1Wct6HAR26MxRhjzFVyW6JQ1eXA35k06YKjjIKq6iqgmIjYhJwxxvgYbxYgK8U/FxnFOB+7bDWsiAzF0eugYMGC9apVq+aRAI3xd3+fuciJs/HeDiNHO3MxAYCC+bzzcXvm73PEn01Akw7+pao3Xs0+/KJSpapOACYAhIWF6bp167wckcmJpq2OZl7EQW+HcUX+2vs3RYAGFYp7O5QcrUvtUvRpUNZj75d8NquI8NFHazl69AyjRjXff7X782aiOIij7EGy0s7HjMlW2fUBv3qvYyTVnz50G1Qo7vEPKeNdBw+e4qGHvqNnz+rcd18oDz10OwCjRl39Pr2ZKOYDw0VkOtAAOOlcdWrMVUsvKWTXB7x96BpfpqpMmrSBp576gfj4RDp0qJJt+3ZbohCRr3AUZLvBWXv/ZRzFzFDV8cBCHFUwdwFngUHuisUElsx6COklBfuAN4Fu9+6/eeCBBSxbto/mzcszcWInKlXKvp6v2xKFqvbO4nkFHnHX+xvfd7VDQpn1ECwpmJxoy5ajrF9/mAkTOjJkSF1EJFv373clPGwy2/e5mgCuZUjIkoHJ6bZuPcqGDYfp399R6zI29iwlShTIsL2IrFfVsKt5L78468n4powSgqsJwL79G3PlLl5MZMyYFYwZs4Kbby5Ejx7VCQrKk2mSuFaWKIxLrmSS2BKAMe6xenUMgwfPZ9u2Y/TtG8q777YlKMj9H+OWKEyGUicHmyQ2xrsOHjxFkyafcPPNhfj229506HCbx97bEoVJkbbXkDo5WFIwxjt27IjltttKUKpUEb7+ujstW1akSJHrPBqDJYocLLPEkPyvJQdjvOPEifOMHPkDkyZt4OefB3LXXeXo2jXYK7FYoshBLDEY4x/mz9/OQw99x5EjcTz9dCNuv/1Wr8ZjiSIHSE4QlhiM8X1Dhsxn8uSN1Kx5E/Pm9SIszLtJAixRBLT0EoQlBmN8T+oifmFht1KuXFGeeeZO8uXL7eXIHCxRBLB5EQeJPHzKEoQxPuzAgZM8+OB39OpVnX79avHgg1e1Js6tLFEEqGmro1m9928aVCjO18MaejscY0waSUnKxx+v45lnlpKYqHTt6rvX2bFEEWDSDjd1qV3KyxEZY9LauTOWIUMWsHz5flq1qsiECR2pUOF6b4eVIUsUASCjhXE23GSMb4qMPMbmzX8yZUpnBg6sne1F/LKbJQo/lVFysARhjG/atOkIERFHGDCgNl26VGPPnnJcf31+b4flEksUfsJWTRvjny5cSGD06OW89tqvlCxZiJ49axAUlMdvkgRYovAL01ZH8/ycLYCtgTDGn/z++wEGD55PVNRf9O9fi3feaeORIn7Zzf8izkHSTkyP6VrTEoMxfuLgwVM0bTqVW24pxMKFfbj77uy7NKmnWaLwYbYOwhj/ExV1jODgGylVqggzZtxLy5YVKFzYs0X8slsubwdg0pe8DiKkZBG+HtbQkoQxPu748XPcf/88QkLGsWLFfgDuuaea3ycJsB6FT0o9J2HrIIzxfXPmRPHwwws5duwMzz13J7ffHlh/t5YofFDy2U02J2GM77v//nl88kkEtWvfwnff9aFu3ZLeDinbWaLwUQ0qFLckYYyPSl3E7447SlOlSnGeeqoRefP6RhG/7GaJwockn+UUefgUISWLeDscY0w69u8/wbBh39KnT03696/F0KH1vB2S29lkto9InpdInsC2uQljfEtSkhIevoYaNT5i5cpo4uMTvR2Sx1iPwkfYvIQxvmv79r8YMmQBK1dG06ZNJT7+uCPlyxfzdlgeY4nCB6QuCW5Jwhjfs317LNu2HWXq1C7071/L54v4ZTdLFF5kJcGN8V0bNx4mIuIIgwbVoXPnquzZ8xjFigV5OyyvsEThJWnrN9nKa2N8w/nzCbzyyi+88cavlCpVhN69axIUlCfHJgmwROE1NidhjO/59ddoBg+ez/btsQwaVJu33/bPIn7ZzY6AF9ichDG+5+DBUzRv/imlShVh8eK+tGlTydsh+QxLFB5m5TmM8S2RkccICXEU8Zs1qwfNm1egUKF83g7Lp9g6Cg9KnSRsyMkY7/r773MMHDiX6tXHsXy5o4hfp05VLUmkw3oUHmTzEsb4hlmzInnkkYXExp7jhReaUL++9e4zY4nCw2xewhjvGjhwLp9+uom6dUuyaFFfate+xdsh+TxLFMaYgJe6iF+jRmUIDr6BJ59sRJ48NvruCrceJRFpJyLbRWSXiDybzvNlRWSZiGwUkc0i0t6d8XhT8plOxhjP2rv3OG3afMFnn20CYOjQejzzzJ2WJK6A246UiOQGwoG7gRCgt4iEpGn2IjBDVesAvYBx7orHW6atjqbnx7/bmU7GeFhiYhJjx66mRo2PWLUqBmenwlwFdw491Qd2qeoeABGZDnQBIlO1USC5nnZR4JAb4/E4W31tjHdERR1j8OD5/P57DHffXZnx4ztStmxRb4flt9yZKEoBB1JtxwAN0rQZBSwRkUeBgkCr9HYkIkOBoQBly/rPB62d5WSMd+za9Tfbt8fy+eddue++mjmuiF928/YgXW9gqqqWBtoDn4vIZTGp6gRVDVPVsBtvvNHjQV4NW31tjGetX3+IKVM2Ao71EHv3PkbfvqGWJLKBOxPFQaBMqu3SzsdSGwzMAFDV34Eg4AY3xuQRtvraGM85dy6eZ59dSoMGk3j11eWcP58AQJEi13k5ssDhzkSxFqgiIhVEJB+Oyer5adpEAy0BRCQYR6I45saY3M5WXxvjOcuX76dWrfG8/vqvDBxYm40bh1kRPzdw2xFV1QQRGQ4sBnIDU1R1m4i8AqxT1fnAk8BEEXkcx8T2QFX/PTfBkoQxnnPw4ClatvyMMmWKsHRpP1q2rOjtkAKWW1Ovqi4EFqZ57D+p7kcCjd0ZgyfZ5LUx7rdly5/UrHkzpUoVYc6cnjRvXp6CBa0+kzt5ezI74NjktTHu8ddfZ+nXbw6hoeNTivh17HibJQkPsESRTWzltTHuoarMmLGNkJBwpk/fyssvN6VBAztJxJNs1icb2FlOxrjPgAFz+fzzzYSF3cqPP3amZs2bvR1SjmOJ4hrZBLYx2S91Eb+mTcsRGnoz//73HVafyUssUVwDSxLGZL89e47zwAML6Nu3JoMG1WHw4LreDinHs/R8DewsJ2OyT2JiEu+9t4qaNT9i7dqD5MplK6p9hfUorsK01dHMizhI5OFTdpaTMdkgMvIY998/j9WrD9KhQxXGj+9I6dJFsn6h8QhLFFchOUmElCxik9fGZIO9e4+ze/dxpk37F7161bD6TD7GEsUVSN2TCClZhK+HNfR2SMb4rbVrDxIRcYQHHqhHhw63sWfPCAoXtvpMvsjmKFyUPHG9eu/f1pMw5hqcPRvPU08t4Y47JvO//61MKeJnScJ3WY/CRTZxbcy1+/nnfQwZMp/du48zbFg9Xn+9lRXx8wP2P+QCu7aEMdcuJuYUrVt/TrlyRfnpp/40b17B2yEZF1miyIKtujbm2mzadIRatW6hdOkizJvXi2bNylOgQF5vh2WugM1RZMGGnIy5OseOnaFPn1nUrv0xv/yyD4D27atYkvBD1qNwgQ05GeM6VWX69K2MGLGIkyfP89//NqNhwzJZv9D4LEsUxphs1a/fHL78cgsNGpRi8uTOVK9+k7dDMtfI5UQhIgVU9aw7g/E1qSexjTEZS0pSRBxF/Jo3L0+9eiUZMaIBuXPb6HYgyPJ/UUQaiUgk8Idzu5aIjHN7ZD4geX7CJrGNydiuXX/TsuVnfPJJBACDB9fl8ccbWpIIIK78T74LtAViAVR1E3CXO4PyJTY/YUz6EhKSeOut36hZ8yM2bjxMvny5vR2ScROXhp5U9UCa2iuJ7gnHd9iwkzEZ27r1KIMGzWPdukN06VKVceM6cOuthb0dlnETVxLFARFpBKiI5AUeA6LcG5b32bCTMRmLjj7J/v0nmD69Gz16VLcifgHOlUTxIPA+UAo4CCwBHnZnUL7Chp2MuWT16hg2bfqToUPr0b59FfbseYxChfJ5OyzjAa7MUVRV1ftU9WZVvUlV+wLB7g7Mm5KHnYwxcObMRZ54YjENG07mjTd+5cIFRxE/SxI5hyuJ4gMXHwsIVrLDmEt++mkvoaHjeffdVTz4YBgbNgzjuuts+VVOk+H/uIg0BBoBN4rIE6meKgIE7OkNVrLDGIeYmFO0bfsFFSoU45dfBnLXXeW8HZLxksy+GuQDCjnbpD6d4RTQ3Z1BeYtViTUGNm48TJ06JSlduggLFvSmadNy5M9v9ZlysgwThar+AvwiIlNVdb8HY/IKG3IyOd2ff8YxYsQiZszYxs8/D6Bp0/K0a1fZ22EZH+DKYONZEXkTqA4EJT+oqi3cFpWHpU4SNuRkchpV5csvt/DYY4uIi7vI6NHNadTIiviZS1xJFF8CXwMdcZwqOwA45s6gPM3mJUxO1qfPbKZP30rDhqWZPLkzwcE3ejsk42NcSRQlVHWyiDyWajhqrbsD8zSblzA5Seoifm3aVKRhw9I88sjtVp/JpMuVRBHv/PewiHQADgFW18IYP7VjRywPPLCA/v1DGTy4LoMG1fF2SMbHuZIoRotIUeBJHOsnigD/dmtUxphsl5CQxDvv/M7LL/9MUFAeO5PJuCzLRKGq3zrvngSaA4hIY3cG5UlW/M/kBJs3/8n9989j/frDdO1ajfDw9pQsaUX8jGsyW3CXG+iBo8bTIlXdKiIdgeeB/EBA9Fet+J/JCWJiTnHgwCm++eZeunULtiJ+5opkNnM1GRgClADGisgXwFvAG6rqUpIQkXYisl1EdonIsxm06SEikSKyTUSmXekPcC1sgZ0JZL/9doDx49cBOIv4jaB79xBLEuaKZTb0FAaEqmqSiAQBR4BKqhrryo6dPZJwoDUQA6wVkfmqGpmqTRXgOaCxqh4XEY9eXNd6EyYQxcVd5IUXfuSDD9ZQqVJxBg2qzXXX5aFgQSviZ65OZj2Ki6qaBKCq54E9riYJp/rALlXdo6oXgelAlzRtHgDCVfW4832OXsH+s4X1JkwgWbJkNzVqjOODD9bwyCO3s2HDUCviZ65ZZr9B1URks/O+AJWc2wKoqoZmse9SwIFU2zFAgzRtbgMQkV9xFBocpaqL0u5IRIYCQwHKls2eD3WbxDaB5sCBk3ToMI1Kla5n+fJB3HmnfQEy2SOzROGJa07kAaoAzYDSwHIRqamqJ1I3UtUJwASAsLAwzY43tmEnEyjWrz9EvXq3UqZMURYu7EOTJuUICrJehMk+GQ49qer+zG4u7PsgkLpgTGnnY6nFAPNVNV5V9wI7cCQOj7BhJ+PPjhyJ4957vyEsbCK//LIPgNatK1mSMNnOnev11wJVRKSCiOQDegHz07SZi6M3gYjcgGMoao8bYwLsCnbGv6kqn34aQUhIOAsWbGfMmBZWxM+4ldu+eqhqgogMBxbjmH+YoqrbROQVYJ2qznc+10ZEIoFE4OkrnDC/KjbsZPxZr16zmDFjG40bl2HSpM5Uq3aDt0MyAc6lRCEi+YGyqrr9SnauqguBhWke+0+q+wo84bx5lA07GX+Suohf+/aVadKkLA8/fDu5ctmaCON+WQ49iUgnIAJY5NyuLSJph5CMMW7yxx9/cdddnzB58kYABgyozfDh9S1JGI9xZY5iFI41EScAVDUCqODGmIwxQHx8ImPGrKBWrfFERh6jUCFbMGe8w6Uy46p6Ms2y/2w5RdUbbP2E8QcREUcYNGgeERFH6N49hA8+uJtbbink7bBMDuVKotgmIn2A3M6SGyOA39wblvvYRLbxB0eOxHHkSByzZvXgX//yxJImYzLmytDToziul30BmIaj3LhfX4/CJrKNL1q5Mppx4xwXj2zXrjK7d4+wJGF8gis9imqq+gLwgruDMSYnOn36As899yPh4WupUqU4gwfX4brr8lCggF1YyPgGV3oUb4tIlIi8KiI13B6RG9lCO+NrFi/eRY0aHzFu3Foee6wBGzYMsyJ+xue4coW75iJyC46LGH0sIkWAr1V1tNujy2Y2P2F8yYEDJ+nY8SsqVy7OypX32+pq47NcKuGhqkdUdSzwII41Ff/J4iU+y+YnjDepKmvWOL6wlClTlO+/v4+NG4dZkjA+zZUFd8EiMkpEtgAf4DjjqbTbI8tmNuxkvO3w4dN06zaDBg0mpRTxa9WqohXxMz7Pld/QKcDXQFtVPeTmeNzGhp2Mt6gqU6dG8MQTSzh/PoHXX29F48bWqzX+w5U5ioaeCMQTbNjJeEOPHjOZOTOSJk3KMmlSZ267rYS3QzLmimSYKERkhqr2cA45pV6J7eoV7ozJsRITkxARcuUSOnW6jRYtyjNsWJjVZzJ+KbMexWPOfzt6IhBjAkVU1DEGD57PoEG1eeCBevTvX8vbIRlzTTK7wt1h592H07m63cOeCc8Y/xEfn8jo0cupXftjtm+PpWjRIG+HZEy2cOX02NbpPHZ3dgdijD/buPEwYWETeemlZXTtWo2oqEfo0aO6t8MyJltkNkfxEI6eQ0UR2ZzqqcLAr+4OzBh/8uefZ/jrr7PMnduTLl2qeTscY7JVZnMU04Dvgf8Bz6Z6/LSq2oIEk+MtX76fLVv+5JFH6tOuXWV27XqU/PmtPpMJPJkNPamq7gMeAU6nuiEidjEHk2OdOnWBhx/+jqZNpzJ27BouXEgAsCRhAlZWPYqOwHocp8emPq9PgYpujMsYn7Rw4U6GDfuWQ4dO88QTd/DKK82tiJ8JeBn+hqtqR+e/dtlTY3AU8evSZTpVq5Zg5sx7adDA7yrZGHNVXKn11FhECjrv9xWRd0TEljebHEFVWbUqBnAU8VuypC8bNgyzJGFyFFdOj/0IOCsitYAngd3A526NKptZQUBzNQ4dOs0993xNw4aTU4r4NW9egXz5cns3MGM8zJVEkaCqCnQBPlTVcBynyPoNKwhoroSqMmnSBkJCwlmyZDdvvdXaiviZHM2VWbjTIvIc0A9oIiK5AL87vcMKAhpXde/+DbNnR9G0aTkmTepM5cp2kp/J2VzpUfQELgD3q+oRHNeieNOtUWUjG3YyrkhMTCIpyVH78p57qjJ+fAd++mmAJQljcCFROJPDl0BREekInFfVz9weWTaxYSeTla1bj9K48RQmT94AQL9+tazSqzGpuHLWUw9gDXAvjutmrxaR7u4OLDsk9yZs2Mmk5+LFRP7735+pW/djdu8+zvXX5/d2SMb4JFfmKF4AblfVowAiciOwFJjpzsCyg/UmTEbWrz/EwIHz2Lr1KH361OS999py440FvR2WMT7JlUSRKzlJOMXi2tyGV1lvwmQmNvYcJ06cZ8GC3nTseJu3wzHGp7mSKBaJyGLgK+d2T2Ch+0LKHtabMGktW7aXLVuOMmJEA9q0qcTOnY8SFGTlN4zJiiuT2U8DHwOhztsEVX3G3YFlB+tNGICTJ88zbNgCWrT4jI8+WpdSxM+ShDGuyex6FFWAt4BKwBbgKVU96KnAjMkOCxZs58EHv+PIkTieeqoh//2vFfEz5kpl1qOYAnwLdMNRQfYDj0SUDWzthAFHEb9u3WZQokR+Vq0azJtvtqFAAb9bK2qM12X21aqwqk503t8uIhs8EVB2sPmJnEtV+f33GBo1KuMs4tePRo3KWH0mY65BZj2KIBGpIyJ1RaQukD/NdpZEpJ2IbBeRXSLybCbtuomIikjYlf4AGbH5iZwnJuYUnTtPp3HjKSlF/Jo1K29JwphrlFmP4jDwTqrtI6m2FWiR2Y5FJDcQDrQGYoC1IjJfVSPTtCsMPAasvrLQjXFISlImTlzP00//QEJCEu+804Y777QvCcZkl8wuXNT8GvddH9ilqnsARGQ6jgq0kWnavQq8Djx9je8H/HP9hMkZunWbwdy5f9CiRQUmTuxExYrXezskYwKKOxfOlQIOpNqOcT6WwjmEVUZVv8tsRyIyVETWici6Y8eOZfqmNj+RMyQkXCri161bMBMndmLp0n6WJIxxA6+tsHaWK38Hx8WQMqWqE1Q1TFXDbrzxxiz3bfMTgW3z5j9p2HAyEyeuB6Bv31CGDKmLiBXxM8Yd3JkoDgJlUm2Xdj6WrDBQA/hZRPYBdwDzs3NC2wSWCxcSePnlZdSrN4H9+09YbSZjPCTLlUfi+Jp2H1BRVV9xXi/7FlVdk8VL1wJVRKQCjgTRC+iT/KSqngRuSPU+P+NY1Lfuin8KE/DWrj3IwIHziIw8Rr9+obz7bltKlCjg7bCMyRFcWaI6DkjCcZbTK8BpYBZwe2YvUtUEERkOLAZyA1NUdZuIvAKsU9X51xS5yVGOHz9PXNxFFi7sw913V/F2OMbkKK4kigaqWldENgKo6nERyefKzlV1IWkKCKrqfzJo28yVfZqc46ef9rJly5889tgdtGlTiR07hlv5DWO8wJU5injnmgiFlOtRJLk1KpOjnThxngcemE/Llp/x8cfrU4r4WZIwxjtcSRRjgTnATSLyf8BKYIxbozI51rx5fxASEs6UKRGMHNmI9euHWoIwxsuy/AtU1S9FZD3QEhDgHlWNcntkV8EW2/m36OiT3HvvNwQH38j8+b0JC7vV2yEZY3DtrKeywFlgQerHVDXanYFdDVts539UlZUro2nSpBxlyxZl6dL+3HFHaavPZIwPcWXo6Tsc5ca/A34E9gDfuzOoa2GL7fxHdPRJOnSYxl13TU0p4nfXXeUsSRjjY1wZeqqZettZduNht0VkAl5SkjJ+/DqeeWYpqsrYse2siJ8xPuyKZwlVdYOINHBHMCZn+Ne/vmbevO20bl2RCRM6Ub58MW+HZIzJhCtzFE+k2swF1AUOuS0iE5ASEpLIlUvIlUvo2bM6XbpUZeDA2lafyRg/4MocReFUt+twzFV0cWdQJrBs2nSEBg0mMWGCo4hf7941GTSojiUJY/xEpj0K50K7wqr6lIfiMQHk/PkERo9ezuuv/0rx4vm55ZZC3g7JGHMVMkwUIpLHWa+psScDulq2hsK3rFlzkAED5vLHH38xYEAt3nmnLcWL5/d2WMaYq5BZj2INjvmICBGZD3wDnEl+UlVnuzm2K2JrKHzLqVMXOHcunkWL7qNt28reDscYcw1cOespCIjFUT1WcazOVsCnEgXYGgpvW7JkN9u2HeXxxxvSqlVFtm+3In7GBILM/opvcp7xtJVLCSKZujUq41eOHz/HE08sYerUCKpXv5GHH76d667LY0nCmACR2V9ybqAQ/0wQySxRGABmz47ikUcWcuzYGZ577k7+85+mliCMCTCZ/UUfVtVXPBaJ8TvR0Sfp1WsmNWrcxMKFfahTp6S3QzLGuEFmicJOcjeXUVWWL99P06blKVu2KD/9NIAGDUqRN6/VZzImUGW24K6lx6IwfmH//hPcffeXNGv2aUoRvzvvLGtJwpgAl2GPQlX/9mQgxnclJSnjxq3l2WeXAvDBB3fTpEk5L0dljPEUm3U0WbrnnuksWLCDtm0r8fHHHSlXzor4GZOTWKIw6YqPTyR37lzkyiX07l2D7t1D6Ncv1OozGZMDuVIU0Ocll+8w2WPDhsPUrz+J8ePXAY4ifv3717IkYUwOFRCJwsp3ZI9z5+J57rml1K8/kSNH4ihTpoi3QzLG+D69RqMAABgbSURBVICAGXqy8h3XZtWqGAYMmMuOHbHcf39t3nqrDddfb0X8jDEBlCjMtTlz5iLx8Yn88EM/WrWq6O1wjDE+xBJFDrZo0S62bTvKk082omXLivzxx3Dy5bM1EcaYfwqIOQpzZWJjzzJgwFzuvvtLPv10ExcvJgJYkjDGpMsSRQ6iqsycGUlIyDimTdvCiy82Ye3aByxBGGMy5fdDT3ZlO9dFR5+kT59ZhIbezJIlfalV6xZvh2SM8QN+nSimrY7m+TlbADs1NiOqyrJl+2jRogLlyhXj558HUr9+KfLksc6kMcY1fv1pkbx+YkzXmnZqbDr27j1OmzZf0LLlZylF/Bo1KmNJwhhzRfy6RwG2fiI9iYlJfPjhGp5//idy5xY++qiDFfEzxlw1v08U5nJdukznu+920r59FcaP70CZMkW9HZIxxo9ZoggQqYv49esXSu/eNejTp6bVZzLGXDO3DlaLSDsR2S4iu0Tk2XSef0JEIkVks4j8KCI2PnIV1q07RFjYRD76aC0APXvW4L77rNKrMSZ7uC1RiEhuIBy4GwgBeotISJpmG4EwVQ0FZgJvuCueQHTuXDzPPPMDDRpM4tixM3adCGOMW7hz6Kk+sEtV9wCIyHSgCxCZ3EBVl6Vqvwro68Z4Asrvvx9gwIC57Nz5N0OG1OHNN9tQrFiQt8MyxgQgdw49lQIOpNqOcT6WkcHA9+k9ISJDRWSdiKw7duwYYNegOHcugaQkZenSfkyc2NmShDHGbXxiMltE+gJhQNP0nlfVCcAEgLCwMIWceQ2KhQt3sm3bUZ5+ujEtWlQgKuoR8ua18hvGGPdyZ4/iIFAm1XZp52P/ICKtgBeAzqp64UreIKesofjrr7P07TubDh2m8eWXW1KK+FmSMMZ4gjsTxVqgiohUEJF8QC9gfuoGIlIH+BhHkjjqxlj8kqoyffpWgoPDmTFjGy+/3JQ1a6yInzHGs9yWKFQ1ARgOLAaigBmquk1EXhGRzs5mbwKFgG9EJEJE5mewu3/IKfMT0dEnGTBgLhUqFGP9+qGMGtXMkoQxxuPcOkehqguBhWke+0+q+62uZr+BPD+hqvz4415atapIuXLF+OWXgdx++63kzm31mYwx3uG3nz6BOD+xe/fftGz5Ga1bf55SxO+OO0pbkjDGeJVPnPWU0yUmJvH++6t58cWfyJs3Nx9/3NGK+BljfIYlCh/QqdNXfP/9Ljp2vI2PPupA6dJFvB2SMcaksEThJRcvJpInj6OI38CBtenXL5RevWpYfSZjjM+xwW8vWLPmIPXqTWDcOEcRvx49qtO7t1V6Ncb4JksUHnT2bDxPPrmYhg0nc/z4OSpVut7bIRljTJZs6MlDVq6MZsCAuezZc5xhw+rx+uutKFrU6jMZY3yfJQoPcVxYSFi2bADNmpX3djjGGOMySxRutGDBdqKi/mLkyMY0b16ByMhHyJPHRvuMMf7FPrXc4NixM/TpM4vOnafz1VdbU4r4WZIwxvgjv/vk+vvMRZ+t86SqTJu2heDgcGbOjOSVV5qxevUQq89kjPFrfjf0dOJsPEXwzTpP0dEnGTRoHnXq3MLkyZ2pXv0mb4dkjDHXzO8SBfhWnaekJOWHH3bTtm1lypUrxooVg6hXr6TVZzLGBAz7NLsGO3fG0qLFp7Rr9yXLl+8HoH79UpYkjDEBxT7RrkJCQhJvvvkroaHjiYg4wuTJnWnSxDd6OMYYk938cujJ2zp2nMbixbvp0qUq48Z14NZbC3s7JOOD4uPjiYmJ4fz5894OxeQgQUFBlC5dmrx582bbPi1RuOjChQTy5s1NrlzCkCF1uf/+Otx7b4jVZzIZiomJoXDhwpQvX95+T4xHqCqxsbHExMRQoUKFbNuvDT25YNWqGOrWnUB4+BoAuncPoUeP6vbHbzJ1/vx5SpQoYb8nxmNEhBIlSmR7L9YSRSbOnLnI448volGjyZw+fYEqVUp4OyTjZyxJGE9zx++cDT1lYMWK/QwYMJe9e0/w8MNh/O9/rShS5Dpvh2WMMR5nPYoMJCQkkTdvbn75ZSDh4R0sSRi/lDt3bmrXrk2NGjXo1KkTJ06cSHlu27ZttGjRgqpVq1KlShVeffVVVDXl+e+//56wsDBCQkKoU6cOTz75pDd+hExt3LiRwYMHezuMDF24cIGePXtSuXJlGjRowL59+9Jt9/7771OjRg2qV6/Oe++9l/L4008/TbVq1QgNDaVr164p/39btmxh4MCBHvgJnFTVr27Xl62mPcb/pu4wZ06UjhmzPGU7Pj7RLe9jcobIyEhvh6AFCxZMud+/f38dPXq0qqqePXtWK1asqIsXL1ZV1TNnzmi7du30ww8/VFXVLVu2aMWKFTUqKkpVVRMSEnTcuHHZGlt8fPw176N79+4aERHh0fe8EuHh4Tps2DBVVf3qq6+0R48el7XZsmWLVq9eXc+cOaPx8fHasmVL3blzp6qqLl68OCXmkSNH6siRI1Ne17JlS92/f3+675ve7x6wTq/yc9eGnoA//4zj0Ue/55tvIqlbtyRPPtmIfPlyWxE/k23+u2AbkYdOZes+Q24twsudqrvcvmHDhmzevBmAadOm0bhxY9q0aQNAgQIF+PDDD2nWrBmPPPIIb7zxBi+88ALVqlUDHD2Thx566LJ9xsXF8eijj7Ju3TpEhJdffplu3bpRqFAh4uLiAJg5cybffvstU6dOZeDAgQQFBbFx40YaN27M7NmziYiIoFixYgBUqVKFlStXkitXLh588EGio6MBeO+992jcuPE/3vv06dNs3ryZWrVqAbBmzRoee+wxzp8/T/78+fnkk0+oWrUqU6dOZfbs2cTFxZGYmMjChQt59NFH2bp1K/Hx8YwaNYouXbqwb98++vXrx5kzZwD48MMPadSokcvHNz3z5s1j1KhRAHTv3p3hw4ejqv+YR4iKiqJBgwYUKFAAgKZNmzJ79mxGjhyZ8v8DcMcddzBz5syU7U6dOjF9+nRGjhx5TTG6IkcnClXliy828+9/LyYu7iL/938tePrpRuTNa0X8TGBJTEzkxx9/TBmm2bZtG/Xq1ftHm0qVKhEXF8epU6fYunWrS0NNr776KkWLFmXLli0AHD9+PMvXxMTE8Ntvv5E7d24SExOZM2cOgwYNYvXq1ZQrV46bb76ZPn368Pjjj3PnnXcSHR1N27ZtiYqK+sd+1q1bR40aNVK2q1WrxooVK8iTJw9Lly7l+eefZ9asWQBs2LCBzZs3U7x4cZ5//nlatGjBlClTOHHiBPXr16dVq1bcdNNN/PDDDwQFBbFz50569+7NunXrLou/SZMmnD59+rLH33rrLVq1avWPxw4ePEiZMmUAyJMnD0WLFiU2NpYbbrghpU2NGjV44YUXiI2NJX/+/CxcuJCwsLDL9j9lyhR69uyZsh0WFsZrr71miSI9Zy4mZNu+oqNPMmTIAsLCbmXy5M5Uq3ZD1i8y5ipcyTf/7HTu3Dlq167NwYMHCQ4OpnXr1tm6/6VLlzJ9+vSU7euvz/ryvvfeey+5czu+jPXs2ZNXXnmFQYMGMX369JQPwqVLlxIZGZnymlOnThEXF0ehQoVSHjt8+DA33nhjyvbJkycZMGAAO3fuRESIj49Pea5169YUL14cgCVLljB//nzeeustwHEac3R0NLfeeivDhw8nIiKC3Llzs2PHjnTjX7FiRZY/45UIDg7mmWeeoU2bNhQsWJDatWunHJ9k//d//0eePHm47777Uh676aabOHToULbGkhG/SxRwbZVjk5KUxYt3cffdVShXrhi//no/dercYvWZTEDKnz8/ERERnD17lrZt2xIeHs6IESMICQlh+fLl/2i7Z88eChUqRJEiRahevTrr169PGda5UqmHVtKe01+wYMGU+w0bNmTXrl0cO3aMuXPn8uKLLwKQlJTEqlWrCArK+HLB+fPn/8e+X3rpJZo3b86cOXPYt28fzZo1S/c9VZVZs2ZRtWrVf+xv1KhR3HzzzWzatImkpKQM3/tKehSlSpXiwIEDlC5dmoSEBE6ePEmJEpefZj948OCU3t7zzz9P6dKlU56bOnUq3377LT/++ONlxzV//vzpxpjd/O7TsWC+PFddOXbHjliaNZtK+/bT+OWXfQCEhd1qScIEvAIFCjB27FjefvttEhISuO+++1i5ciVLly4FHD2PESNGpAxjPP3004wZMyblW3VSUhLjx4+/bL+tW7cmPDw8ZTt56Onmm28mKiqKpKQk5syZk2FcIkLXrl154oknCA4OTvkQbdOmDR988EFKu4iIiMteGxwczK5du1K2T548SalSji+RU6dOzfA927ZtywcffJByhtfGjRtTXl+yZEly5crF559/TmJiYrqvX7FiBREREZfd0iYJgM6dO/Ppp58CjrmaFi1apLvO4ejRowBER0cze/Zs+vTpA8CiRYt44403mD9/fsocRrIdO3b8Y+jNnXLEJ2RCQhKvv76S0NCP2LLlKJ980oW77irn7bCM8ag6deoQGhrKV199Rf78+Zk3bx6jR4+matWq1KxZk9tvv53hw4cDEBoaynvvvUfv3r0JDg6mRo0a7Nmz57J9vvjiixw/fpwaNWpQq1Ytli1bBsBrr71Gx44dadSoESVLlsw0rp49e/LFF1/8Y/x97NixrFu3jtDQUEJCQtJNUtWqVePkyZMp3+5HjhzJc889R506dUhIyHiI+qWXXiI+Pp7Q0FCqV6/OSy+9BMDDDz/Mp59+Sq1atfjjjz/+0Qu5WoMHDyY2NpbKlSvzzjvv8NprrwFw6NAh2rdvn9KuW7duhISE0KlTJ8LDw1Mm94cPH87p06dp3bo1tWvX5sEHH0x5zbJly+jQocM1x+gKSc6q/qJ4uWD9e39U1g1Tadv2C5Ys2c2//hVMeHh7brmlUNYvMuYaRUVFERwc7O0wAtq7775L4cKFGTJkiLdD8agLFy7QtGlTVq5cSZ48l88gpPe7JyLrVfXyWXIXBGyP4vz5BBITkwAYOrQuM2fey6xZPSxJGBNAHnroIa67Luctho2Ojua1115LN0m4Q0Amil9/jaZ27fGEh68FoFu3ELp1C/FyVMaY7BYUFES/fv28HYbHValS5R+T9e4WUIkiLu4iI0Z8T5Mmn3D+fALBwXa6q/EufxvaNf7PHb9zfnl6bHp++WUfAwbMJTr6JMOH12fMmJYUKpTP22GZHCwoKIjY2FgrNW48Rp3Xo8jstOKrETCJAqBAgbysWDGIxo3tsqTG+0qXLk1MTAzHjh3zdigmB0m+wl128uuznmbPjuKPP/7i+eebAJCYmGRrIowxJh0+e9aTiLQTke0isktEnk3n+etE5Gvn86tFpLwr+z1yJI7u3WfQrdsM5sz5g4sXHQtjLEkYY0z2c9vQk4jkBsKB1kAMsFZE5qtqZKpmg4HjqlpZRHoBrwM9L9/bJRfOXCQ4OJxz5+L53/9a8uSTDa2InzHGuJE7v4LXB3ap6h5VvQhMB7qkadMF+NR5fybQUrKY9Tsbe54aNW5i06YHefbZOy1JGGOMm7lzMrsUcCDVdgzQIKM2qpogIieBEsBfqRuJyFBgqHPzwsqV9291lsnP6W4gzbHKwexYXGLH4hI7FpdUzbpJ+vzirCdVnQBMABCRdVc7IRNo7FhcYsfiEjsWl9ixuERELr+4hovcOfR0ECiTaru087F024hIHqAoEOvGmIwxxlwhdyaKtUAVEakgIvmAXsD8NG3mAwOc97sDP6m/na9rjDEBzm1DT845h+HAYiA3MEVVt4nIKzgu8j0fmAx8LiK7gL9xJJOsTHBXzH7IjsUldiwusWNxiR2LS676WPjdgjtjjDGeZSvUjDHGZMoShTHGmEz5bKJwV/kPf+TCsXhCRCJFZLOI/CgiAXud16yORap23URERSRgT4105ViISA/n78Y2EZnm6Rg9xYW/kbIiskxENjr/Ttqntx9/JyJTROSoiGzN4HkRkbHO47RZROq6tGNV9bkbjsnv3UBFIB+wCQhJ0+ZhYLzzfi/ga2/H7cVj0Rwo4Lz/UE4+Fs52hYHlwCogzNtxe/H3ogqwEbjeuX2Tt+P24rGYADzkvB8C7PN23G46FncBdYGtGTzfHvgeEOAOYLUr+/XVHoVbyn/4qSyPhaouU9Wzzs1VONasBCJXfi8AXsVRN+y8J4PzMFeOxQNAuKoeB1DVox6O0VNcORYKFHHeLwoc8mB8HqOqy3GcQZqRLsBn6rAKKCYiJbPar68mivTKf5TKqI2qJgDJ5T8CjSvHIrXBOL4xBKIsj4WzK11GVb/zZGBe4MrvxW3AbSLyq4isEpF2HovOs1w5FqOAviISAywEHvVMaD7nSj9PAD8p4WFcIyJ9gTCgqbdj8QYRyQW8Awz0cii+Ig+O4admOHqZy0Wkpqqe8GpU3tEbmKqqb4tIQxzrt2qoapK3A/MHvtqjsPIfl7hyLBCRVsALQGdVveCh2Dwtq2NRGKgB/Cwi+3CMwc4P0AltV34vYoD5qhqvqnuBHTgSR6Bx5VgMBmYAqOrvQBCOgoE5jUufJ2n5aqKw8h+XZHksRKQO8DGOJBGo49CQxbFQ1ZOqeoOqllfV8jjmazqr6lUXQ/NhrvyNzMXRm0BEbsAxFLXHk0F6iCvHIhpoCSAiwTgSRU68Ru18oL/z7Kc7gJOqejirF/nk0JO6r/yH33HxWLwJFAK+cc7nR6tqZ68F7SYuHoscwcVjsRhoIyKRQCLwtKoGXK/bxWPxJDBRRB7HMbE9MBC/WIrIVzi+HNzgnI95GcgLoKrjcczPtAd2AWeBQS7tNwCPlTHGmGzkq0NPxhhjfIQlCmOMMZmyRGGMMSZTliiMMcZkyhKFMcaYTFmiMD5JRBJFJCLVrXwmbeOy4f2mishe53ttcK7evdJ9TBKREOf959M899u1xujcT/Jx2SoiC0SkWBbtawdqpVTjOXZ6rPFJIhKnqoWyu20m+5gKfKuqM0WkDfCWqoZew/6uOaas9isinwI7VPX/Mmk/EEcF3eHZHYvJOaxHYfyCiBRyXmtjg4hsEZHLqsaKSEkRWZ7qG3cT5+NtROR352u/EZGsPsCXA5Wdr33Cua+tIvJv52MFReQ7EdnkfLyn8/GfRSRMRF4D8jvj+NL5XJzz3+ki0iFVzFNFpLuI5BaRN0VkrfM6AcNcOCy/4yzoJiL1nT/jRhH5TUSqOlcpvwL0dMbS0xn7FBFZ42ybXvVdY/7J2/XT7Wa39G44VhJHOG9zcFQRKOJ87gYcK0uTe8Rxzn+fBF5w3s+No/bTDTg++As6H38G+E867zcV6O68fy+wGqgHbAEK4lj5vg2oA3QDJqZ6bVHnvz/jvP5Fckyp2iTH2BX41Hk/H45KnvmBocCLzsevA9YBFdKJMy7Vz/cN0M65XQTI47zfCpjlvD8Q+DDV68cAfZ33i+Go/1TQ2//fdvPtm0+W8DAGOKeqtZM3RCQvMEZE7gKScHyTvhk4kuo1a4EpzrZzVTVCRJriuFDNr87yJvlwfBNPz5si8iKOGkCDcdQGmqOqZ5wxzAaaAIuAt0XkdRzDVSuu4Of6HnhfRK4D2gHLVfWcc7grVES6O9sVxVHAb2+a1+cXkQjnzx8F/JCq/aciUgVHiYq8Gbx/G6CziDzl3A4Cyjr3ZUy6LFEYf3EfcCNQT1XjxVEdNih1A1Vd7kwkHYCpIvIOcBz4QVV7u/AeT6vqzOQNEWmZXiNV3SGO6160B0aLyI+q+oorP4SqnheRn4G2QE8cF9kBxxXHHlXVxVns4pyq1haRAjhqGz0CjMVxsaZlqtrVOfH/cwavF6Cbqm53JV5jwOYojP8oChx1JonmwGXXBRfHtcL/VNWJwCQcl4RcBTQWkeQ5h4IicpuL77kCuEdECohIQRzDRitE5FbgrKp+gaMgY3rXHY539mzS8zWOYmzJvRNwfOg/lPwaEbnN+Z7pUscVDUcAT8qlMvvJ5aIHpmp6GscQXLLFwKPi7F6Jo/KwMZmyRGH8xZdAmIhsAfoDf6TTphmwSUQ24vi2/r6qHsPxwfmViGzGMexUzZU3VNUNOOYu1uCYs5ikqhuBmsAa5xDQy8DodF4+AdicPJmdxhIcF5daqo5Ld4IjsUUCG0RkK46y8Zn2+J2xbMZxUZ43gP85f/bUr1sGhCRPZuPoeeR1xrbNuW1Mpuz0WGOMMZmyHoUxxphMWaIwxhiTKUsUxhhjMmWJwhhjTKYsURhjjMmUJQpjjDGZskRhjDEmU/8P/JdmIlxtmK4AAAAASUVORK5CYII=\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":["## Experiment 2: one dense layer with 256 neurons and two dropouts\n","In the last experiment we can still see overfitting, we try to insert another dropout before the dense layer with 256 neurons."],"metadata":{"id":"3wXjRrGPwaLV"}},{"cell_type":"code","source":["inputs = keras.Input(shape=(IMAGE_HEIGHT, IMAGE_WIDTH, 3))\n","x = layers.Rescaling(1./255)(inputs)\n","x = layers.Conv2D(filters=32, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=64, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=128, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(3,3))(x)\n","x = layers.Conv2D(filters=256, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(5,5))(x)\n","x = layers.Flatten()(x)\n","x = layers.Dropout(0.5)(x)\n","x = layers.Dense(256, activation='relu') (x)\n","x = layers.Dropout(0.5)(x)\n","outputs = layers.Dense(1, activation='sigmoid')(x)\n","\n","model = keras.Model(inputs=inputs, outputs=outputs)\n","compile_model(model)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"lDQuJIYAwfnP","executionInfo":{"status":"ok","timestamp":1640723893103,"user_tz":-60,"elapsed":438,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"3cd750bd-d5e3-49af-9f9a-2c38989bb96e"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Model: \"model\"\n","_________________________________________________________________\n"," Layer (type)                Output Shape              Param #   \n","=================================================================\n"," input_1 (InputLayer)        [(None, 300, 300, 3)]     0         \n","                                                                 \n"," rescaling (Rescaling)       (None, 300, 300, 3)       0         \n","                                                                 \n"," conv2d (Conv2D)             (None, 300, 300, 32)      896       \n","                                                                 \n"," max_pooling2d (MaxPooling2D  (None, 150, 150, 32)     0         \n"," )                                                               \n","                                                                 \n"," conv2d_1 (Conv2D)           (None, 150, 150, 64)      18496     \n","                                                                 \n"," max_pooling2d_1 (MaxPooling  (None, 75, 75, 64)       0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_2 (Conv2D)           (None, 75, 75, 128)       73856     \n","                                                                 \n"," max_pooling2d_2 (MaxPooling  (None, 25, 25, 128)      0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_3 (Conv2D)           (None, 25, 25, 256)       295168    \n","                                                                 \n"," max_pooling2d_3 (MaxPooling  (None, 5, 5, 256)        0         \n"," 2D)                                                             \n","                                                                 \n"," flatten (Flatten)           (None, 6400)              0         \n","                                                                 \n"," dropout (Dropout)           (None, 6400)              0         \n","                                                                 \n"," dense (Dense)               (None, 256)               1638656   \n","                                                                 \n"," dropout_1 (Dropout)         (None, 256)               0         \n","                                                                 \n"," dense_1 (Dense)             (None, 1)                 257       \n","                                                                 \n","=================================================================\n","Total params: 2,027,329\n","Trainable params: 2,027,329\n","Non-trainable params: 0\n","_________________________________________________________________\n"]}]},{"cell_type":"code","source":["run_model(model, \"OneDense256_2dropout\", epochs=30)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"uiKmnaSnxpck","executionInfo":{"status":"ok","timestamp":1640256989571,"user_tz":-60,"elapsed":1959276,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"bfdb2bc5-0d56-4d7a-c35b-5e178885422b"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Epoch 1/30\n","200/200 [==============================] - ETA: 0s - loss: 0.6763 - accuracy: 0.5647\n","Epoch 00001: val_loss improved from inf to 0.62381, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 44s 216ms/step - loss: 0.6763 - accuracy: 0.5647 - val_loss: 0.6238 - val_accuracy: 0.6548\n","Epoch 2/30\n","200/200 [==============================] - ETA: 0s - loss: 0.5846 - accuracy: 0.6925\n","Epoch 00002: val_loss improved from 0.62381 to 0.56114, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.5846 - accuracy: 0.6925 - val_loss: 0.5611 - val_accuracy: 0.7387\n","Epoch 3/30\n","200/200 [==============================] - ETA: 0s - loss: 0.5461 - accuracy: 0.7328\n","Epoch 00003: val_loss did not improve from 0.56114\n","200/200 [==============================] - 44s 217ms/step - loss: 0.5461 - accuracy: 0.7328 - val_loss: 0.6001 - val_accuracy: 0.6904\n","Epoch 4/30\n","200/200 [==============================] - ETA: 0s - loss: 0.5261 - accuracy: 0.7394\n","Epoch 00004: val_loss improved from 0.56114 to 0.45071, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.5261 - accuracy: 0.7394 - val_loss: 0.4507 - val_accuracy: 0.7974\n","Epoch 5/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4842 - accuracy: 0.7666\n","Epoch 00005: val_loss did not improve from 0.45071\n","200/200 [==============================] - 43s 216ms/step - loss: 0.4842 - accuracy: 0.7666 - val_loss: 0.4639 - val_accuracy: 0.7936\n","Epoch 6/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4669 - accuracy: 0.7852\n","Epoch 00006: val_loss did not improve from 0.45071\n","200/200 [==============================] - 43s 215ms/step - loss: 0.4669 - accuracy: 0.7852 - val_loss: 0.4592 - val_accuracy: 0.8035\n","Epoch 7/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4460 - accuracy: 0.7957\n","Epoch 00007: val_loss improved from 0.45071 to 0.42017, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.4460 - accuracy: 0.7957 - val_loss: 0.4202 - val_accuracy: 0.8096\n","Epoch 8/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4191 - accuracy: 0.8039\n","Epoch 00008: val_loss improved from 0.42017 to 0.34586, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 44s 216ms/step - loss: 0.4191 - accuracy: 0.8039 - val_loss: 0.3459 - val_accuracy: 0.8541\n","Epoch 9/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4242 - accuracy: 0.8063\n","Epoch 00009: val_loss did not improve from 0.34586\n","200/200 [==============================] - 43s 215ms/step - loss: 0.4242 - accuracy: 0.8063 - val_loss: 0.3462 - val_accuracy: 0.8499\n","Epoch 10/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3766 - accuracy: 0.8338\n","Epoch 00010: val_loss improved from 0.34586 to 0.32483, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3766 - accuracy: 0.8338 - val_loss: 0.3248 - val_accuracy: 0.8616\n","Epoch 11/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3601 - accuracy: 0.8368\n","Epoch 00011: val_loss improved from 0.32483 to 0.29188, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3601 - accuracy: 0.8368 - val_loss: 0.2919 - val_accuracy: 0.8752\n","Epoch 12/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3528 - accuracy: 0.8449\n","Epoch 00012: val_loss did not improve from 0.29188\n","200/200 [==============================] - 43s 216ms/step - loss: 0.3528 - accuracy: 0.8449 - val_loss: 0.2995 - val_accuracy: 0.8771\n","Epoch 13/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3344 - accuracy: 0.8529\n","Epoch 00013: val_loss did not improve from 0.29188\n","200/200 [==============================] - 44s 217ms/step - loss: 0.3344 - accuracy: 0.8529 - val_loss: 0.3114 - val_accuracy: 0.8635\n","Epoch 14/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3146 - accuracy: 0.8634\n","Epoch 00014: val_loss improved from 0.29188 to 0.28452, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 44s 216ms/step - loss: 0.3146 - accuracy: 0.8634 - val_loss: 0.2845 - val_accuracy: 0.8879\n","Epoch 15/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2932 - accuracy: 0.8735\n","Epoch 00015: val_loss did not improve from 0.28452\n","200/200 [==============================] - 43s 215ms/step - loss: 0.2932 - accuracy: 0.8735 - val_loss: 0.2992 - val_accuracy: 0.8888\n","Epoch 16/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2984 - accuracy: 0.8699\n","Epoch 00016: val_loss improved from 0.28452 to 0.27159, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.2984 - accuracy: 0.8699 - val_loss: 0.2716 - val_accuracy: 0.8818\n","Epoch 17/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2616 - accuracy: 0.8899\n","Epoch 00017: val_loss did not improve from 0.27159\n","200/200 [==============================] - 43s 215ms/step - loss: 0.2616 - accuracy: 0.8899 - val_loss: 0.2873 - val_accuracy: 0.8780\n","Epoch 18/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2691 - accuracy: 0.8866\n","Epoch 00018: val_loss improved from 0.27159 to 0.27083, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 44s 216ms/step - loss: 0.2691 - accuracy: 0.8866 - val_loss: 0.2708 - val_accuracy: 0.8940\n","Epoch 19/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2679 - accuracy: 0.8802\n","Epoch 00019: val_loss improved from 0.27083 to 0.25833, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.2679 - accuracy: 0.8802 - val_loss: 0.2583 - val_accuracy: 0.8959\n","Epoch 20/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2471 - accuracy: 0.8940\n","Epoch 00020: val_loss did not improve from 0.25833\n","200/200 [==============================] - 43s 215ms/step - loss: 0.2471 - accuracy: 0.8940 - val_loss: 0.3068 - val_accuracy: 0.8851\n","Epoch 21/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2365 - accuracy: 0.9034\n","Epoch 00021: val_loss did not improve from 0.25833\n","200/200 [==============================] - 43s 215ms/step - loss: 0.2365 - accuracy: 0.9034 - val_loss: 0.2958 - val_accuracy: 0.8748\n","Epoch 22/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2208 - accuracy: 0.9057\n","Epoch 00022: val_loss did not improve from 0.25833\n","200/200 [==============================] - 43s 215ms/step - loss: 0.2208 - accuracy: 0.9057 - val_loss: 0.3479 - val_accuracy: 0.8766\n","Epoch 23/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2383 - accuracy: 0.8974\n","Epoch 00023: val_loss did not improve from 0.25833\n","200/200 [==============================] - 44s 216ms/step - loss: 0.2383 - accuracy: 0.8974 - val_loss: 0.3425 - val_accuracy: 0.8537\n","Epoch 24/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2165 - accuracy: 0.9115\n","Epoch 00024: val_loss improved from 0.25833 to 0.24445, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 44s 216ms/step - loss: 0.2165 - accuracy: 0.9115 - val_loss: 0.2444 - val_accuracy: 0.8977\n","Epoch 25/30\n","200/200 [==============================] - ETA: 0s - loss: 0.1963 - accuracy: 0.9174\n","Epoch 00025: val_loss did not improve from 0.24445\n","200/200 [==============================] - 44s 216ms/step - loss: 0.1963 - accuracy: 0.9174 - val_loss: 0.2893 - val_accuracy: 0.8982\n","Epoch 26/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2142 - accuracy: 0.9128\n","Epoch 00026: val_loss did not improve from 0.24445\n","200/200 [==============================] - 44s 216ms/step - loss: 0.2142 - accuracy: 0.9128 - val_loss: 0.3175 - val_accuracy: 0.8593\n","Epoch 27/30\n","200/200 [==============================] - ETA: 0s - loss: 0.1954 - accuracy: 0.9215\n","Epoch 00027: val_loss did not improve from 0.24445\n","200/200 [==============================] - 43s 216ms/step - loss: 0.1954 - accuracy: 0.9215 - val_loss: 0.2872 - val_accuracy: 0.8785\n","Epoch 28/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2136 - accuracy: 0.9140\n","Epoch 00028: val_loss improved from 0.24445 to 0.24360, saving model to /content/models/OneDense256_L2reg.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.2136 - accuracy: 0.9140 - val_loss: 0.2436 - val_accuracy: 0.8987\n","Epoch 29/30\n","200/200 [==============================] - ETA: 0s - loss: 0.1812 - accuracy: 0.9226\n","Epoch 00029: val_loss did not improve from 0.24360\n","200/200 [==============================] - 44s 216ms/step - loss: 0.1812 - accuracy: 0.9226 - val_loss: 0.2481 - val_accuracy: 0.9006\n","Epoch 30/30\n","200/200 [==============================] - ETA: 0s - loss: 0.1618 - accuracy: 0.9375\n","Epoch 00030: val_loss did not improve from 0.24360\n","200/200 [==============================] - 44s 216ms/step - loss: 0.1618 - accuracy: 0.9375 - val_loss: 0.3003 - val_accuracy: 0.8785\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"}},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":["evaluate_model(tf.keras.models.load_model(GLOBAL_MODELS_FOLDER + '/OneDense256_2dropout.h5'))"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":741},"id":"ZPc1fNcsftEe","executionInfo":{"status":"ok","timestamp":1640701830027,"user_tz":-60,"elapsed":30647,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"bbe6b3c7-cbdd-4aac-ecd5-834a38a64a6c"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Classification report: \n","              precision    recall  f1-score   support\n","\n","           0     0.8754    0.8376    0.8561       696\n","           1     0.9230    0.9422    0.9325      1437\n","\n","    accuracy                         0.9081      2133\n","   macro avg     0.8992    0.8899    0.8943      2133\n","weighted avg     0.9074    0.9081    0.9076      2133\n","\n"]},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 2 Axes>"]},"metadata":{"needs_background":"light"}},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":["## Experiment 3: one dense layer with 128 neurons and dropout\n","To fight overfitting, we try to decrease the number of neurons in the dense layer"],"metadata":{"id":"pWXCYJ4WJidn"}},{"cell_type":"code","source":["inputs = keras.Input(shape=(IMAGE_HEIGHT, IMAGE_WIDTH, 3))\n","x = layers.Rescaling(1./255)(inputs)\n","x = layers.Conv2D(filters=32, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=64, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=128, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(3,3))(x)\n","x = layers.Conv2D(filters=256, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(5,5))(x)\n","x = layers.Flatten()(x)\n","x = layers.Dropout(0.5)(x)\n","x = layers.Dense(128, activation='relu') (x)\n","x = layers.Dropout(0.5)(x)\n","outputs = layers.Dense(1, activation='sigmoid')(x)\n","\n","model = keras.Model(inputs=inputs, outputs=outputs)\n","compile_model(model)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"AgeHuV3cJqjD","executionInfo":{"status":"ok","timestamp":1641672854364,"user_tz":-60,"elapsed":321,"user":{"displayName":"Daniele Cioffo","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ghy77quHt2uDag8a3PWA5edFZQ2rZw6_PdtwCwLzw=s64","userId":"13448622176976839838"}},"outputId":"7d4844a8-72b7-4780-d6a6-d7ea4719baa4"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Model: \"model\"\n","_________________________________________________________________\n"," Layer (type)                Output Shape              Param #   \n","=================================================================\n"," input_1 (InputLayer)        [(None, 300, 300, 3)]     0         \n","                                                                 \n"," rescaling (Rescaling)       (None, 300, 300, 3)       0         \n","                                                                 \n"," conv2d (Conv2D)             (None, 300, 300, 32)      896       \n","                                                                 \n"," max_pooling2d (MaxPooling2D  (None, 150, 150, 32)     0         \n"," )                                                               \n","                                                                 \n"," conv2d_1 (Conv2D)           (None, 150, 150, 64)      18496     \n","                                                                 \n"," max_pooling2d_1 (MaxPooling  (None, 75, 75, 64)       0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_2 (Conv2D)           (None, 75, 75, 128)       73856     \n","                                                                 \n"," max_pooling2d_2 (MaxPooling  (None, 25, 25, 128)      0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_3 (Conv2D)           (None, 25, 25, 256)       295168    \n","                                                                 \n"," max_pooling2d_3 (MaxPooling  (None, 5, 5, 256)        0         \n"," 2D)                                                             \n","                                                                 \n"," flatten (Flatten)           (None, 6400)              0         \n","                                                                 \n"," dropout (Dropout)           (None, 6400)              0         \n","                                                                 \n"," dense (Dense)               (None, 128)               819328    \n","                                                                 \n"," dropout_1 (Dropout)         (None, 128)               0         \n","                                                                 \n"," dense_1 (Dense)             (None, 1)                 129       \n","                                                                 \n","=================================================================\n","Total params: 1,207,873\n","Trainable params: 1,207,873\n","Non-trainable params: 0\n","_________________________________________________________________\n"]}]},{"cell_type":"code","source":["run_model(model, \"OneDense128_2dropout\", epochs=100)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"QRJYn_hQUtVw","executionInfo":{"status":"ok","timestamp":1641676023052,"user_tz":-60,"elapsed":1833588,"user":{"displayName":"Daniele Cioffo","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ghy77quHt2uDag8a3PWA5edFZQ2rZw6_PdtwCwLzw=s64","userId":"13448622176976839838"}},"outputId":"4b94a328-7749-4c00-e697-c704aa96c98a"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Epoch 1/100\n","200/200 [==============================] - ETA: 0s - loss: 0.6942 - accuracy: 0.4983\n","Epoch 00001: val_loss improved from inf to 0.69427, saving model to /content/models/prova.h5\n","200/200 [==============================] - 73s 299ms/step - loss: 0.6942 - accuracy: 0.4983 - val_loss: 0.6943 - val_accuracy: 0.3419\n","Epoch 2/100\n","200/200 [==============================] - ETA: 0s - loss: 0.6932 - accuracy: 0.4948\n","Epoch 00002: val_loss improved from 0.69427 to 0.69401, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 279ms/step - loss: 0.6932 - accuracy: 0.4948 - val_loss: 0.6940 - val_accuracy: 0.3419\n","Epoch 3/100\n","200/200 [==============================] - ETA: 0s - loss: 0.6881 - accuracy: 0.5261\n","Epoch 00003: val_loss improved from 0.69401 to 0.62614, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 277ms/step - loss: 0.6881 - accuracy: 0.5261 - val_loss: 0.6261 - val_accuracy: 0.6637\n","Epoch 4/100\n","200/200 [==============================] - ETA: 0s - loss: 0.6253 - accuracy: 0.6646\n","Epoch 00004: val_loss improved from 0.62614 to 0.62046, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 276ms/step - loss: 0.6253 - accuracy: 0.6646 - val_loss: 0.6205 - val_accuracy: 0.6979\n","Epoch 5/100\n","200/200 [==============================] - ETA: 0s - loss: 0.5969 - accuracy: 0.6915\n","Epoch 00005: val_loss improved from 0.62046 to 0.54727, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 278ms/step - loss: 0.5969 - accuracy: 0.6915 - val_loss: 0.5473 - val_accuracy: 0.7434\n","Epoch 6/100\n","200/200 [==============================] - ETA: 0s - loss: 0.5763 - accuracy: 0.7051\n","Epoch 00006: val_loss did not improve from 0.54727\n","200/200 [==============================] - 56s 277ms/step - loss: 0.5763 - accuracy: 0.7051 - val_loss: 0.6223 - val_accuracy: 0.7073\n","Epoch 7/100\n","200/200 [==============================] - ETA: 0s - loss: 0.5564 - accuracy: 0.7233\n","Epoch 00007: val_loss did not improve from 0.54727\n","200/200 [==============================] - 56s 277ms/step - loss: 0.5564 - accuracy: 0.7233 - val_loss: 0.9287 - val_accuracy: 0.4334\n","Epoch 8/100\n","200/200 [==============================] - ETA: 0s - loss: 0.5578 - accuracy: 0.7247\n","Epoch 00008: val_loss improved from 0.54727 to 0.50250, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 276ms/step - loss: 0.5578 - accuracy: 0.7247 - val_loss: 0.5025 - val_accuracy: 0.7781\n","Epoch 9/100\n","200/200 [==============================] - ETA: 0s - loss: 0.5179 - accuracy: 0.7556\n","Epoch 00009: val_loss did not improve from 0.50250\n","200/200 [==============================] - 56s 277ms/step - loss: 0.5179 - accuracy: 0.7556 - val_loss: 0.5315 - val_accuracy: 0.7641\n","Epoch 10/100\n","200/200 [==============================] - ETA: 0s - loss: 0.5009 - accuracy: 0.7620\n","Epoch 00010: val_loss did not improve from 0.50250\n","200/200 [==============================] - 56s 276ms/step - loss: 0.5009 - accuracy: 0.7620 - val_loss: 0.5538 - val_accuracy: 0.7458\n","Epoch 11/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4957 - accuracy: 0.7645\n","Epoch 00011: val_loss improved from 0.50250 to 0.47891, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 277ms/step - loss: 0.4957 - accuracy: 0.7645 - val_loss: 0.4789 - val_accuracy: 0.7913\n","Epoch 12/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4904 - accuracy: 0.7644\n","Epoch 00012: val_loss did not improve from 0.47891\n","200/200 [==============================] - 55s 276ms/step - loss: 0.4904 - accuracy: 0.7644 - val_loss: 0.4839 - val_accuracy: 0.7908\n","Epoch 13/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4924 - accuracy: 0.7727\n","Epoch 00013: val_loss improved from 0.47891 to 0.46952, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 277ms/step - loss: 0.4924 - accuracy: 0.7727 - val_loss: 0.4695 - val_accuracy: 0.8030\n","Epoch 14/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4676 - accuracy: 0.7819\n","Epoch 00014: val_loss did not improve from 0.46952\n","200/200 [==============================] - 56s 277ms/step - loss: 0.4676 - accuracy: 0.7819 - val_loss: 0.5022 - val_accuracy: 0.7838\n","Epoch 15/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4585 - accuracy: 0.7903\n","Epoch 00015: val_loss improved from 0.46952 to 0.40968, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 276ms/step - loss: 0.4585 - accuracy: 0.7903 - val_loss: 0.4097 - val_accuracy: 0.8293\n","Epoch 16/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4577 - accuracy: 0.7861\n","Epoch 00016: val_loss improved from 0.40968 to 0.40695, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 277ms/step - loss: 0.4577 - accuracy: 0.7861 - val_loss: 0.4069 - val_accuracy: 0.8293\n","Epoch 17/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4502 - accuracy: 0.7974\n","Epoch 00017: val_loss did not improve from 0.40695\n","200/200 [==============================] - 56s 276ms/step - loss: 0.4502 - accuracy: 0.7974 - val_loss: 0.4815 - val_accuracy: 0.8016\n","Epoch 18/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4587 - accuracy: 0.7886\n","Epoch 00018: val_loss improved from 0.40695 to 0.40615, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 278ms/step - loss: 0.4587 - accuracy: 0.7886 - val_loss: 0.4061 - val_accuracy: 0.8269\n","Epoch 19/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4418 - accuracy: 0.8024\n","Epoch 00019: val_loss did not improve from 0.40615\n","200/200 [==============================] - 56s 278ms/step - loss: 0.4418 - accuracy: 0.8024 - val_loss: 0.4102 - val_accuracy: 0.8283\n","Epoch 20/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4370 - accuracy: 0.8014\n","Epoch 00020: val_loss improved from 0.40615 to 0.39575, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 278ms/step - loss: 0.4370 - accuracy: 0.8014 - val_loss: 0.3958 - val_accuracy: 0.8330\n","Epoch 21/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4334 - accuracy: 0.8068\n","Epoch 00021: val_loss did not improve from 0.39575\n","200/200 [==============================] - 56s 277ms/step - loss: 0.4334 - accuracy: 0.8068 - val_loss: 0.4107 - val_accuracy: 0.8194\n","Epoch 22/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4210 - accuracy: 0.8078\n","Epoch 00022: val_loss did not improve from 0.39575\n","200/200 [==============================] - 56s 277ms/step - loss: 0.4210 - accuracy: 0.8078 - val_loss: 0.3985 - val_accuracy: 0.8340\n","Epoch 23/100\n","200/200 [==============================] - ETA: 0s - loss: 0.4156 - accuracy: 0.8133\n","Epoch 00023: val_loss did not improve from 0.39575\n","200/200 [==============================] - 56s 277ms/step - loss: 0.4156 - accuracy: 0.8133 - val_loss: 0.4037 - val_accuracy: 0.8283\n","Epoch 24/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3978 - accuracy: 0.8252\n","Epoch 00024: val_loss improved from 0.39575 to 0.36465, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 278ms/step - loss: 0.3978 - accuracy: 0.8252 - val_loss: 0.3647 - val_accuracy: 0.8494\n","Epoch 25/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3898 - accuracy: 0.8269\n","Epoch 00025: val_loss did not improve from 0.36465\n","200/200 [==============================] - 56s 277ms/step - loss: 0.3898 - accuracy: 0.8269 - val_loss: 0.3903 - val_accuracy: 0.8283\n","Epoch 26/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3834 - accuracy: 0.8299\n","Epoch 00026: val_loss improved from 0.36465 to 0.36273, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 278ms/step - loss: 0.3834 - accuracy: 0.8299 - val_loss: 0.3627 - val_accuracy: 0.8527\n","Epoch 27/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3848 - accuracy: 0.8294\n","Epoch 00027: val_loss did not improve from 0.36273\n","200/200 [==============================] - 56s 277ms/step - loss: 0.3848 - accuracy: 0.8294 - val_loss: 0.3827 - val_accuracy: 0.8447\n","Epoch 28/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3538 - accuracy: 0.8415\n","Epoch 00028: val_loss improved from 0.36273 to 0.32474, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 278ms/step - loss: 0.3538 - accuracy: 0.8415 - val_loss: 0.3247 - val_accuracy: 0.8720\n","Epoch 29/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3615 - accuracy: 0.8369\n","Epoch 00029: val_loss improved from 0.32474 to 0.31521, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 278ms/step - loss: 0.3615 - accuracy: 0.8369 - val_loss: 0.3152 - val_accuracy: 0.8734\n","Epoch 30/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3527 - accuracy: 0.8476\n","Epoch 00030: val_loss did not improve from 0.31521\n","200/200 [==============================] - 60s 296ms/step - loss: 0.3527 - accuracy: 0.8476 - val_loss: 0.3326 - val_accuracy: 0.8677\n","Epoch 31/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3499 - accuracy: 0.8476\n","Epoch 00031: val_loss did not improve from 0.31521\n","200/200 [==============================] - 56s 278ms/step - loss: 0.3499 - accuracy: 0.8476 - val_loss: 0.3260 - val_accuracy: 0.8720\n","Epoch 32/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3225 - accuracy: 0.8609\n","Epoch 00032: val_loss did not improve from 0.31521\n","200/200 [==============================] - 56s 279ms/step - loss: 0.3225 - accuracy: 0.8609 - val_loss: 0.3311 - val_accuracy: 0.8621\n","Epoch 33/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3158 - accuracy: 0.8651\n","Epoch 00033: val_loss did not improve from 0.31521\n","200/200 [==============================] - 56s 277ms/step - loss: 0.3158 - accuracy: 0.8651 - val_loss: 0.3499 - val_accuracy: 0.8588\n","Epoch 34/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3089 - accuracy: 0.8694\n","Epoch 00034: val_loss improved from 0.31521 to 0.31155, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 278ms/step - loss: 0.3089 - accuracy: 0.8694 - val_loss: 0.3115 - val_accuracy: 0.8757\n","Epoch 35/100\n","200/200 [==============================] - ETA: 0s - loss: 0.3116 - accuracy: 0.8669\n","Epoch 00035: val_loss did not improve from 0.31155\n","200/200 [==============================] - 56s 277ms/step - loss: 0.3116 - accuracy: 0.8669 - val_loss: 0.3268 - val_accuracy: 0.8640\n","Epoch 36/100\n","200/200 [==============================] - ETA: 0s - loss: 0.2843 - accuracy: 0.8813\n","Epoch 00036: val_loss did not improve from 0.31155\n","200/200 [==============================] - 56s 277ms/step - loss: 0.2843 - accuracy: 0.8813 - val_loss: 0.3426 - val_accuracy: 0.8607\n","Epoch 37/100\n","200/200 [==============================] - ETA: 0s - loss: 0.2814 - accuracy: 0.8818\n","Epoch 00037: val_loss improved from 0.31155 to 0.30558, saving model to /content/models/prova.h5\n","200/200 [==============================] - 56s 278ms/step - loss: 0.2814 - accuracy: 0.8818 - val_loss: 0.3056 - val_accuracy: 0.8771\n","Epoch 38/100\n","200/200 [==============================] - ETA: 0s - loss: 0.2732 - accuracy: 0.8827\n","Epoch 00038: val_loss did not improve from 0.30558\n","200/200 [==============================] - 56s 278ms/step - loss: 0.2732 - accuracy: 0.8827 - val_loss: 0.3181 - val_accuracy: 0.8738\n","Epoch 39/100\n","200/200 [==============================] - ETA: 0s - loss: 0.2860 - accuracy: 0.8749\n","Epoch 00039: val_loss did not improve from 0.30558\n","200/200 [==============================] - 56s 278ms/step - loss: 0.2860 - accuracy: 0.8749 - val_loss: 0.3500 - val_accuracy: 0.8621\n","Epoch 40/100\n","200/200 [==============================] - ETA: 0s - loss: 0.2639 - accuracy: 0.8898\n","Epoch 00040: val_loss did not improve from 0.30558\n","200/200 [==============================] - 56s 278ms/step - loss: 0.2639 - accuracy: 0.8898 - val_loss: 0.3419 - val_accuracy: 0.8626\n","Epoch 41/100\n","200/200 [==============================] - ETA: 0s - loss: 0.2528 - accuracy: 0.8931\n","Epoch 00041: val_loss did not improve from 0.30558\n","200/200 [==============================] - 56s 278ms/step - loss: 0.2528 - accuracy: 0.8931 - val_loss: 0.3295 - val_accuracy: 0.8705\n","Epoch 42/100\n","200/200 [==============================] - ETA: 0s - loss: 0.2609 - accuracy: 0.8957\n","Epoch 00042: val_loss did not improve from 0.30558\n","200/200 [==============================] - 56s 278ms/step - loss: 0.2609 - accuracy: 0.8957 - val_loss: 0.3381 - val_accuracy: 0.8677\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"}},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":["evaluate_model(tf.keras.models.load_model(GLOBAL_MODELS_FOLDER + '/OneDense128_2dropout.h5'))"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":731},"id":"rKZWu4cZhGUw","executionInfo":{"status":"ok","timestamp":1641676061667,"user_tz":-60,"elapsed":11520,"user":{"displayName":"Daniele Cioffo","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ghy77quHt2uDag8a3PWA5edFZQ2rZw6_PdtwCwLzw=s64","userId":"13448622176976839838"}},"outputId":"f6f119c2-3ae3-496e-cfe6-7f40a9568f2c"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Classification report: \n","              precision    recall  f1-score   support\n","\n","           0     0.8682    0.6911    0.7696       696\n","           1     0.8638    0.9492    0.9045      1437\n","\n","    accuracy                         0.8650      2133\n","   macro avg     0.8660    0.8201    0.8371      2133\n","weighted avg     0.8653    0.8650    0.8605      2133\n","\n"]},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 2 Axes>"]},"metadata":{"needs_background":"light"}},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":["## Experiment 4: two dense layers \n","We can try to increase the number of dense layers."],"metadata":{"id":"L4Oc_4UcBSlg"}},{"cell_type":"code","source":["inputs = keras.Input(shape=(IMAGE_HEIGHT, IMAGE_WIDTH, 3))\n","x = layers.Rescaling(1./255)(inputs)\n","x = layers.Conv2D(filters=32, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=64, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=128, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(3,3))(x)\n","x = layers.Conv2D(filters=256, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(5,5))(x)\n","x = layers.Flatten()(x)\n","x = layers.Dense(256, activation='relu') (x)\n","x = layers.Dense(256, activation='relu') (x)\n","outputs = layers.Dense(1, activation='sigmoid')(x)\n","\n","model = keras.Model(inputs=inputs, outputs=outputs)\n","compile_model(model)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"4N0fRNQuBYYf","executionInfo":{"status":"ok","timestamp":1641668028546,"user_tz":-60,"elapsed":259,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"d48cd70a-27f2-4624-828c-105518b99dab"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Model: \"model_2\"\n","_________________________________________________________________\n"," Layer (type)                Output Shape              Param #   \n","=================================================================\n"," input_3 (InputLayer)        [(None, 300, 300, 3)]     0         \n","                                                                 \n"," rescaling_2 (Rescaling)     (None, 300, 300, 3)       0         \n","                                                                 \n"," conv2d_8 (Conv2D)           (None, 300, 300, 32)      896       \n","                                                                 \n"," max_pooling2d_8 (MaxPooling  (None, 150, 150, 32)     0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_9 (Conv2D)           (None, 150, 150, 64)      18496     \n","                                                                 \n"," max_pooling2d_9 (MaxPooling  (None, 75, 75, 64)       0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_10 (Conv2D)          (None, 75, 75, 128)       73856     \n","                                                                 \n"," max_pooling2d_10 (MaxPoolin  (None, 25, 25, 128)      0         \n"," g2D)                                                            \n","                                                                 \n"," conv2d_11 (Conv2D)          (None, 25, 25, 256)       295168    \n","                                                                 \n"," max_pooling2d_11 (MaxPoolin  (None, 5, 5, 256)        0         \n"," g2D)                                                            \n","                                                                 \n"," flatten_2 (Flatten)         (None, 6400)              0         \n","                                                                 \n"," dense_4 (Dense)             (None, 256)               1638656   \n","                                                                 \n"," dense_5 (Dense)             (None, 256)               65792     \n","                                                                 \n"," dense_6 (Dense)             (None, 1)                 257       \n","                                                                 \n","=================================================================\n","Total params: 2,093,121\n","Trainable params: 2,093,121\n","Non-trainable params: 0\n","_________________________________________________________________\n"]}]},{"cell_type":"code","source":["run_model(model, \"TwoDense_256\", epochs=30)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"Jdo1EhXTCTc3","executionInfo":{"status":"ok","timestamp":1641669218890,"user_tz":-60,"elapsed":1141210,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"9a8c0d4c-0f98-4857-ee10-99f6015dab13"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Epoch 1/30\n","200/200 [==============================] - ETA: 0s - loss: 0.6550 - accuracy: 0.5907\n","Epoch 00001: val_loss improved from inf to 0.62138, saving model to /content/models/TwoDense_256.h5\n","200/200 [==============================] - 48s 230ms/step - loss: 0.6550 - accuracy: 0.5907 - val_loss: 0.6214 - val_accuracy: 0.7420\n","Epoch 2/30\n","200/200 [==============================] - ETA: 0s - loss: 0.5826 - accuracy: 0.6907\n","Epoch 00002: val_loss did not improve from 0.62138\n","200/200 [==============================] - 47s 235ms/step - loss: 0.5826 - accuracy: 0.6907 - val_loss: 0.6256 - val_accuracy: 0.7036\n","Epoch 3/30\n","200/200 [==============================] - ETA: 0s - loss: 0.5494 - accuracy: 0.7169\n","Epoch 00003: val_loss improved from 0.62138 to 0.56394, saving model to /content/models/TwoDense_256.h5\n","200/200 [==============================] - 48s 240ms/step - loss: 0.5494 - accuracy: 0.7169 - val_loss: 0.5639 - val_accuracy: 0.7434\n","Epoch 4/30\n","200/200 [==============================] - ETA: 0s - loss: 0.5118 - accuracy: 0.7534\n","Epoch 00004: val_loss improved from 0.56394 to 0.47005, saving model to /content/models/TwoDense_256.h5\n","200/200 [==============================] - 48s 236ms/step - loss: 0.5118 - accuracy: 0.7534 - val_loss: 0.4701 - val_accuracy: 0.7805\n","Epoch 5/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4838 - accuracy: 0.7656\n","Epoch 00005: val_loss improved from 0.47005 to 0.43419, saving model to /content/models/TwoDense_256.h5\n","200/200 [==============================] - 48s 239ms/step - loss: 0.4838 - accuracy: 0.7656 - val_loss: 0.4342 - val_accuracy: 0.8105\n","Epoch 6/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4572 - accuracy: 0.7841\n","Epoch 00006: val_loss improved from 0.43419 to 0.41935, saving model to /content/models/TwoDense_256.h5\n","200/200 [==============================] - 48s 239ms/step - loss: 0.4572 - accuracy: 0.7841 - val_loss: 0.4194 - val_accuracy: 0.8157\n","Epoch 7/30\n","200/200 [==============================] - ETA: 0s - loss: 0.4445 - accuracy: 0.7941\n","Epoch 00007: val_loss did not improve from 0.41935\n","200/200 [==============================] - 48s 239ms/step - loss: 0.4445 - accuracy: 0.7941 - val_loss: 0.4447 - val_accuracy: 0.7988\n","Epoch 8/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3983 - accuracy: 0.8175\n","Epoch 00008: val_loss did not improve from 0.41935\n","200/200 [==============================] - 47s 234ms/step - loss: 0.3983 - accuracy: 0.8175 - val_loss: 0.4392 - val_accuracy: 0.8265\n","Epoch 9/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3896 - accuracy: 0.8258\n","Epoch 00009: val_loss did not improve from 0.41935\n","200/200 [==============================] - 49s 241ms/step - loss: 0.3896 - accuracy: 0.8258 - val_loss: 0.4318 - val_accuracy: 0.8152\n","Epoch 10/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3667 - accuracy: 0.8352\n","Epoch 00010: val_loss did not improve from 0.41935\n","200/200 [==============================] - 48s 240ms/step - loss: 0.3667 - accuracy: 0.8352 - val_loss: 0.4256 - val_accuracy: 0.8391\n","Epoch 11/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3492 - accuracy: 0.8490\n","Epoch 00011: val_loss improved from 0.41935 to 0.37439, saving model to /content/models/TwoDense_256.h5\n","200/200 [==============================] - 48s 239ms/step - loss: 0.3492 - accuracy: 0.8490 - val_loss: 0.3744 - val_accuracy: 0.8537\n","Epoch 12/30\n","200/200 [==============================] - ETA: 0s - loss: 0.3188 - accuracy: 0.8580\n","Epoch 00012: val_loss did not improve from 0.37439\n","200/200 [==============================] - 48s 238ms/step - loss: 0.3188 - accuracy: 0.8580 - val_loss: 0.3769 - val_accuracy: 0.8579\n","Epoch 13/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2954 - accuracy: 0.8716\n","Epoch 00013: val_loss did not improve from 0.37439\n","200/200 [==============================] - 47s 235ms/step - loss: 0.2954 - accuracy: 0.8716 - val_loss: 0.4349 - val_accuracy: 0.8574\n","Epoch 14/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2927 - accuracy: 0.8732\n","Epoch 00014: val_loss did not improve from 0.37439\n","200/200 [==============================] - 47s 233ms/step - loss: 0.2927 - accuracy: 0.8732 - val_loss: 0.3997 - val_accuracy: 0.8382\n","Epoch 15/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2526 - accuracy: 0.8934\n","Epoch 00015: val_loss did not improve from 0.37439\n","200/200 [==============================] - 47s 235ms/step - loss: 0.2526 - accuracy: 0.8934 - val_loss: 0.4632 - val_accuracy: 0.8363\n","Epoch 16/30\n","200/200 [==============================] - ETA: 0s - loss: 0.2473 - accuracy: 0.8956\n","Epoch 00016: val_loss did not improve from 0.37439\n","200/200 [==============================] - 47s 234ms/step - loss: 0.2473 - accuracy: 0.8956 - val_loss: 0.4786 - val_accuracy: 0.8213\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"}},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":["evaluate_model(tf.keras.models.load_model(GLOBAL_MODELS_FOLDER + '/TwoDense_256.h5'))"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":741},"id":"j2wWdDf3HVQU","executionInfo":{"status":"ok","timestamp":1641669316355,"user_tz":-60,"elapsed":11041,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"168a1470-d69a-470d-9cee-27a9c9b392d6"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Classification report: \n","              precision    recall  f1-score   support\n","\n","           0     0.7884    0.7227    0.7541       696\n","           1     0.8709    0.9061    0.8881      1437\n","\n","    accuracy                         0.8462      2133\n","   macro avg     0.8297    0.8144    0.8211      2133\n","weighted avg     0.8440    0.8462    0.8444      2133\n","\n"]},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 2 Axes>"]},"metadata":{"needs_background":"light"}},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":["## Experiment 5: two dense layer, the first with less neurons"],"metadata":{"id":"TGKzfiON8Hiq"}},{"cell_type":"code","source":["inputs = keras.Input(shape=(IMAGE_HEIGHT, IMAGE_WIDTH, 3))\n","x = layers.Rescaling(1./255)(inputs)\n","x = layers.Conv2D(filters=32, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=64, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=128, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(3,3))(x)\n","x = layers.Conv2D(filters=256, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(5,5))(x)\n","x = layers.Flatten()(x)\n","x = layers.Dense(128, activation='relu') (x)\n","x = layers.Dense(256, activation='relu') (x)\n","outputs = layers.Dense(1, activation='sigmoid')(x)\n","\n","model = keras.Model(inputs=inputs, outputs=outputs)\n","compile_model(model)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"OzVwn0P98PI-","executionInfo":{"status":"ok","timestamp":1641719321445,"user_tz":-60,"elapsed":217,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"a7331601-d14f-488d-83bd-9e41e28d6300"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Model: \"model_4\"\n","_________________________________________________________________\n"," Layer (type)                Output Shape              Param #   \n","=================================================================\n"," input_5 (InputLayer)        [(None, 300, 300, 3)]     0         \n","                                                                 \n"," rescaling_4 (Rescaling)     (None, 300, 300, 3)       0         \n","                                                                 \n"," conv2d_16 (Conv2D)          (None, 300, 300, 32)      896       \n","                                                                 \n"," max_pooling2d_16 (MaxPoolin  (None, 150, 150, 32)     0         \n"," g2D)                                                            \n","                                                                 \n"," conv2d_17 (Conv2D)          (None, 150, 150, 64)      18496     \n","                                                                 \n"," max_pooling2d_17 (MaxPoolin  (None, 75, 75, 64)       0         \n"," g2D)                                                            \n","                                                                 \n"," conv2d_18 (Conv2D)          (None, 75, 75, 128)       73856     \n","                                                                 \n"," max_pooling2d_18 (MaxPoolin  (None, 25, 25, 128)      0         \n"," g2D)                                                            \n","                                                                 \n"," conv2d_19 (Conv2D)          (None, 25, 25, 256)       295168    \n","                                                                 \n"," max_pooling2d_19 (MaxPoolin  (None, 5, 5, 256)        0         \n"," g2D)                                                            \n","                                                                 \n"," flatten_4 (Flatten)         (None, 6400)              0         \n","                                                                 \n"," dense_12 (Dense)            (None, 128)               819328    \n","                                                                 \n"," dense_13 (Dense)            (None, 256)               33024     \n","                                                                 \n"," dense_14 (Dense)            (None, 1)                 257       \n","                                                                 \n","=================================================================\n","Total params: 1,241,025\n","Trainable params: 1,241,025\n","Non-trainable params: 0\n","_________________________________________________________________\n"]}]},{"cell_type":"code","source":["run_model(model, \"TwoDense_128_256\", epochs=50)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"UZtmbIzb88EQ","executionInfo":{"status":"ok","timestamp":1641720241077,"user_tz":-60,"elapsed":915961,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"049a80d2-ba33-41ee-8ba2-188a41948583"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Epoch 1/50\n","200/200 [==============================] - ETA: 0s - loss: 0.6911 - accuracy: 0.5167\n","Epoch 00001: val_loss improved from inf to 0.60424, saving model to /content/models/TwoDense_128_256.h5\n","200/200 [==============================] - 44s 216ms/step - loss: 0.6911 - accuracy: 0.5167 - val_loss: 0.6042 - val_accuracy: 0.7369\n","Epoch 2/50\n","200/200 [==============================] - ETA: 0s - loss: 0.6322 - accuracy: 0.6420\n","Epoch 00002: val_loss improved from 0.60424 to 0.57427, saving model to /content/models/TwoDense_128_256.h5\n","200/200 [==============================] - 43s 214ms/step - loss: 0.6322 - accuracy: 0.6420 - val_loss: 0.5743 - val_accuracy: 0.7575\n","Epoch 3/50\n","200/200 [==============================] - ETA: 0s - loss: 0.5956 - accuracy: 0.6843\n","Epoch 00003: val_loss improved from 0.57427 to 0.55075, saving model to /content/models/TwoDense_128_256.h5\n","200/200 [==============================] - 43s 213ms/step - loss: 0.5956 - accuracy: 0.6843 - val_loss: 0.5508 - val_accuracy: 0.7547\n","Epoch 4/50\n","200/200 [==============================] - ETA: 0s - loss: 0.5837 - accuracy: 0.6942\n","Epoch 00004: val_loss did not improve from 0.55075\n","200/200 [==============================] - 43s 214ms/step - loss: 0.5837 - accuracy: 0.6942 - val_loss: 0.5775 - val_accuracy: 0.7242\n","Epoch 5/50\n","200/200 [==============================] - ETA: 0s - loss: 0.5672 - accuracy: 0.7067\n","Epoch 00005: val_loss improved from 0.55075 to 0.53796, saving model to /content/models/TwoDense_128_256.h5\n","200/200 [==============================] - 43s 212ms/step - loss: 0.5672 - accuracy: 0.7067 - val_loss: 0.5380 - val_accuracy: 0.7594\n","Epoch 6/50\n","200/200 [==============================] - ETA: 0s - loss: 0.5499 - accuracy: 0.7219\n","Epoch 00006: val_loss did not improve from 0.53796\n","200/200 [==============================] - 43s 214ms/step - loss: 0.5499 - accuracy: 0.7219 - val_loss: 0.6727 - val_accuracy: 0.6379\n","Epoch 7/50\n","200/200 [==============================] - ETA: 0s - loss: 0.5217 - accuracy: 0.7427\n","Epoch 00007: val_loss did not improve from 0.53796\n","200/200 [==============================] - 43s 213ms/step - loss: 0.5217 - accuracy: 0.7427 - val_loss: 0.5646 - val_accuracy: 0.7416\n","Epoch 8/50\n","200/200 [==============================] - ETA: 0s - loss: 0.4938 - accuracy: 0.7624\n","Epoch 00008: val_loss did not improve from 0.53796\n","200/200 [==============================] - 43s 212ms/step - loss: 0.4938 - accuracy: 0.7624 - val_loss: 0.5754 - val_accuracy: 0.7237\n","Epoch 9/50\n","200/200 [==============================] - ETA: 0s - loss: 0.4581 - accuracy: 0.7861\n","Epoch 00009: val_loss improved from 0.53796 to 0.44332, saving model to /content/models/TwoDense_128_256.h5\n","200/200 [==============================] - 43s 213ms/step - loss: 0.4581 - accuracy: 0.7861 - val_loss: 0.4433 - val_accuracy: 0.8180\n","Epoch 10/50\n","200/200 [==============================] - ETA: 0s - loss: 0.4329 - accuracy: 0.8014\n","Epoch 00010: val_loss did not improve from 0.44332\n","200/200 [==============================] - 43s 214ms/step - loss: 0.4329 - accuracy: 0.8014 - val_loss: 0.5211 - val_accuracy: 0.7631\n","Epoch 11/50\n","200/200 [==============================] - ETA: 0s - loss: 0.4021 - accuracy: 0.8155\n","Epoch 00011: val_loss did not improve from 0.44332\n","200/200 [==============================] - 43s 214ms/step - loss: 0.4021 - accuracy: 0.8155 - val_loss: 0.6553 - val_accuracy: 0.7111\n","Epoch 12/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3554 - accuracy: 0.8454\n","Epoch 00012: val_loss did not improve from 0.44332\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3554 - accuracy: 0.8454 - val_loss: 0.4514 - val_accuracy: 0.8213\n","Epoch 13/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3540 - accuracy: 0.8452\n","Epoch 00013: val_loss did not improve from 0.44332\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3540 - accuracy: 0.8452 - val_loss: 0.5122 - val_accuracy: 0.7988\n","Epoch 14/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3131 - accuracy: 0.8694\n","Epoch 00014: val_loss did not improve from 0.44332\n","200/200 [==============================] - 43s 214ms/step - loss: 0.3131 - accuracy: 0.8694 - val_loss: 0.4901 - val_accuracy: 0.8171\n"]},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAXoAAAEICAYAAABRSj9aAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deXyU1dXA8d9hN+ybG5EAyiIWIRBRUCqIUhQLggqBKCBV1IpWXlvFF6tUS19rraitC7gEFTQgRYoKWhQVd4kIlkWQVYKKkX0ny3n/uE/CJGSZSWbP+X4+85l57jzLmcnkzJ1773MfUVWMMcbEr2qRDsAYY0xoWaI3xpg4Z4neGGPinCV6Y4yJc5bojTEmzlmiN8aYOGeJvgoSkYUiMirY60aSiGwWkYtDsF8VkTO8x0+LyB/9WbcCx0kTkf9UNE5jyiI2jj42iMh+n8UE4AiQ5y3fqKozwx9V9BCRzcD1qvpOkPerQFtVXR+sdUWkFbAJqKmqucGI05iy1Ih0AMY/qlqv4HFZSU1EaljyMNHCPo/RwZpuYpyI9BaRLBG5S0R+BNJFpLGIvCEi2SKyy3uc6LPN+yJyvfd4tIh8JCIPe+tuEpFLK7huaxFZIiL7ROQdEXlCRGaUErc/MT4gIh97+/uPiDTzef5aEdkiIjtEZGIZ78+5IvKjiFT3KRssIl97j7uLyKcisltEfhCRf4pIrVL2NV1E/uyz/Advm+9FZEyxdQeIyFcisldEtorIJJ+nl3j3u0Vkv4j0KHhvfbbvKSJLRWSPd9/T3/cmwPe5iYike69hl4jM83lukIgs917DBhHp75UXaSYTkUkFf2cRaeU1Yf1GRL4DFnvlr3p/hz3eZ+Qsn+1PEJG/e3/PPd5n7AQReVNEbi32er4WkcElvVZTOkv08eFkoAmQBIzF/V3TveWWwCHgn2Vsfy6wFmgGPAQ8JyJSgXVfBr4AmgKTgGvLOKY/MY4ArgNOBGoBvwcQkY7AU97+T/WOl0gJVPVz4ABwUbH9vuw9zgPGe6+nB9AX+G0ZcePF0N+L5xKgLVC8f+AAMBJoBAwAbhaRK7znfundN1LVeqr6abF9NwHeBB73XtsjwJsi0rTYazjuvSlBee/zS7imwLO8fU3xYugOvAj8wXsNvwQ2l/Z+lOBC4EzgV97yQtz7dCKwDPBtanwY6Ab0xH2O7wTygReAawpWEpHOQAvce2MCoap2i7Eb7h/uYu9xb+AoUKeM9bsAu3yW38c1/QCMBtb7PJcAKHByIOvikkgukODz/Axghp+vqaQY7/FZ/i3wlvf4XiDD57m63ntwcSn7/jPwvPe4Pi4JJ5Wy7u3Aaz7LCpzhPZ4O/Nl7/DzwoM967XzXLWG/jwJTvMetvHVr+Dw/GvjIe3wt8EWx7T8FRpf33gTyPgOn4BJq4xLWm1oQb1mfP295UsHf2ee1tSkjhkbeOg1xX0SHgM4lrFcH2IXr9wD3hfBkuP/f4uFmNfr4kK2qhwsWRCRBRKZ6P4X34poKGvk2XxTzY8EDVT3oPawX4LqnAjt9ygC2lhawnzH+6PP4oE9Mp/ruW1UPADtKOxau9j5ERGoDQ4BlqrrFi6Od15zxoxfHX3C1+/IUiQHYUuz1nSsi73lNJnuAm/zcb8G+txQr24KrzRYo7b0popz3+TTc32xXCZueBmzwM96SFL43IlJdRB70mn/2cuyXQTPvVqekY3mf6VnANSJSDRiO+wViAmSJPj4UHzp1B9AeOFdVG3CsqaC05phg+AFoIiIJPmWnlbF+ZWL8wXff3jGblrayqq7GJcpLKdpsA64J6BtcrbEB8L8ViQH3i8bXy8B84DRVbQg87bPf8oa6fY9ravHVEtjmR1zFlfU+b8X9zRqVsN1W4PRS9nkA92uuwMklrOP7GkcAg3DNWw1xtf6CGH4GDpdxrBeANFyT2kEt1sxl/GOJPj7Vx/0c3u21994X6gN6NeRMYJKI1BKRHsCvQxTjHOByEbnA6zi9n/I/yy8Dv8MluleLxbEX2C8iHYCb/YxhNjBaRDp6XzTF46+Pqy0f9tq7R/g8l41rMmlTyr4XAO1EZISI1BCRYUBH4A0/YyseR4nvs6r+gGs7f9LrtK0pIgVfBM8B14lIXxGpJiItvPcHYDmQ6q2fAlzlRwxHcL+6EnC/mgpiyMc1gz0iIqd6tf8e3q8vvMSeD/wdq81XmCX6+PQocAKutvQZ8FaYjpuG69DcgWsXn4X7By9JhWNU1VXALbjk/QOuHTernM1ewXUQLlbVn33Kf49LwvuAZ7yY/YlhofcaFgPrvXtfvwXuF5F9uD6F2T7bHgQmAx+LG+1zXrF97wAux9XGd+A6Jy8vFre/ynufrwVycL9qfsL1UaCqX+A6e6cAe4APOPYr44+4Gvgu4E8U/YVUkhdxv6i2Aau9OHz9HvgvsBTYCfyVornpRaATrs/HVICdMGVCRkRmAd+oash/UZj4JSIjgbGqekGkY4lVVqM3QSMi54jI6d5P/f64dtl55W1nTGm8ZrHfAtMiHUsss0Rvgulk3NC//bgx4Der6lcRjcjELBH5Fa4/YzvlNw+ZMljTjTHGxDmr0RtjTJyLuknNmjVrpq1atYp0GMYYE1O+/PLLn1W1eUnPRV2ib9WqFZmZmZEOwxhjYoqIFD+bupA13RhjTJyzRG+MMXHOEr0xxsS5qGujL0lOTg5ZWVkcPny4/JVNlVCnTh0SExOpWbNmpEMxJurFRKLPysqifv36tGrVitKvh2GqClVlx44dZGVl0bp160iHY0zUi4mmm8OHD9O0aVNL8gYAEaFp06b2C8/EjZkzoVUrqFbN3c+cWd4WgYmJGj1gSd4UYZ8HEy9mzoSxY+Ggd8meLVvcMkBaWnCOERM1emOMiVcTJx5L8gUOHnTlwWKJ3g87duygS5cudOnShZNPPpkWLVoULh89erTMbTMzM7ntttvKPUbPnj2DFa4xJoZ8911g5RURl4k+2O1dTZs2Zfny5SxfvpybbrqJ8ePHFy7XqlWL3NzcUrdNSUnh8ccfL/cYn3zySeWCjIC8vLxIh2BMzGtZ/CKU5ZRXRNwl+oL2ri1bQPVYe1ewOzdGjx7NTTfdxLnnnsudd97JF198QY8ePUhOTqZnz56sXbsWgPfff5/LL78cgEmTJjFmzBh69+5NmzZtinwB1KtXr3D93r17c9VVV9GhQwfS0tIomGF0wYIFdOjQgW7dunHbbbcV7tfX5s2b6dWrF127dqVr165FvkD++te/0qlTJzp37syECRMAWL9+PRdffDGdO3ema9eubNiwoUjMAOPGjWP69OmAm6LirrvuomvXrrz66qs888wznHPOOXTu3Jkrr7ySg95v0O3btzN48GA6d+5M586d+eSTT7j33nt59NFHC/c7ceJEHnvssUr/LYyJZZMnQ0JC0bKEBFceNKoaVbdu3bppcatXrz6urDRJSaouxRe9JSX5vYsy3Xffffq3v/1NR40apQMGDNDc3FxVVd2zZ4/m5OSoquqiRYt0yJAhqqr63nvv6YABAwq37dGjhx4+fFizs7O1SZMmevToUVVVrVu3buH6DRo00K1bt2peXp6ed955+uGHH+qhQ4c0MTFRN27cqKqqqamphfv1deDAAT106JCqqq5bt04L3s8FCxZojx499MCBA6qqumPHDlVV7d69u86dO1dVVQ8dOqQHDhwoErOq6i233KLp6emqqpqUlKR//etfC5/7+eefCx9PnDhRH3/8cVVVHTp0qE6ZMkVVVXNzc3X37t26adMmTU5OVlXVvLw8bdOmTZHtAxXI58KYaDZjhstRIu5+xozA9wFkail5NWZG3fgrHO1dBa6++mqqV68OwJ49exg1ahTffvstIkJOTk6J2wwYMIDatWtTu3ZtTjzxRLZv305iYmKRdbp3715Y1qVLFzZv3ky9evVo06ZN4bjx4cOHM23a8RfdycnJYdy4cSxfvpzq1auzbt06AN555x2uu+46EryqQ5MmTdi3bx/btm1j8ODBgDsJyR/Dhg0rfLxy5Uruuecedu/ezf79+/nVr34FwOLFi3nxxRcBqF69Og0bNqRhw4Y0bdqUr776iu3bt5OcnEzTpk39OqYx8SwtLXgjbEoSd4m+ZUvXXFNSebDVrVu38PEf//hH+vTpw2uvvcbmzZvp3bt3idvUrl278HH16tVLbN/3Z53STJkyhZNOOokVK1aQn5/vd/L2VaNGDfLz8wuXi49X933do0ePZt68eXTu3Jnp06fz/vvvl7nv66+/nunTp/Pjjz8yZsyYgGMzxgQu7trow9LeVYI9e/bQokULgML27GBq3749GzduZPPmzQDMmjWr1DhOOeUUqlWrxksvvVTYYXrJJZeQnp5e2Ia+c+dO6tevT2JiIvPmucu6HjlyhIMHD5KUlMTq1as5cuQIu3fv5t133y01rn379nHKKaeQk5PDTJ+OkL59+/LUU08BrtN2z549AAwePJi33nqLpUuXFtb+jTGhFXeJPi0Npk2DpCQQcffTpoX2ZxHAnXfeyd13301ycnJANXB/nXDCCTz55JP079+fbt26Ub9+fRo2bHjcer/97W954YUX6Ny5M998801h7bt///4MHDiQlJQUunTpwsMPPwzASy+9xOOPP87ZZ59Nz549+fHHHznttNMYOnQov/jFLxg6dCjJycmlxvXAAw9w7rnncv7559OhQ4fC8scee4z33nuPTp060a1bN1avXg1ArVq16NOnD0OHDi1s9jLGhJZf14wVkf7AY0B14FlVfbDY8y2BF4BG3joTVHWBiLQC1gBrvVU/U9WbyjpWSkqKFr/wyJo1azjzzDP9eT1xbf/+/dSrVw9V5ZZbbqFt27aMHz8+0mEFJD8/v3DETtu2bSu1L/tcGHOMiHypqiklPVdujV5EqgNPAJcCHYHhItKx2Gr3ALNVNRlIBZ70eW6DqnbxbmUmeVO2Z555hi5dunDWWWexZ88ebrzxxkiHFJDVq1dzxhln0Ldv30oneWOM//zpjO0OrFfVjQAikgEMAlb7rKNAA+9xQ+D7YAZpnPHjx8dcDd5Xx44d2bhxY6TDMKbK8aeNvgWw1Wc5yyvzNQm4RkSygAXArT7PtRaRr0TkAxHpVdIBRGSsiGSKSGZ2drb/0RtjjCfUM0CGev+hFKzO2OHAdFVNBC4DXhKRasAPQEuvSed/gJdFpEHxjVV1mqqmqGpK8+YlXsTcGGNKFeoz4sN1xn2o+JPotwGn+SwnemW+fgPMBlDVT4E6QDNVPaKqO7zyL4ENQLvKBm2MMb5CPQNkOGaYDCV/Ev1SoK2ItBaRWrjO1vnF1vkO6AsgImfiEn22iDT3OnMRkTZAW8AaaY0xQRXqM+LDecZ9KJSb6FU1FxgHvI0bKjlbVVeJyP0iMtBb7Q7gBhFZAbwCjPbmXvgl8LWILAfmADep6s5QvJBQ6tOnD2+//XaRskcffZSbb7651G169+5NwTDRyy67jN27dx+3zqRJkwrHs5dm3rx5hWPQAe69917eeeedQMI3Ju6FegbIcMwwGUp+tdGr6gJVbaeqp6vqZK/sXlWd7z1erarnq2pnbxjlf7zyf6nqWV5ZV1V9PXQvJXSGDx9ORkZGkbKMjAyGDx/u1/YLFiygUaNGFTp28UR///33c/HFF1doX5Fi0xmbUAv1GfGROuM+WOLuzNhQuOqqq3jzzTcLLzKyefNmvv/+e3r16sXNN99MSkoKZ511Fvfdd1+J27dq1Yqff/4ZgMmTJ9OuXTsuuOCCwqmMgRKn+/3kk0+YP38+f/jDH+jSpQsbNmxg9OjRzJkzB4B3332X5ORkOnXqxJgxYzhy5Ejh8e677z66du1Kp06d+Oabb46LyaYzNvEk1GfER+qM+2CJuUnNbr8dli8P7j67dAGfvHKcJk2a0L17dxYuXMigQYPIyMhg6NChiAiTJ0+mSZMm5OXl0bdvX77++mvOPvvsEvfz5ZdfkpGRwfLly8nNzaVr165069YNgCFDhnDDDTcAcM899/Dcc89x6623MnDgQC6//HKuuuqqIvs6fPgwo0eP5t1336Vdu3aMHDmSp556ittvvx2AZs2asWzZMp588kkefvhhnn322SLbn3jiiSxatIg6derw7bffMnz4cDIzM1m4cCH//ve/+fzzz0lISGDnTtfSlpaWxoQJExg8eDCHDx8mPz+frVu3UpamTZuybNkywF2lq6TXd9ttt3HhhRfy2muvkZeXx/79+zn11FMZMmQIt99+O/n5+WRkZPDFF1+UeSxjQj0DZKj3H0pWo/eTb/ONb7PN7Nmz6dq1K8nJyaxatapIM0txH374IYMHDyYhIYEGDRowcODAwudWrlxJr1696NSpEzNnzmTVqlVlxrN27Vpat25Nu3ZuENOoUaNYsmRJ4fNDhgwBoFu3boUTofnKycnhhhtuoFOnTlx99dWFcfs7nXFC8d+xJSg+nXFJr2/x4sWFfR0F0xm3atWqcDrj//znPzadsTGVFHM1+rJq3qE0aNAgxo8fz7Jlyzh48CDdunVj06ZNPPzwwyxdupTGjRszevTo46b09Veg0/2Wp2Cq49KmObbpjI2pOqxG76d69erRp08fxowZU1ib37t3L3Xr1qVhw4Zs376dhQsXlrmPX/7yl8ybN49Dhw6xb98+Xn/9WN90adP91q9fn3379h23r/bt27N582bWr18PuFkoL7zwQr9fj01nbEzVYYk+AMOHD2fFihWFib5z584kJyfToUMHRowYwfnnn1/m9l27dmXYsGF07tyZSy+9lHPOOafwudKm+01NTeVvf/sbycnJbNiwobC8Tp06pKenc/XVV9OpUyeqVavGTTf5P2ecTWdsTNXh1zTF4WTTFBvwbzpj+1wYc0ylpik2JtxsOmNjgivmOmNN/LPpjI0Jrpip0UdbE5OJLPs8mJKsXw/PPguHDkU6kugSE4m+Tp067Nixw/65DeCS/I4dOyo0JNREVqjndL/9drjhBmjbFp57DkJw+eaYFBOdsTk5OWRlZVV4jLqJP3Xq1CExMZGaNWtGOpSIeOklqFkTUlMjHYn/CuZ0953uNyEheFMJ/PADJCbC4MGwbRt89hmceSb83//BwIFu6oJotm8fZGW5mCuirM5YVDWqbt26dVNjTOl++km1dm3VatVUFy+OdDT+S0pSdZftKHpLSgrO/v/6V7e/tWtV8/NV585Vbd/elfXsqfrRR8E5TrBt3Kg6frxqgwaqXbq42CsCyNRS8mpMNN0YY4556ik4cgROOw2GD3c12VgQyjndVSE9HXr2hHbtXO198GBYudL9Yti0CS64AAYNgjJmKQkbVViyBIYMgTPOgH/8AwYMgKlTQ/PLwxK9MTHkyBF48km49FJ44w3Yu9cl+1hoiw7lnO6ffw7ffAPXXVe0vEYN12a/fj385S/w/vvQqRP85jeumSTcjhxxzW4pKXDhhfDBB3DXXe6L6OWXoXv30BzXEr0xMeSVV2D7dhg/Hn7xC3j6aZcs7r030pGVL5RzuqenwwknwNChJT+fkAB33w0bN7oO2xkzXIftXXfBrl2VP355fvoJHnjAdUCPHOlGBU2dClu3ui+gxMQQB1Bam06kbtZGb0zJ8vNVO3VS/cUvirbjXn+9a4d+443IxeavGTNcm7yIu58xo/L7PHDAtW9fe63/22zerDpypIujUSPVhx5SPXiw8rEUt2KF6pgxrk8FVC+9VPXttyveDl8Wymijj3hiL36zRG9Myd55x/3HPvdc0fKDB10nXuPGLoFVNTNmuPelIh3TK1aoXnaZ2z4xUfX551VzcysXT16e6vz5qhdd5PZ7wgmqN9+sumZN5fZbnrISvTXdGBMjHnkETjwRRowoWn7CCfDqq66dftgw8C6EViGhHuceCunpLtYAJm8tdPbZ8Oab8N57cOqpMGYMdO4Mr7/uOkwDsX8//POf0L69G865bh08+KDrC3jySfCZyy/8SvsG8L0B/YG1wHpgQgnPtwTeA74CvgYu83nubm+7tcCvyjuW1eiNOd6aNa52OGlS6evMmePWue22ih1jxgzVhAQtMvQxISE4zSuhsnmza34p633xV36+ew/btXOv/YILVD/+uPztNm1SveMO1YYN3XbnnaeakaF69GjlYwoElWm6AaoDG4A2QC1gBdCx2DrTgJu9xx2BzT6PVwC1gdbefqqXdTxL9MYc76abXDvv9u1lr/e737n/6tmzAz9GqMe5h8Kf/uRi3LQpePs8elT16adVTz7Z7fuKK1RXry66Tn6+6ocfql55pTufoXp11dRU1U8/DV4cgSor0fvTdNMdWK+qG1X1KJABDCr+wwBo4D1uCHzvPR4EZKjqEVXd5NXsQzSAyJj4tGMHvPACXHONa7opy0MPwXnnueGD69YFdpxQjnMPhfx8mD4dLrrINd0ES82acOONbkjmn/8M777rRjjdcIMbBjljBpxzDvTqBYsXw513uvJXXnHvfTTyJ9G3AHyvAp3llfmaBFwjIlnAAuDWALZFRMaKSKaIZGZnZ/sZujFVw9Spbjied933MtWqBbNmuWR19dWBTe4VynHuobBkiUuwxcfOB0vdujBxohuSedtt7su2TRu49lo4cMANbd261U2xcNppoYkhWILVGTscmK6qicBlwEsi4ve+VXWaqqaoakrz5s2DFJIxse/oUdfB16+fq1X6o2VLV+v8+msYN87/Y4VynHsopKdDgwbu7NJQatYMpkyBtWvhvvtg4UJYtcrV+n0uixzV/EnG2wDf76tEr8zXb4DZAKr6KVAHaObntsaYUsya5aY4GD8+sO0uvdTVRp9/3jVv+CMtzU0XkJTkTsNPSgrehGPBtm8fzJnjRhkV/3IKldatYdIk6N/fjUqKJeXOXikiNYB1QF9ckl4KjFDVVT7rLARmqep0ETkTeBfXRNMReBnXLn+qV95WVfNKO15Js1caUxWpQteu7rT5VasCnwMlLw8uucTN4vj55+7U/3jx3HNw/fXwySfQo0eko4kOlbqUoKrmAuOAt4E1wGxVXSUi94vIQG+1O4AbRGQF8Aow2usIXoWr6a8G3gJuKSvJG2OO+eADWL7c1eYrMtFV9epu/pSGDeGqq1wtOF6kp7vx6tHa+Rlt/PoBoqoLVLWdqp6uqpO9sntVdb73eLWqnq+qnVW1i6r+x2fbyd527VV1YWhehjHxZ8oU1z58zTVFywM5qenkk91okPXr3aiRQE8Cikbr1sHHH7tO2GifYz5axFhLkzFVw7ffurMzb77ZnflaoODiHVu2uKS9ZYtbLivZ9+7thgnOmuXO0Ix106e7L7lrr410JLEjJq4wZUxVM24cPPOMS+Qnn3ysvFUrV1ZcUhJs3lz6/vLz4de/hkWLXG34nHOCHXF45OW519q5s5u6wBxTqTZ6Y0x47dzp2qBHjCia5KHiJzVVqwYvvginnOLG1+/cGZxYw23RIneZwFCNnY9XluiNiTLPPOOuq1rSCVKVOampaVM3+dn338OoUa6WH2vS06FJE/frxPjPEr0xUSQnx11Wrm9f1zxRXGVPaureHf7+d3d1qr/9rfLxhtPOnTBvnhvXX7t2pKOJLZbojYkir77qmiZKO0EqGCc1jRvnmm8mTnTTCMSKV15xZwpbs03grDPWmCih6mrc+/a5C1iH8uzLvXvddUv374evvoKTTgrdsYIlJcXNub98eaQjiU7WGWtMDPjoI8jMdG3zoT7FvkEDN4XArl2u0zcvyk9j/O9/4csvrTZfUZbojYkSU6a4jsaRI8NzvLPPhieecFPt/ulP4TlmRaWnuxk5o3HenVhgid6YKLBhg+tovOmm8E3SBe7SeaNHuxOq3n47fMcNRE6Om43z1792ZwqbwFmiNyYKPP441KgBt9wS/mM/8YSbAjktzc2vHm3efBOys63ZpjIs0RsTYbt3u+mEU1PdBarDLSHBjfY5csRN+5uTE/4YypKe7k4c698/0pHELkv0xoRAIBOPPfusG/0S6JzzwdS+vYvj00/hrrsiF0dx27e7Gv3Ike4Xj6kYe+uMCbKCiccOHnTLBROPwfGdibm5rtnmwgshOTm8cRY3bBh8+KHrFL7ggtBfuckfM2a4EUHWbFM5No7emCALZOKxWbNck82//w0DBx6/TbgdOeIuer12LSxdCu3aRS4WVXexlPr13S8NUzYbR29MGPk78ZgqPPIInHEGXH556OPyR+3arr2+Zk244orIXqwkM9NdWctq85Vnid5UyowZ7if/kSORjiR6+Dvx2KefwhdfhOcEqUAkJcHs2e4CHyNHRm7ys/R0Nxf/sGGROX48sTZ6P2RlwcMPw4oV0LjxsVuTJqUvN2rkLuUWr/Lz4Z574P/+zy2npUVH00M0mDy5aBs9lDzx2JQp7nMyalR44/PHRRe5Sc/+539c3H/8Y3iPf/iwm9tmyBB3KURTOZboy/Ddd/Dgg+5CxPn5bq6NdevcaeO7dsGhQ2Vv37Ch/18Mp54KHTuG53VV1oEDrqY3d667PN3cuZCRYYm+QEGH68SJ7jPUsqVLlr4dsZs3u/ftD3+AevUiEma5br8dli2De++FLl3COzXwvHlu2Kk12wSHX4leRPoDjwHVgWdV9cFiz08B+niLCcCJqtrIey4P+K/33HeqGvXpYPNml+Cff94tjxkDEya4TjZfhw8fS/q7drlpVEtb3rkTVq48tnz06PHHveYad6m3+vVD/Qorbts2l9CXL3c10t/9zjU7vPSS+wKoWzfSEUaHtLSyT9d//HH3vo0bF76YAiXiZsZcvdp9Nr/4wg3DDIf0dNeE1KdP+esaP6hqmTdcct8AtAFqASuAjmWsfyvwvM/y/vKO4Xvr1q2bRsqGDarXX69ao4ZqrVqqN9+sumVL8I+Tn6+6f7/q1q2qX3+t+sEHqvfco1qtmuoZZ6hmZgb/mMGQmal66qmq9eqpvvHGsfL33lMF1VmzIhZaTNmzR7V+fdURIyIdiX+2bFFt3ly1fXvV3bvDczwR1XvvDf2x4gmQqaXl5dKe0GOJugfwts/y3cDdZaz/CXCJz3LUJ/pvv1W97jrV6tVVa9dWvfVWl4TDbckS1cRE1Zo1Vf/+d9W8vPDHUFYMsS0AABmiSURBVJp//Uv1hBNUW7ZUXbGi6HO5uaqnnKI6eHBkYos1jzzi/vOWLo10JP577z33//HrX4f+c/nAA+792bgxtMeJN5VN9FfhmmsKlq8F/lnKuknAD0B1n7JcIBP4DLiilO3GeutktmzZMjzviqquXat67bWuJl2njurtt6tu2xa2w5doxw7VK65wf5lLL1Xdvj2y8eTnq/7lLy6e885T/fHHktf73e/cl2Q4anyxLDdXtVUr1QsuiHQkgXv8cfc5uO++0B0jP1/19NNVe/cO3THiVTgT/V3AP4qVtfDu2wCbgdPLOl44avSrV7ufzdWquVrqHXeo/vBDyA/rt/x81SeecInz5JNVFy2KTByHD6uOHOk+JcOHqx46VPq6n37q1ps+PXzxVcaMGapJSa6JICnJLYfDnDnufZo7NzzHC6b8fNVRo1z8r70WmmN88IHb/wsvhGb/8SxsTTfAV0DPMvY1HbiqrOOFMtGvXKk6bJj7565bV/XOOyNfYy7L11+rnnmmi3fCBNWjR8N37J9+crVOUP3Tn9w/eVny811NtX//8MRXGTNmqCYkuNdWcEtICE+y79lTtU0bV7OPRYcOqZ5zjuunWb06+PsfPdr1X+zfH/x9x7vKJvoawEagtU9n7FklrNfBq7GLT1ljoLb3uBnwbVkduRqiRL9ihepVV7lXW6+e6t13q2ZnB/0wIXHggOoNN7jYzz03PO2Wq1aptm7tmrMyMvzfbsIE144b7e9tUlLRJF9wS0oK7XE/+8wd57HHQnucUNu6VfXEE1XbtlXdtSt4+923z1XAfvOb4O2zKqlUonfbcxmwzht9M9Erux8Y6LPOJODBYtv1xA2tXOHd/6a8YwUz0S9b5joIQbVBAzey5eefg7b7sJo9W7VhQ/c6Akm+gXrrLXeMk05yiSkQy5e79/rpp0MTW7CIlJzoRUJ73GHD3Hu7d29ojxMOS5a40WkDBgSvc/b5593f4aOPgrO/qqbSiT6ct2Ak+sxM1YED3atr2NB1Hu3cWendRtymTao9erjXdf31wf95+49/uH6Ls8+u2LDS/HzVDh2ivyOttBp9o0ah+zWyZYv7tfP734dm/5HwxBPufbvnnuDsr1cv1Xbtym8mNCWrMon+889dDQNUGzdWvf/++BsFcvSo6v/+r6t9duhw/FDHisjJUb3lFve+/frXlatxTprkYov06KWyzJjhzpPwTfLVqrn72rXdUNtly4J7zN//3iX6UJyXESn5+a6ZBdzw28r49lu3n7/8JTixVUVVItGvW+deTZMmqpMnu5NS4tk777ix67Vrq/7znxWvBe3erdqvn3vv7rij8p2Ea9a4fT36aOX2E2qdO7vEW9A2P2OG66y/6aZjHbUXXOBOAqtsJ/i+fe6X5bBhQQk9qhw+7PqO6tZ1719FTZzovmyzsoIXW1VTJRK9qvunjIf2T3/99JPqZZe5v+KgQYH3P2zY4Eb11Kih+swzwYurSxc35j5abdrkfnX88Y8lP79zpzthrU0b9962aOFO4qnoCK2C8eeB9nnEiqwsNwz4jDMq1kSam+tOFIyFEVvRrMok+qooP191yhR3Nm1iohuH7I8lS1SbNnVNXIsXBzemBx/UqD6zsaD2WF4zSm6u6uuvH/vFU6uWO68gkDNac3PdF0aPHpWLOdp99JH7DPbvH/ivwrffVptCIwgs0VcBX37phrtVq+Y6n3NySl93+nT3T9munWvyCrZNm9wn68EHg7/vyjp61NU+BwwIbLs1a1w/Rr16WniW8MyZqkeOlL3da6+59V99teIxx4qnn3av9e67A9suNdVVOMo6Ic+UzxJ9FbF377EzWXv1Uv3uu6LP5+W5se6getFFoR2JdN55rgkn2vzrX+71z59fse1373bj4M84w+3n5JNdB3RpZ1b36uX6AMr64o0nY8e692X2bP/W37nT9TPdckto46oKLNFXMS+95GqejRsfO1V9/37VIUPcX3zs2NCfZfvoo+5Ya9aE9jiB6tfPNXFVNvHm5akuWODmIwL3C2nEiKLt8JmZ7rlHHqncsWLJ4cOumSohwb8RYU8+6d6jaJ2xNZZYoq+Cvv1WtVs39xe+8UbVrl1dB+Qjj4RnnPK2be54kyZVfB/Bno9mwwYtnNIhmNauVb3tNnfqPrgpAl58UXXoUFcW7yPAivv+ezcirHVrN0lfWc45R7VTJxs7HwyW6KuoI0fckMmCqR9efz28x+/d2431r8g/cSjmo7nrLjekMlRD+PbudUNd27c/FvP48aE5VrT79FPXed2vX+mdsytXVr1fPKFkib6K++ST0HS6lqegc2758sC3DfZ8NEeOuPlZBg2q2PaByMtzI0nGjSt9Wueq4Jln3N/szjtLfv6OO9zQ3p9+Cm9c8aqsRB9F1543odKjB7RtG/7jXnmlu0B6Rkbg2373XWDl5Zk3D376CW68sWLbB6JaNejXD/7xDzjppNAfL1pdfz3cfDM89NDxn4GcHHf5ycsvh+bNIxNfVWKJ3oRMs2ZwySXun1w1sG1btgysvDxTp7prkPbrV7HtTcU8+ihccIG77vKKFcfKFy50X7x28e/wsERvQio11V1s/YsvAttu8mRISChalpDgygO1bh0sXgxjx7pfGCZ8atWCOXOgSRO44grYscOVp6fDiSfCpZdGNr6qwhK9CakrrnD/7IE236SlwbRprhYu4u6nTXPlgZo2DWrUcLVKE34nnQRz58IPP8CwYe7+jTfg2muhZs1IR1c1iAb6mzrEUlJSNDMzM9JhmCAaPBg+/xy2bg1/jfrwYUhMhN69Xc3SRE56uvuybdfO/cpauRLOOivSUcUPEflSVVNKes5q9KbCZs6EVq1c52OrVm65JKmprhb30UfhjM6ZO9c1F4SjE9aU7brrYNw4l+TPOceSfDjViHQAJjbNnOnavA8edMtbtrhlOL555fLLXft6RgZceGF445w6Fdq0gb59w3tcU7JHHnG/6q64ItKRVC1WozcVMnHisSRf4OBBV15c3bowcKBrOsnJCU98AGvWwJIl7guomn3So0LNmm4kTu/ekY6karGPv6mQQMe5p6bCzz+70S/hMm2aSyw2hM9UdX4lehHpLyJrRWS9iEwo4fkpIrLcu60Tkd0+z40SkW+926hgBm8iJ9Bx7v37Q8OGFTt5qiIOHYLp02HIEDeMz5iqrNxELyLVgSeAS4GOwHAR6ei7jqqOV9UuqtoF+Acw19u2CXAfcC7QHbhPRBoH9yWYSAh0nHvt2i7pzp0LR46EPr5XX4Xdu60T1hjwr0bfHVivqhtV9SiQAQwqY/3hwCve418Bi1R1p6ruAhYB/SsTsIkOFRnnnpoKe/fCW2+FPr6pU90wPmsLNsa/RN8C2OqznOWVHUdEkoDWQEFLrF/bishYEckUkczs7Gx/4jZ+8ncIZEWkpbmzXvPz3X15JzNddJGbFiHUzTcrV8Inn7hOWJHQHsuYWBDszthUYI6q5gWykapOU9UUVU1pbjMcBU3BEMgtW9xcMwVDIIOZ7ANRowZcfTXMnw8HDoTuOFOnuqaiUdYjZAzgX6LfBpzms5zolZUklWPNNoFua4IskCGQ4ZKa6mJ4443Q7P/AAXjxRbjqKvfrwRjjX6JfCrQVkdYiUguXzOcXX0lEOgCNgU99it8G+olIY68Ttp9XZsIg2FP9BsMFF8Cpp4au+WbWLNcPYJ2wxhxTbqJX1VxgHC5BrwFmq+oqEblfRAb6rJoKZKjP5DmquhN4APdlsRS43yszYRDsqX6DoVo1N7HVggWwZ0/w9z91Kpx5pvtCMcY4frXRq+oCVW2nqqer6mSv7F5Vne+zziRVPW6Mvao+r6pneLf04IVuyhPMqX6DKTUVjh51FwMJpuXL3XTIN95onbDG+LIzYyMs1KNigjXVbzCdcw60bh385pupU6FOHRg5Mrj7NSbW2aRmERTIxGAVlZYW+cRenIir1T/0kJsWIRidpvv2wYwZrlmosZ2SZ0wRVqOPoGgcFRMuqamQlxe8OeJfeQX277dOWGNKYok+gqJxVEy4dOrkOk2D1Xwzdarb53nnBWd/xsQTS/QRFI2jYsKloPlmyRLYVskzKzIzYdky64Q1pjSW6CMoWkfFhMuwYe6M3Vdfrdx+pk5179s11wQnLmPijSX6CIrWUTHh0r49JCdXrvlmzx7XPj98uJsG2RhzPEv0ERboxGDxJjXVXTh806aKbT9zppv2wDphjSmdJXoTUUOHuvtZswLfVtU12yQnQ0pKcOMyJp5YojcR1aoV9OhRseabzz+Hr7+2TlhjymOJ3kRcaiqsWOEu5h2IqVOhXj0YMSI0cRkTLyzRm4i7+mpXIw+k+Wb3brd+WhrUrx+62IyJB5boTcSdcoq75F9Ghmt398dLL7kLgFsnrDHls0RvokJqKqxd65pwyqMKTz/tJkdLTg59bMbEOkv0JipceaW71KA/nbIffwyrV1tt3hh/WaI3UaFpU+jXz7/mm6lToUED9yvAGFM+S/QmaqSmuqmaP/+89HV27HBTJlxzDdStG77YjIllluhN1Bg0CGrXLrv55sUX4cgRa7YxJhCW6E3UaNAABgyA2bPdXPXFFZwJ26MHnH12+OMzJlb5lehFpL+IrBWR9SJy3HVhvXWGishqEVklIi/7lOeJyHLvNr+kbY0pkJoKP/wAH354/HMffOBG5lht3pjAlHspQRGpDjwBXAJkAUtFZL6qrvZZpy1wN3C+qu4SkRN9dnFIVbsEOW4TpwYMcG3vGRlubL2vqVOhUaNj8+MYY/zjT42+O7BeVTeq6lEgAxhUbJ0bgCdUdReAqv4U3DBNVZGQ4Nrq58yBnJxj5dnZ8K9/uQt/n3BC5OIzJhb5k+hbAFt9lrO8Ml/tgHYi8rGIfCYi/X2eqyMimV75FSUdQETGeutkZmdnB/QCTPxJTXWja95991jZ9Oku8VuzjTGBC1ZnbA2gLdAbGA48IyKNvOeSVDUFGAE8KiKnF99YVaepaoqqpjRv3jxIIZlY1a+fa6IpGH2Tn+8uyNKrF3TsGNnYjIlF/iT6bcBpPsuJXpmvLGC+quao6iZgHS7xo6rbvPuNwPtAzJ20PnOmm063WjV3P3NmpCOKb7Vrw5Ah8NprcPgwLF4M69dbbd6YivIn0S8F2opIaxGpBaQCxUfPzMPV5hGRZrimnI0i0lhEavuUnw+sJobMnAljx7oTeVTd/dixluxDLTUV9u6Ft95ynbBNm7ppEowxgSs30atqLjAOeBtYA8xW1VUicr+IDPRWexvYISKrgfeAP6jqDuBMIFNEVnjlD/qO1okFEyfCwYNFyw4edOUmdPr0gebN4fHHYd48GDUK6tSJdFTGxCZRf+eFDZOUlBTNzMyMdBiFqlUree4VEdd2bELnllvgySfd42++cRcTN8aUTES+9PpDj2NnxpajZcvAyk3wFExa1qePJXljKsMSfTkmT3Zju30lJLhyE1rnn+/6Qx54INKRGBPbyj0ztqpLS3P3EyfCd9+5mvzkycfKTehUq+Y6Yo0xlWOJ3g9paZbYjTGxy5pujDEmzlmiN8aYOGeJ3hhj4pwlemOMiXOW6I0xJs5ZojfGmDhnid4YY+KcJXpjjIlzluiNMSbOWaI3xpg4Z4neGGPinCV6Y4yJc5bojTEmzlmiN8aYOGeJ3hhj4pxfiV5E+ovIWhFZLyITSllnqIisFpFVIvKyT/koEfnWu40KVuDGGGP8U+6FR0SkOvAEcAmQBSwVkfmqutpnnbbA3cD5qrpLRE70ypsA9wEpgAJfetvuCv5LMcYYUxJ/avTdgfWqulFVjwIZwKBi69wAPFGQwFX1J6/8V8AiVd3pPbcI6B+c0I0xxvjDn0TfAtjqs5zllflqB7QTkY9F5DMR6R/AtojIWBHJFJHM7Oxs/6M3xhhTrmB1xtYA2gK9geHAMyLSyN+NVXWaqqaoakrz5s2DFJIxxhjwL9FvA07zWU70ynxlAfNVNUdVNwHrcInfn22NMcaEkD+JfinQVkRai0gtIBWYX2ydebjaPCLSDNeUsxF4G+gnIo1FpDHQzyszxhgTJuWOulHVXBEZh0vQ1YHnVXWViNwPZKrqfI4l9NVAHvAHVd0BICIP4L4sAO5X1Z2heCHGGGNKJqoa6RiKSElJ0czMzEiHYYwxMUVEvlTVlJKeszNjjTEmzlmiN8aYOGeJ3hhj4pwlemOMiXOW6I0xJs5ZojfGmDhnid4YY+KcJXpjjIlzluiNMSbOWaI3xpg4Z4neGGPinCV6Y4yJc5bojTEmzlmiN8aYOGeJ3hhj4pwlemOMiXOW6I0xJs5ZojfGmDhnid4YY+KcX4leRPqLyFoRWS8iE0p4frSIZIvIcu92vc9zeT7l84MZvDHGmPLVKG8FEakOPAFcAmQBS0VkvqquLrbqLFUdV8IuDqlql8qHaowxpiL8qdF3B9ar6kZVPQpkAINCG5Yxxphg8SfRtwC2+ixneWXFXSkiX4vIHBE5zae8johkishnInJFSQcQkbHeOpnZ2dn+R2+MMaZcweqMfR1opapnA4uAF3yeS1LVFGAE8KiInF58Y1WdpqopqprSvHnzIIVkjDEG/Ev02wDfGnqiV1ZIVXeo6hFv8Vmgm89z27z7jcD7QHIl4jXGGBMgfxL9UqCtiLQWkVpAKlBk9IyInOKzOBBY45U3FpHa3uNmwPlA8U5cY4wxIVTuqBtVzRWRccDbQHXgeVVdJSL3A5mqOh+4TUQGArnATmC0t/mZwFQRycd9qTxYwmgdY4wxISSqGukYikhJSdHMzMxIh2GMMTFFRL70+kOPY2fGGmNMnLNEb4wxcc4SvTHGxDlL9MYYE+cs0RtjTJyzRG+MMXHOEr0xxsQ5S/TGGBPnLNEbY0ycs0RvjDFxLm4S/cyZ0KoVVKvm7mfOjHRExhgTHcqd1CwWzJwJY8fCwYNuecsWtwyQlha5uIwxJhrERY1+4sRjSb7AwYOu3Bhjqrq4SPTffRdYuTHGVCVxkehbtgys3BhjqpK4SPSTJ0NCQtGyhARXbowxVV1cJPq0NJg2DZKSQMTdT5tmHbHGGANxMuoGXFK3xG6MMceLixq9McaY0vmV6EWkv4isFZH1IjKhhOdHi0i2iCz3btf7PDdKRL71bqOCGbwxxpjyldt0IyLVgSeAS4AsYKmIzFfV1cVWnaWq44pt2wS4D0gBFPjS23ZXUKI3xhhTLn9q9N2B9aq6UVWPAhnAID/3/ytgkaru9JL7IqB/xUI1xhhTEf4k+hbAVp/lLK+suCtF5GsRmSMipwWyrYiMFZFMEcnMzs72M3RjjDH+CNaom9eBV1T1iIjcCLwAXOTvxqo6DZgG4LX1b6lELM2AnyuxfaTEatxgsUeKxR4Z0Rp7UmlP+JPotwGn+SwnemWFVHWHz+KzwEM+2/Yutu37ZR1MVZv7EVOpRCRTVVMqs49IiNW4wWKPFIs9MmIxdn+abpYCbUWktYjUAlKB+b4riMgpPosDgTXe47eBfiLSWEQaA/28MmOMMWFSbo1eVXNFZBwuQVcHnlfVVSJyP5CpqvOB20RkIJAL7ARGe9vuFJEHcF8WAPer6s4QvA5jjDGl8KuNXlUXAAuKld3r8/hu4O5Stn0eeL4SMQZqWhiPFUyxGjdY7JFisUdGzMUuqhrpGIwxxoSQTYFgjDFxzhK9McbEubhJ9OXNxxOtROQ0EXlPRFaLyCoR+V2kYwqUiFQXka9E5I1IxxIIEWnkneD3jYisEZEekY7JHyIy3vusrBSRV0SkTqRjKouIPC8iP4nISp+yJiKyyJsDa5E3Ki+qlBL337zPy9ci8pqINIpkjP6Ki0TvMx/PpUBHYLiIdIxsVH7LBe5Q1Y7AecAtMRR7gd9xbEhtLHkMeEtVOwCdiYHXICItgNuAFFX9BW4kXGpkoyrXdI6f+mQC8K6qtgXe9ZajzXSOj3sR8AtVPRtYRymDUKJNXCR6KjcfT0Sp6g+qusx7vA+XbEqaYiIqiUgiMAB3olzMEJGGwC+B5wBU9aiq7o5sVH6rAZwgIjWABOD7CMdTJlVdght27WsQ7gx6vPsrwhqUH0qKW1X/o6q53uJnuJNAo168JHp/5+OJaiLSCkgGPo9sJAF5FLgTyI90IAFqDWQD6V6z07MiUjfSQZVHVbcBDwPfAT8Ae1T1P5GNqkJOUtUfvMc/AidFMpgKGgMsjHQQ/oiXRB/zRKQe8C/gdlXdG+l4/CEilwM/qeqXkY6lAmoAXYGnVDUZOEB0Nh8U4bVlD8J9UZ0K1BWRayIbVeWoG+MdU+O8RWQirtl1ZqRj8Ue8JPpy5+OJZiJSE5fkZ6rq3EjHE4DzgYEishnXXHaRiMyIbEh+ywKyVLXg19McXOKPdhcDm1Q1W1VzgLlAzwjHVBHbC6ZO8e5/inA8fhOR0cDlQJrGyIlI8ZLoy52PJ1qJiODaideo6iORjicQqnq3qiaqaivce75YVWOidqmqPwJbRaS9V9QXKH4xnWj0HXCeiCR4n52+xEAncgnmAwVXnBsF/DuCsfhNRPrjmioHqurBSMfjr7hI9F7nSMF8PGuA2aq6KrJR+e184FpcbbjgUoyXRTqoKuJWYKaIfA10Af4S4XjK5f0CmQMsA/6L+x+O6lPyReQV4FOgvYhkichvgAeBS0TkW9yvlAcjGWNJSon7n0B9YJH3v/p0RIP0k02BYIwxcS4uavTGGGNKZ4neGGPinCV6Y4yJc5bojTEmzlmiN8aYOGeJ3hhj4pwlemOMiXP/DzpNH57FdZuNAAAAAElFTkSuQmCC\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAXoAAAEICAYAAABRSj9aAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deXhU9fX48fcJewAFJCqyBLAsgkWQAEGqglYBQXBBAqYKP6ooaFXUIi4Vi9LWav1aW9RGXGgFWVxxXxHcIWAAQVBk0SAihl1AEnJ+f3zuwBCyTJKZuTOT83qeeSZz5y5nMsmZO+d+FlFVjDHGJK4kvwMwxhgTWZbojTEmwVmiN8aYBGeJ3hhjEpwlemOMSXCW6I0xJsFZojflIiKvi8iIcK/rJxFZLyK/jcB+VUR+5f38qIj8KZR1K3CcTBF5q6JxlrLf3iKSG+79muir7ncAJvJEZHfQw2TgF+CA9/gqVZ0e6r5UtX8k1k10qnp1OPYjIi2BdUANVS3w9j0dCPk9NFWPJfoqQFXrBX4WkfXAFar6TtH1RKR6IHkYYxKHlW6qsMBXcxG5RUR+AJ4UkYYi8oqIbBGRbd7PzYK2eV9ErvB+HikiH4rI/d6660SkfwXXbSUiC0Rkl4i8IyJTROTpEuIOJca7ReQjb39viUjjoOcvE5ENIpInIreX8vvpISI/iEi1oGUXisgy7+fuIvKJiGwXkU0i8m8RqVnCvp4SkXuCHv/R2+Z7ERlVZN0BIvK5iOwUke9E5K6gpxd499tFZLeI9Az8boO2P01EFonIDu/+tFB/N6URkZO87beLyAoRGRT03HkistLb50YRudlb3th7f7aLyFYR+UBELO9Emf3CzfFAIyAVGI37m3jSe9wC2Av8u5TtewCrgcbA34HHRUQqsO4MYCFwDHAXcFkpxwwlxkuB/wccC9QEAomnA/CIt/8TvOM1oxiq+hnwM3BWkf3O8H4+AIzzXk9P4GxgbClx48XQz4vnHKANUPT6wM/A5UADYAAwRkQu8J47w7tvoKr1VPWTIvtuBLwKPOS9tgeAV0XkmCKv4YjfTRkx1wBeBt7ytvsDMF1E2nmrPI4rA9YHTgbe85bfBOQCKcBxwG2AjbsSZZboTSEwUVV/UdW9qpqnqs+p6h5V3QVMBs4sZfsNqvqYqh4ApgFNcP/QIa8rIi2AbsCdqrpfVT8E5pZ0wBBjfFJVv1LVvcBsoLO3fAjwiqouUNVfgD95v4OSPAMMBxCR+sB53jJUdbGqfqqqBaq6HvhPMXEUZ6gX3xeq+jPugy349b2vqstVtVBVl3nHC2W/4D4YvlbV/3lxPQOsAs4PWqek301p0oF6wN+89+g94BW83w2QD3QQkaNUdZuqLgla3gRIVdV8Vf1AbYCtqLNEb7ao6r7AAxFJFpH/eKWNnbhSQYPg8kURPwR+UNU93o/1yrnuCcDWoGUA35UUcIgx/hD0856gmE4I3reXaPNKOhbu7P0iEakFXAQsUdUNXhxtvbLED14cf8Gd3ZflsBiADUVeXw8RmeeVpnYAV4e438C+NxRZtgFoGvS4pN9NmTGravCHYvB+L8Z9CG4Qkfki0tNbfh+wBnhLRNaKyITQXoYJJ0v0pujZ1U1AO6CHqh7FoVJBSeWYcNgENBKR5KBlzUtZvzIxbgret3fMY0paWVVX4hJafw4v24ArAa0C2nhx3FaRGHDlp2AzcN9omqvq0cCjQfst62z4e1xJK1gLYGMIcZW13+ZF6usH96uqi1R1MK6s8yLumwKquktVb1LV1sAg4EYRObuSsZhyskRviqqPq3lv9+q9EyN9QO8MORu4S0RqemeD55eySWVifBYYKCK/8S6cTqLs/4MZwPW4D5Q5ReLYCewWkfbAmBBjmA2MFJEO3gdN0fjr477h7BOR7rgPmIAtuFJT6xL2/RrQVkQuFZHqIpIBdMCVWSrjM9zZ/3gRqSEivXHv0UzvPcsUkaNVNR/3OykEEJGBIvIr71rMDtx1jdJKZSYCLNGboh4E6gA/AZ8Cb0TpuJm4C5p5wD3ALFx7/+JUOEZVXQFcg0vem4BtuIuFpQnUyN9T1Z+Clt+MS8K7gMe8mEOJ4XXvNbyHK2u8V2SVscAkEdkF3Il3duxtuwd3TeIjryVLepF95wEDcd968oDxwMAicZebqu7HJfb+uN/7w8DlqrrKW+UyYL1Xwroa936Cu9j8DrAb+AR4WFXnVSYWU35i10VMLBKRWcAqVY34NwpjEp2d0ZuYICLdROREEUnymh8OxtV6jTGVZD1jTaw4Hnged2E0Fxijqp/7G5IxicFKN8YYk+CsdGOMMQku5ko3jRs31pYtW/odhjHGxJXFixf/pKopxT0Xc4m+ZcuWZGdn+x2GMcbEFREp2iP6oJBKNyLST0RWi8ia4rowi8j/iUiOd/tKRLYHPTdCRL72bjE/CYUxxiSaMs/ovfFDpuBG2ssFFonIXK9rOACqOi5o/T8AXbyfA70W03Bdtxd7224L66swxhhTolDO6LsDa1R1rdc7biaujXNJhuON7gf0Bd5W1a1ecn8b6FeZgI0xxpRPKIm+KYePtJfL4SPhHSQiqUArDnXpDnlbY4wxkRHu5pXDgGe98cZDJiKjRSRbRLK3bNkS5pCMMaZqCyXRb+TwIVWbUfKQp8M4VLYJeVtVzVLVNFVNS0kptnVQmaZPh5YtISnJ3U+3qZKNMQYILdEvAtqIm9OzJi6ZHzH7jzdMa0PcCHUBbwLnipvjsyFwrrcsrKZPh9GjYcMGUHX3o0dbsjfGGAgh0atqAXAtLkF/CcxW1RUiMil4cmDcB8DM4GnCVHUrcDfuw2IRMMlbFla33w579hy+bM8et9wYY6q6mBvrJi0tTcvbYSopyZ3JFyUChTbFgTGmChCRxaqaVtxzCTHWTYuiE7GVsdwYY6qShEj0kydDcvLhy5KT3XJjjKnqEiLRZ2ZCVhakprpyTWqqe5yZWfa2xhiT6GJuULOKysy0xG6MMcVJiDN6Y4wxJbNEb4wxCc4SvTHGJDhL9MYYk+As0ZuYlZ8PZ58NL77odyTGxLeEaXVjEs9777nb1q0weLBrOmuMKT87ozcxa84cd5+TA59+6m8sxsQzS/QmJuXnwwsvuDP5+vXh4Yf9jsiY+GWJ3sSkefNcyeb//T8YMQJmzwabk8aYirFEb2LSnDnuTL5vXxgzBvbvhyee8DsqY+KTJXoTcwJlm/PPh9q1oUMH6N0bHn0UDpRrkkpjDFiiNzHo/fchLw8uueTQsrFjYf16eOMNv6IyJn5ZojcxZ84cqFfPlW0CLrgAjj/eLsoaUxGW6E1MKSg4VLapU+fQ8ho13DzAr78Oa9f6F5+pHFX3Pr76qt+RVC0hJXoR6Sciq0VkjYhMKGGdoSKyUkRWiMiMoOUHRCTHux0xqbgxwd5/H3766fCyTcCVV7ppI//zn6iHZcJk6VJ47DGYOtXvSKqWMhO9iFQDpgD9gQ7AcBHpUGSdNsCtQC9V7QjcEPT0XlXt7N2CJxM35gizZ7uyTb9+Rz7XrJlrV//447BvX/RjM5UX6AT36afFz/NsIiOUM/ruwBpVXauq+4GZwOAi61wJTFHVbQCq+mN4wzRVQaBsM3Dg4WWbYGPHugu1gYRh4oeqe9+qVYMffoBvv/U7oqojlETfFPgu6HGutyxYW6CtiHwkIp+KSPD5WG0RyfaWX1DcAURktLdO9hbrFVNllVa2CTjrLGjXzi7KxqOlS+Hrr+GKK9zjTz7xN56qJFwXY6sDbYDewHDgMRFp4D2XqqppwKXAgyJyYtGNVTVLVdNUNS0lJSVMIZl4M2cO1K0L/fuXvI6I60D16aewZEn0YjOVFzibnzjRfWOz8YuiJ5REvxFoHvS4mbcsWC4wV1XzVXUd8BUu8aOqG737tcD7QJdKxmwSUEEBPP986WWbgBEj3DqPPBKd2EzlBco2ffpAkyaQlmaJPppCSfSLgDYi0kpEagLDgKKtZ17Enc0jIo1xpZy1ItJQRGoFLe8FrAxT7CaBzJ9fdtkmoEEDNxH89OmwfXvkYzOVFyjbBN7fnj3h88/hl1/8jauqKDPRq2oBcC3wJvAlMFtVV4jIJBEJtKJ5E8gTkZXAPOCPqpoHnARki8hSb/nfVNUSvTnCnDmQnFx62SbYmDGwdy9MmxbZuEx4BMo2F17oHqenu/GLPv/c37iqCtEYa+OUlpam2dnZfodhoqigAE44wX2tnzUr9O169nQjXK5aZZOSxDJVaNsWWraEt992y77/Hpo2hQcegHHjfA0vYYjIYu966BGsZ2wIpk93f6RJSe5++nS/I0osCxa4IYhDKdsEGzsWvvrKzUJlYtfSpbBmDQwdemjZCSdAixZWp48WS/RlmD7dddnesMGdmWzY4B5bsg+fQNnmvPPKt90ll8Axx1hTy1hXtGwTkJ5uiT5aLNGX4fbbYc+ew5ft2eOWm8o7cMC1thkwwCX78qhdG37/e3jpJcjNjUx8pnJUXW/nPn2gcePDn+vZ03Wa+v57f2KrSizRl6Gk3nvWqy88FiyAH38sf9km4KqroLDQjZ9iYk9xZZuA9HR3/9ln0Y2pKrJEX4YWLcq33JTPnDmuTXx5yzYBrVu7ljpZWW7CEhNbZs8uvmwD0KUL1KxpPWSjwRJ9GSZPPrKkkJzslpvKCS7b1K1b8f2MHevGTnnxxfDFZiov0EnqrLOOLNsA1Krlkr3V6SPPEn0ZMjPd2WJqqmvCl5rqHmdm+h1Z/PvgA9i8ueJlm4B+/VxrKLsoG1sCZZvS3t/0dMjOtm9jkWaJPgSZmW4au8JCd29JPjwCZZsBAyq3n2rV4Oqr3aBoK607XsworWwT0LOn6/i2fHn04qqKLNEbXxw4AM89V/myTcCoUa7ea+PfxIayyjYBgQuyVqePLEv0xhcffhiesk1ASopr2TFtGuzeHZ59mooLpWwDrlHD8cdbnT7SLNEbX4SrbBNs7FjYtcs6s8WCUMo24K57WcepyLNEb6IuULY577zwlG0C0tOhc2d3UTbGhnCqUkIt2wSkp7uz/59+inxsVZUlehN1H33kmkOGq2wTIOLO6pcts5qvn3JyQivbBPTs6e6t41TkWKI3UTdnjhu+IJxlm4BLL4WjjrKmln4qaWybknTt6ta3D+fIsURvourAAXj2WVe2qVcv/PuvWxdGjnTJ5keboj7qylu2AfeedepkdfpIskRvoipSZZtgY8a4SS2eeCJyxzDFK2/ZJiA9HRYudCcCJvwSKtGvWmUX4WJdoGwzcGDkjtG+vRst8dFHLXFEW3nLNgHp6a7F1JdfRiauqi6kRC8i/URktYisEZEJJawzVERWisgKEZkRtHyEiHzt3UaEK/Ci1qxxLS4uusidMZrYU1joWtv07x+Zsk2wsWPd3AGvvx7Z45hDAkMSl6dsExC4IGvlm8goM9GLSDVgCtAf6AAMF5EORdZpA9wK9FLVjsAN3vJGwESgB9AdmCgiDcP6CjytWsE997h/7I4d4Zln7Ow+1nz0EWzaFNmyTcDgwdCkiV2UjaacHPjmm+KHJC7Lr34FjRrZBdlICeWMvjuwRlXXqup+YCYwuMg6VwJTVHUbgKoGLoP1Bd5W1a3ec28D/cIT+uGqVYObb3Z/bG3auNYXQ4a43pexrCpNUzhnjhuxMJJlm4AaNdxMYG+84ZKPibxA2eaCC8q/rXWciqxQEn1T4Lugx7nesmBtgbYi8pGIfCoi/cqxLSIyWkSyRSR7y5YtoUdfjPbt3ZnjvffCq6+6s/vZsyu1y4ipStMUBpdt6tePzjGvvNJ9gP7nP9E5XlVWmbJNQHq6G5Ru+/bwxmbCdzG2OtAG6A0MBx4TkQahbqyqWaqapqppKSkplQ6mWjUYPx6WLIETT4SMDFcuqORnSNhVpWkKP/7YTRkXjbJNQNOm7uzy8cfdCIkmcipTtgkIDHC2aFF4YjKHhJLoNwLNgx4385YFywXmqmq+qq4DvsIl/lC2jZgOHdzZ/V//CnPnusdz5kTr6GWrStMUBso2558f3eOOHQtbt8bW+56IKlO2Ceje3ZVwrHwTfqEk+kVAGxFpJSI1gWHA3CLrvIg7m0dEGuNKOWuBN4FzRaShdxH2XG9Z1FSvDhMmuLP71FR3xpGRERvjalSVaQoLC10nqX79ole2CejTB9q1s4uykRQo25x9dsXLNgBHH+1OxuyCbPiVmehVtQC4FpegvwRmq+oKEZkkIoO81d4E8kRkJTAP+KOq5qnqVuBu3IfFImCStyzqOnZ0ZwqTJ8MLL7jHzz/vRySHVJVpCj/5JPplm4DA+DeffQaLF0f/+FVBoGwTjvc3cEHWWsyFmarG1K1r164aacuWqZ56qiqoDhum+tNPET9kiZ5+WjU1VVXE3T/9dHztPxTXX69aq5bqjh3RP7aq6rZtqsnJqr//vT/HT3QTJqhWq6a6ZUvl9/XYY+7/cvXqyu+rqgGytYS86ntiL3qLRqJXVd2/X/Xuu1Vr1FA97jjVF16IymGj6umnXYJz50fulpwc3WR/4IBq06aqgwZF75jFufJK1Tp1VLdu9TeORFNYqHriiarnnhue/S1f7v5Op00Lz/6qktISfUINgVAeNWrAHXe4iYmbNHFdtjMzIS/P78jCJxZa9Xz6KWzc6E/ZJtjYsa7lzbRp/saRaMJZtgFXoz/qKLsgG25VNtEHdOrkBlP685/dBaWTT3YtdBJBLLTqmTPHzeUa7dY2RXXu7LrZP/ywuzhswiPUmaRClZTkWt/YBdnwqvKJHtzZ/Z13uva7xx7rus9ffjls2+Z3ZJXjd6ueQGubvn1diwq/jR0LX38N773ndySJITAk8dlnwzHHhG+/6elu8piffw7fPqs6S/RBOnd2yf7OO2HGDNcy55VX/I6q4vxu1fPZZ5Cb63/ZJmDIENf8z5pahsfnn4e3bBOQnu5OErKzw7vfqswSfRE1a7oyzsKFLimcf76byCIeu2UPH+7G/Eny3uVmzSAry12LiIZA2WbQoLLXjYbateH3v4eXXnIfQJGW6E0EKzokcVl69HD3VqcPH0v0JTj1VHdGcccd8PTT7uz+pZfi5583O9vVpKdOda+lTh3XMzXwTxRpsVa2CbjqKvceZmWFf995ee5v5KaboFs3N3PSxx+H/zixIFJlG3AnWG3aWKIPJ0v0pahZE+6+2/3BNWzouneffjosWOB3ZCXLy4Orr3YXtDZscK1MFi50dekdO1zyj8YkzAsXwnffxU7ZJqBVKzeN4WOPuVmoKmPTJncx8tpr3UX9xo3d38iUKa5ElpSUuLNcBco2lRnbpjTp6e6CbLycWMW8ktpd+nWLVjv68tq/X/XRR1VPOMG18+3XT3XxYr+jOqSgwMXXqJHrvHLDDarbtx++zldfuTbPdeqovvRS5Y9ZWmesG29UrVnzyBhiwauvuvdw1qzybbdunWvf/fvfq7Zpc6hvQt26rh35PfeoLligunevW/93v1Nt2FD1l1/C/hJ8F+gkFanOhlOmuN/tunWR2X8iwjpMhc+ePar33ecSKqheconql1/6G9Onn6qmpbl4zjjD9fwtyebNqt26qSYlqT78cMWPWVpnrMJC1RYtVAcOrPj+I6mgQLVVK9Uzzyx5ncJC1VWrVLOyXMJu0eLQ62zQQPX8893fwcKF7iSgOHPnuvVfey0iL8M34e4kVZzFi93v7plnIneMRGOJPgK2b1f905/c2VxSkuqoUaobNkQ3hh9/dGeXoNqkier06e6fsCy7d7skDO7M7MCB8h87NfXwJB+4paa6D55Y7914770uxi++cI8PHFBdulT1oYdUhwxRPfbYQ6/puOPcB/q//uXWCfX3tW+f6tFHq44YEbGX4YtAEp46NXLH2L/fffO8/vrIHSPRWKKPoM2bVceNc2O51Kzp/jA3b47sMQsK3Ffbhg1Vq1dXvemm8o8jk5+vOnq0+wvIzCx/eUGk+EQv4so2NWq4MWZi1ZYt7j0780x3dt6gwaHX0Ly5O4vPynJjroTy4VmSkSNdst+3L2yh+y7SZZuAM85Q7d49ssdIJJboo+Dbb1WvuML9A9Stq3rHHZGpT3/8sWqXLu6d69NHdcWKiu+rsFB18mS3r7POKl+8JZ3Rt2jhbgMGVDyuaAl8G2rb1r13//2v6vr14T3G66+7Y4TjmkgsKCxUbd1atW/fyB9r/Hh3whC45mFKZ4k+ilatUh061P1mGzZ0JYKff678fjdvdmeH4AYJmzmzcmeawaZNc98Mfv1r1e++C22bkmr0d93lfn7qqfDEFkn794dnxMWyjtGokfvWlAiiUbYJeP55d6yPP478sRJBaYnemleGWbt2MGuWm+ikZ0+45RY3w/0jj1SsOV9BAfzrX9C2rZtLdvx4WLXKTZ4iEp6YL78cXnsN1q93MX/xRdnbZGa6tuipqS6O1FT3eNcuN6TE4KLTx8egGjUqN1FGqMe46CLXvj4RpjMMx0xSobKOU2FU0ieAX7d4P6MvasEC1d/8xp2ZtG6t+r//uRp7qNt26uS2PeecyLfuyclxzUePPlr1vffKv31hoSvpnHde2EOLa2+/7d7D557zO5LKiWbZJqBFC/cN2ZQNO6P3T6CD1WuvueFXL7vMjalTWi/bTZvcemec4QZWe/ZZePNNaN8+srGecorrpNK0qevROmNG+bZftMh10oq1TlJ+690bUlLcN7149vnnsHZtdN/fnj1tJMtwCCnRi0g/EVktImtEZEIxz48UkS0ikuPdrgh67kDQ8gQZALh8RKB/fzeV3axZroRzwQXujzh4JMX8fPi//3Pln9mz4bbb4Msv4eKLw1emKUuLFvDhhy62zEz4+99D7504Z054yzbTp0PLlq6HacuW7nE8ql7dvYevvBLfIzIGhiSORtkmID3d9bDeuDF6x0xIJZ3qB25ANeAboDVQE1gKdCiyzkjg3yVsv7usYwTfEq10U5z8fHcxq3lz95X+7LPdBdGTT9aDvW6/+srfGPftU83IcPFcc03Z5aZA2aZ///AcPxZmxwqnefO0Qr1xY4UfZRtV1U8+SYyyVzRQydJNd2CNqq5V1f3ATCAOLrXFrurV3SiKX33lzuCXLoURI9yFzBdecGWeNm38jbFWLVe6uflmN3bLxRcfOVtVsOzs8JZtYmF2rHA6/XQ4/vj4Ld/4UbYB6NLFjTllF2QrJ5RE3xT4LuhxrresqItFZJmIPCsizYOW1xaRbBH5VESK/dInIqO9dbK3bNkSevRxrnZtuOEG9w/0+uuwcqX7WhytMk1ZkpLgvvvgn/90s26dfTb89FPx686Z4z7AwlW2iYXZscKpWjWXJF97zX2gx5vZs937G82yDbgTji5dLNFXVrguxr4MtFTVTsDbQPDMnKmqmgZcCjwoIicW3VhVs1Q1TVXTUlJSwhRS/KhfH/r1O3KSkFhx3XXugnBODpx2mvtgCqbekLW//S00ahSeY/o9O1YkDB0K+/bByy/7HUn5BN7fSAxJHIqePd03xvz86B87UYSS6DcCwWfozbxlB6lqnqr+4j2cCnQNem6jd78WeB/oUol4jU8uugjeeccNg9yzp2thE7B4sWuDH86v9X7PjhUJp53mWjTFW/nGr7JNQHq664OwbJk/x08EoST6RUAbEWklIjWBYcBhrWdEpEnQw0HAl97yhiJSy/u5MdALWBmOwE309erlJtJITnZNBl991S0PlG3C+bW+pA5Z0ZodKxKSktxZ/RtvxNeMZX6VbQLS0929lW8qrsxEr6oFwLXAm7gEPltVV4jIJBEJTBJ3nYisEJGlwHW4VjgAJwHZ3vJ5wN9U1RJ9HGvXzrVrPukkN0VgVtahr/XhKtsEZGa6bwqFhe4+npN8wNChrnntSy/5HUlo/C7bgCvXHX+8JfrKEA21kXSUpKWlabbNChzzdu92X+XfeMM9njrVtSQypVN1s1x17HjoG1EsW7wY0tLg8cdh1Cj/4rjwQjc0x9df+xdDrBORxd710CNYz1hTIfXquZY4V14Jxx3n39f6eCPizurfegu2bvU7mrJFoixXET17wpo1UIUa5YWVJXpTYTVquNLNxo3+fa2PRxkZbrC6F17wO5LSBZdtwl2WK69AnT4a8x0nIkv0ptKqVfM7gvhy6qnQunXst75ZssS1tonUBODl0bWr+zuzOn3FWKI3JspE3Fn9e+/FdikiVso2AHXrQqdOlugryhK9MT7IyIADB+D55/2OpHiqrlllODvBVVZ6Oixc6H5vpnws0Rvjg06dDk1SE4uWLIF162JryOmePd3wESutgXa5WaI3VZLfQyAHWt/Mnw8//BDdY4cilso2AdZxquIs0ZsqZ/p0GD3ajbap6u5Hj45+ss/IcJ3BnnsuusctSyyWbcBNydmokSX6irBEb6qcWBkCuWNHd4u18s3ixbFXtgH3LSg93RJ9RViiN1VOLA2BnJHhZvSKpRmUJk1yI6peeKHfkRwpPd3V6ONprKBYYIneVDmxNATy0KGHOibFgrfecsMo/+lP0LCh39EcqWdPd79wob9xxBtL9KbKiaUhkNu1c5Oyz54d/WMXVVAA48bBiSe6OQhiUbduroRj5ZvysURvqpxYGwI5I8ONCOr37FmPPurKIv/4h5vZKRYdfTR06GCJvrws0ZsqKZaGQA4MMeDnWX1eHtx5p2tpM2hQ2ev7KXBBNtoD765Z4y6ejx3r/4dyeVmiN8ZnJ57oxnLxs/XNXXfBjh1usvpYmbO4JOnpsG1bdIcs/uUXGDbMNcWdOtU19bzqKneSEA8s0RsTAzIy3LyoRefjjYYVK+CRR+Dqq+Hkk6N//PIKXJD95JPoHXPCBNfs9Omn4ZtvXL+Lp56CNm3cPAzffBO9WCrCEr0xMcCv8o2quwBbvz78+c/RPXZFnXQSHHVU9Or0L78MDz4I117rego3bw7//rf7UB47FmbMcBfVR46M3YlRQkr0ItJPRFaLyBoRmVDM8yNFZIuI5Hi3K4KeGyEiX3u3EeEM3phEkZoKPXpEv3zzyivw9tsuyTduHN1jV1RSEnTvHp1En5vrEnjnznDffYc/17Qp/POfLuFff737kG7fHi67DFatinxs5VFmoheRasAUoD/QARguIvNE3zMAABsySURBVB2KWXWWqnb2blO9bRsBE4EeQHdgoojEYOtcY/yXkQE5OfDVV9E53v79cNNN7gx5zJjoHDNc0tNh2TL4+efIHaOgAC691NXnZ82C2rWLX69JE9dSad069/t8/nnXMmj4cFcWiwWhnNF3B9ao6lpV3Q/MBAaHuP++wNuqulVVtwFvA/0qFqoxiS0w5EC0yjf/+pcrNTzwgJstLJ707OlaTC1aFLlj3H03fPCBa3batm3Z6x93HPz97+4C7S23uG9Lv/61K8stWxa5OEMRSqJvCnwX9DjXW1bUxSKyTESeFZHm5dlWREaLSLaIZG+J5ZkYjAlRRUbHbNYMfvOb6JRvfvzRDXUwYAD0i8NTrx493H2kyjfz5rlEP2IE/O535ds2JQX++leX8G+/Hd5803WKu+gi943ND+G6GPsy0FJVO+HO2qeVZ2NVzVLVNFVNS0lJCVNIxvijMqNjDh0KX3wR+THX77jDDeT2j39E9jiRcswxrsVLJBL9jz+6fhVt27qLrhV1zDHuw2L9epg40c0o1qULDB7sWlhFUyiJfiPQPOhxM2/ZQaqap6q/eA+nAl1D3daYRFOZ0TGHDHHt2CN5Vp+T49qC/+EPrrVIvIpEx6nCQnfxdetW9x7Uq1f5fTZs6PopbNhwqBzUrZv7NhWtyc5DSfSLgDYi0kpEagLDgLnBK4hIk6CHg4AvvZ/fBM4VkYbeRdhzvWXGJKzKjI7ZpAmceaar00ei56cq3HCDO9u8887w7z+a0tNh82aXQMPlgQfg9dfd/SmnhG+/4IZvuOMOd4b/l7+4JJ+e7kpnH38c3mMVVWaiV9UC4Fpcgv4SmK2qK0RkkogEOktfJyIrRGQpcB0w0tt2K3A37sNiETDJW2ZMwqrs6JgZGa553vLl4Ysp4Lnn3KxW99wDDRqEf//RFO6OU599BrfeChdfHNlWSEcd5Y6zbh3ce6+btrFXLzjnHHe2HxGqGlO3rl27qjHx7OmnVZOTVd35s7slJ7vlodi8WTUpSfW228Ib1969qi1bqnbqpFpQEN59+yE/X7VOHdXrrqv8vrZtc7+b1FTVrVsrv7/y2L1b9f77VY87TvXUU1ULCyu2HyBbS8ir1jPWmDCr7OiYxx4LZ50V/vLNAw+4ssGDD0K1auHbr1+qV3e17spekFV1F8u/+w6eeSb64/DXreva369d664LRGKsIUv0xkRAZUfHzMhwoyV+/nl44vn+e1cXvugi6NMnPPuMBenp7ne0b1/F95GV5SZ+mTz5UDnID8nJbrC0SLBEb0wMuvBCd8YartY3t94K+flHduOPd+np7nVV9ANx+XJ3cfrcc+GPfwxvbLHEEr0xMeiYY9zY8OEo3yxcCP/9L9x4I7RuHZ74YkV6uruvyAXZn39235waNHC/n6QEzoYJ/NKMiW8ZGa7sU5lu/oHmlMcfD7fdFrbQYkaTJu4aSEXq9Ndd51o3Pf20G74gkVmiNyZGXXCBG4OmMuWbZ55xZ7t//asbijgRBTpOlceMGfDEE+7D7+yzIxNXLLFEb0yMatAA+vZ15ZvCwvJv//PPMH48pKXB5ZeHP75YkZ7uWsxsDLHP/Zo1bnaoXr1cj9WqwBK9MTEsI8ONiV6RGvTf/+6S34MPJnb9OVCnD2U4gV9+cb/TGjXcWX316pGNLVYk8NtvTPwbNAhq1Sr/0MXffusS/bBh7sw1kXXpAjVrhvZhOGGC64n61FOh91ROBJbojYlhRx0F553n2nkfOBD6duPHu443994budhiRa1acOqpZdfpA1MCXned+wCtSizRGxPjhg6FTZvgww9DW//DD90F3PHjq85Za3q6G/o3P7/457/7zo1K2aWL+6ZT1ViiNybGDRwIdeqE1vqmsNA1p2zWzCX6qiI93fWOLW4mp4IC1zN5/373O6xVK/rx+c0SvTExrl49l+yfe84lrdJMmwaLF7uz1uTk6MQXCwIXZIsr30yadGhKwDZtohtXrLBEb0wcyMhwMx/Nn1/yNIU7d7qhDk47zV2ErUpatHCdp4pekH3vPTck88iR5R9vKJFUkcZFxsS3/v3dKIeTJ7tmhIEZrALTFIIbt2XzZnfRMRIjIMYykSM7TgWmBGzXrnJTAiYCS/TGxIHkZNdSZNasIztP7dnj6vE//eQms+7WzZ8Y/ZaeDi+8AFu2uLGCRoyAbdvc5Nx16/odnb+sdGNMnMjIKLmH7Pffu7bkf/1rdGOKJcEdp/7xD3jjDdecslMnf+OKBSElehHpJyKrRWSNiEwoZb2LRURFJM173FJE9opIjnd7NFyBG1PV9O1beknmtttcnbqq6trVTagyZYr7XVx8sRvqwIRQuhGRasAU4BwgF1gkInNVdWWR9eoD1wNFOyJ/o6qdwxSvMVVW7dqul2vR9vQi0LgxjBvnT1yxom5dN6H3G2+4i9RTp1a9axUlCeWMvjuwRlXXqup+YCYwuJj17gbuBSox14sxpjQTvO/TKSkuiTVq5IYifuQR90FQ1f3mN278mmeeif/Jz8MplETfFPgu6HGut+wgETkVaK6qrxazfSsR+VxE5ovI6cUdQERGi0i2iGRv2bIl1NiNqXLOOcfNadqvH+TluWTfu7ebItDAn//s+hEE6vXGqXSrGxFJAh4ARhbz9CagharmiUhX4EUR6aiqO4NXUtUsIAsgLS0tjNMhG5NYatZ00wzOmeM6Um3b5i44WonCadDAzuSLE8oZ/UagedDjZt6ygPrAycD7IrIeSAfmikiaqv6iqnkAqroY+AZoG47Ajamqhg6FXbtcueaKK1xd2pjShJLoFwFtRKSViNQEhgFzA0+q6g5VbayqLVW1JfApMEhVs0UkxbuYi4i0BtoAa8P+KoypQs46y7UTP/po1+vTmLKUWbpR1QIRuRZ4E6gGPKGqK0RkEpCtqnNL2fwMYJKI5AOFwNWqujUcgRtTVdWo4cZTr1nTXZQ1piyilZ1iPszS0tI0Ozvb7zCMMSauiMhiVU0r7jnrGWuMOaikAdNMfLOxbowxgEvqo0cXP2BaVR75MRHYGb0xBoDbbz+U5AP27HHLTXyzRG+MAdyE4uVZbuKHJXpjDFDy/LJVZd7ZRGaJ3hgDuElNik4/mJzslpv4ZoneGAO4C65ZWZCa6oZUSE11j+1CbPyzVjfGmIMyMy2xJyI7ozfGmARnid4YYxKcJXpjjElwluiNMSbBWaI3xpgEZ4neGGMSnCV6Y4xJcJbojTEmwVmiN8ZEjY1374+QEr2I9BOR1SKyRkQmlLLexSKiIpIWtOxWb7vVItI3HEEbY+JPYLz7DRtA9dB495bsI6/MRO9N7j0F6A90AIaLSIdi1qsPXA98FrSsA24y8Y5AP+DhwGThxpiqxca7908oZ/TdgTWqulZV9wMzgcHFrHc3cC+wL2jZYGCmqv6iquuANd7+jDFVjI13759QEn1T4Lugx7nesoNE5FSguaq+Wt5tve1Hi0i2iGRv2bIlpMCNMfHFxrv3T6UvxopIEvAAcFNF96GqWaqapqppKSkplQ3JGBODbLx7/4SS6DcCzYMeN/OWBdQHTgbeF5H1QDow17sgW9a2xpgqwsa7908o49EvAtqISCtckh4GXBp4UlV3AI0Dj0XkfeBmVc0Wkb3ADBF5ADgBaAMsDF/4xph4YuPd+6PMRK+qBSJyLfAmUA14QlVXiMgkIFtV55ay7QoRmQ2sBAqAa1T1QJhiN8YYEwJRVb9jOExaWppmZ2f7HYYxxsQVEVmsqmnFPWc9Y40xJsFZojfGmARnid4YYxKcJXpjjElwluiNMSbBWaI3xpgEZ4neGGMSnCV6Y4xJcJbojTEmwVmiN8aYBGeJ3hhjEpwlemOMSXChDFPsu/z8fHJzc9m3b1/ZKxvf1a5dm2bNmlGjRg2/QzHGECeJPjc3l/r169OyZUtExO9wTClUlby8PHJzc2nVqpXf4RhjiJPSzb59+zjmmGMsyccBEeGYY46xb1/GxJC4SPSAJfk4Yu+V8cP06dCyJSQlufvp0/2OKHbETaI3xpiSTJ8Oo0fDhg2g6u5Hjw5vso/nD5KQEr2I9BOR1SKyRkQmFPP81SKyXERyRORDEengLW8pInu95Tki8mi4X0Bxwv2G5OXl0blzZzp37szxxx9P06ZNDz7ev39/qdtmZ2dz3XXXlXmM0047rXJBet5//30GDhwYln0ZEy9uvx327Dl82Z49bnk4ROODJKJUtdQbbp7Yb4DWQE1gKdChyDpHBf08CHjD+7kl8EVZxwi+de3aVYtauXLlEctK8vTTqsnJqu7tcLfkZLc8HCZOnKj33XffYcvy8/PDs/MwmDdvng4YMMDvMMr1nhlTWSKH/88HbiLh2X9qavH7T00Nz/7DATeHd7F5NZQz+u7AGlVdq6r7gZnA4CIfFjuDHtYFfJuINtKf7AEjR47k6quvpkePHowfP56FCxfSs2dPunTpwmmnncbq1auBw8+w77rrLkaNGkXv3r1p3bo1Dz300MH91atX7+D6vXv3ZsiQIbRv357MzMzAByivvfYa7du3p2vXrlx33XVlnrlv3bqVCy64gE6dOpGens6yZcsAmD9//sFvJF26dGHXrl1s2rSJM844g86dO3PyySfzwQcfhPcXZkwEtWhRvuXl9e235Vsea0JpXtkU+C7ocS7Qo+hKInINcCPurP+soKdaicjnwE7gDlWNaAaJ5huSm5vLxx9/TLVq1di5cycffPAB1atX55133uG2227jueeeO2KbVatWMW/ePHbt2kW7du0YM2bMEe3NP//8c1asWMEJJ5xAr169+Oijj0hLS+Oqq65iwYIFtGrViuHDh5cZ38SJE+nSpQsvvvgi7733Hpdffjk5OTncf//9TJkyhV69erF7925q165NVlYWffv25fbbb+fAgQPsKfppaUwMmzzZlVKC/2yTk93ycGjRwpVrilseD8J2MVZVp6jqicAtwB3e4k1AC1XtgvsQmCEiRxXdVkRGi0i2iGRv2bKlUnFE+pM92CWXXEK1atUA2LFjB5dccgknn3wy48aNY8WKFcVuM2DAAGrVqkXjxo059thj2bx58xHrdO/enWbNmpGUlETnzp1Zv349q1atonXr1gfbpoeS6D/88EMuu+wyAM466yzy8vLYuXMnvXr14sYbb+Shhx5i+/btVK9enW7duvHkk09y1113sXz5curXr1/RX4sxUZeZCVlZkJoKIu4+K8stD4fJk90HR7BwfpBEWiiJfiPQPOhxM29ZSWYCFwCo6i+qmuf9vBhX629bdANVzVLVNFVNS0lJCTX2YkXzDalbt+7Bn//0pz/Rp08fvvjiC15++eUS25HXqlXr4M/VqlWjoKCgQutUxoQJE5g6dSp79+6lV69erFq1ijPOOIMFCxbQtGlTRo4cyX//+9+wHtOYSMvMhPXrobDQ3YcryQf2HckPkkgLJdEvAtqISCsRqQkMA+YGryAibYIeDgC+9paniEg17+fWQBtgbTgCL4lfb8iOHTto2rQpAE899VTY99+uXTvWrl3L+vXrAZg1a1aZ25x++ulM95oFvP/++zRu3JijjjqKb775hl//+tfccsstdOvWjVWrVrFhwwaOO+44rrzySq644gqWLFkS9tdgTDyL5AdJpJVZo1fVAhG5FngT1wLnCVVdISKTcFd55wLXishvgXxgGzDC2/wMYJKI5AOFwNWqujUSLyRYZmb034Tx48czYsQI7rnnHgYMGBD2/depU4eHH36Yfv36UbduXbp161bmNoGLv506dSI5OZlp06YB8OCDDzJv3jySkpLo2LEj/fv3Z+bMmdx3333UqFGDevXq2Rm9MQlEAi06YkVaWppmZ2cftuzLL7/kpJNO8imi2LF7927q1auHqnLNNdfQpk0bxo0b53dYxbL3zJjoEpHFqppW3HPWMzaOPPbYY3Tu3JmOHTuyY8cOrrrqKr9DMsbEgbgYvdI448aNi9kzeGNM7LIzemOMSXCW6I0xJsFZojfGmARnid4YYxKcJfoQ9OnThzfffPOwZQ8++CBjxowpcZvevXsTaCZ63nnnsX379iPWueuuu7j//vtLPfaLL77IypUrDz6+8847eeedd8oTfrFsOGNjqg5L9CEYPnw4M2fOPGzZzJkzQxpvBtyokw0aNKjQsYsm+kmTJvHb3/62QvsyxlRNcde88oYbICcnvPvs3BkefLDk54cMGcIdd9zB/v37qVmzJuvXr+f777/n9NNPZ8yYMSxatIi9e/cyZMgQ/vznPx+xfcuWLcnOzqZx48ZMnjyZadOmceyxx9K8eXO6du0KuDbyWVlZ7N+/n1/96lf873//Iycnh7lz5zJ//nzuuecennvuOe6++24GDhzIkCFDePfdd7n55pspKCigW7duPPLII9SqVYuWLVsyYsQIXn75ZfLz85kzZw7t27cv8fVt3bqVUaNGsXbtWpKTk8nKyqJTp07Mnz+f66+/HnDTAy5YsIDdu3eTkZHBzp07KSgo4JFHHuH000+v3BtgjIkoO6MPQaNGjejevTuvv/464M7mhw4diogwefJksrOzWbZsGfPnzz845ntxFi9ezMyZM8nJyeG1115j0aJFB5+76KKLWLRoEUuXLuWkk07i8ccf57TTTmPQoEHcd9995OTkcOKJJx5cf9++fYwcOZJZs2axfPnyg0k3oHHjxixZsoQxY8aUWR4KDGe8bNky/vKXv3D55ZcDHBzOOCcnhw8++IA6deowY8YM+vbtS05ODkuXLqVz584V+p0aY6In7s7oSzvzjqRA+Wbw4MHMnDmTxx9/HIDZs2eTlZVFQUEBmzZtYuXKlXTq1KnYfXzwwQdceOGFJHvDaw4aNOjgc1988QV33HEH27dvZ/fu3fTt27fUeFavXk2rVq1o29YNBjpixAimTJnCDTfcALgPDoCuXbvy/PPPl7qvDz/88ODY+cUNZ5yZmclFF11Es2bN6NatG6NGjSI/P58LLrjAEr0xccDO6EM0ePBg3n33XZYsWcKePXvo2rUr69at4/777+fdd99l2bJlDBgwoMThicsycuRI/v3vf7N8+XImTpxY4f0EBIY6rswwxzacsTHREemJxy3Rh6hevXr06dOHUaNGHbwIu3PnTurWrcvRRx/N5s2bD5Z2SnLGGWfw4osvsnfvXnbt2sXLL7988Lldu3bRpEkT8vPzDw4tDFC/fn127dp1xL7atWvH+vXrWbNmDQD/+9//OPPMMyv02mw4Y2P8E42Jx+OudOOn4cOHc+GFFx5sgXPKKafQpUsX2rdvT/PmzenVq1ep25966qlkZGRwyimncOyxxx421PDdd99Njx49SElJoUePHgeT+7Bhw7jyyit56KGHePbZZw+uX7t2bZ588kkuueSSgxdjr7766gq9LhvO2Bj/lDbPdbiGW7dhik1E2HtmTGiSktyZfFEibpKTUNkwxcYYE6OiMc+1JXpjjPFRNOa5jptEH2slJlMye6+MCV005rkOKdGLSD8RWS0ia0RkQjHPXy0iy0UkR0Q+FJEOQc/d6m23WkRKbxxegtq1a5OXl2cJJA6oKnl5edSuXdvvUIyJG5GeeLzMVjciUg2YApwD5AKLRGSuqq4MWm2Gqj7qrT8IeADo5yX8YUBH4ATgHRFpq6oHyhNks2bNyM3NZcuWLeXZzPikdu3aNGvWzO8wjDGeUJpXdgfWqOpaABGZCQwGDiZ6Vd0ZtH5dIHDqPRiYqaq/AOtEZI23v0/KE2SNGjVo1apVeTYxxhjjCSXRNwW+C3qcC/QoupKIXAPcCNQEzgra9tMi2zYtZtvRwGiAFuG81GyMMSZ8F2NVdYqqngjcAtxRzm2zVDVNVdNSUlLCFZIxxhhCS/QbgeZBj5t5y0oyE7iggtsaY4wJszJ7xopIdeAr4Gxckl4EXKqqK4LWaaOqX3s/nw9MVNU0EekIzMDV5U8A3gXalHYxVkS2ABsq8ZoaAz9VYnu/xGvcYLH7xWL3R6zGnqqqxZZEyqzRq2qBiFwLvAlUA55Q1RUiMgnIVtW5wLUi8lsgH9gGjPC2XSEis3EXbguAa8pqcVNSoKESkeySugHHsniNGyx2v1js/ojH2EMa1ExVXwNeK7LszqCfry9l28lAGPt4GWOMKY+46RlrjDGmYhIx0Wf5HUAFxWvcYLH7xWL3R9zFHnPDFBtjjAmvRDyjN8YYE8QSvTHGJLiESfRljbAZq0SkuYjME5GVIrJCREpswRSrRKSaiHwuIq/4HUt5iEgDEXlWRFaJyJci0tPvmEIhIuO8v5UvROQZEYnpoUJF5AkR+VFEvgha1khE3haRr737hn7GWJwS4r7P+3tZJiIviEgDP2MMVUIk+qARNvsDHYDhwUMlx7gC4CZV7QCkA9fEUewB1wNf+h1EBfwTeENV2wOnEAevQUSaAtcBaap6Mq5vyzB/oyrTU0C/IssmAO+qahtcR8pYPDl7iiPjfhs4WVU74TqS3hrtoCoiIRI9QSNsqup+3DAMg32OKSSquklVl3g/78IlmyMGfotVItIMGABM9TuW8hCRo4EzgMcBVHW/qm73N6qQVQfqeL3Wk4HvfY6nVKq6ANhaZPFgYJr38zQODZsSM4qLW1XfUtUC7+GnuGFdYl6iJPriRtiMm2QZICItgS7AZ/5GUi4PAuOBckxjHBNaAVuAJ72y01QRqet3UGVR1Y3A/cC3wCZgh6q+5W9UFXKcqm7yfv4BOM7PYCpoFPC630GEIlESfdwTkXrAc8ANRcb3j1kiMhD4UVUX+x1LBVQHTgUeUdUuwM/EZvngMF4tezDug+oEoK6I/M7fqCpHXRvvuGrnLSK348qu0/2OJRSJkujjepRMEamBS/LTVfV5v+Mph17AIBFZjyuXnSUiT/sbUshygVxVDXx7ehaX+GPdb4F1qrpFVfOB54HTfI6pIjaLSBMA7/5Hn+MJmYiMBAYCmRonHZESJdEvAtqISCsRqYm7ODXX55hCIiKCqxN/qaoP+B1PeajqraraTFVb4n7n76lqXJxdquoPwHci0s5bdDZBs6bFsG+BdBFJ9v52ziYOLiIXYy7e4Ife/Us+xhIyEemHK1UOUtU9fscTqoRI9N7FkcAIm18Cs4OHUY5xvYDLcGfDOd7tPL+DqiL+AEwXkWVAZ+AvPsdTJu8byLPAEmA57n84prvki8gzuOlD24lIroj8HvgbcI6IfI37lvI3P2MsTglx/xuoD7zt/a8+6muQIbIhEIwxJsElxBm9McaYklmiN8aYBGeJ3hhjEpwlemOMSXCW6I0xJsFZojfGmARnid4YYxLc/weZ0i4Svg210AAAAABJRU5ErkJggg==\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":["evaluate_model(tf.keras.models.load_model(GLOBAL_MODELS_FOLDER + '/TwoDense_128_256.h5'))"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":743},"id":"FYyBulD-HouS","executionInfo":{"status":"ok","timestamp":1642021763213,"user_tz":-60,"elapsed":21465,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"e0cdde52-2eeb-4ca1-83b0-edb9bd220b00"},"execution_count":6,"outputs":[{"output_type":"stream","name":"stdout","text":["Classification report: \n","              precision    recall  f1-score   support\n","\n","           0     0.7578    0.6250    0.6850       696\n","           1     0.8326    0.9033    0.8665      1437\n","\n","    accuracy                         0.8125      2133\n","   macro avg     0.7952    0.7641    0.7758      2133\n","weighted avg     0.8082    0.8125    0.8073      2133\n","\n"]},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 2 Axes>"]},"metadata":{"needs_background":"light"}},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAYoAAAEWCAYAAAB42tAoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3gUVffA8e+hGTqKivQmQqgBAgiI9CJVpCNVEBQRX8XeRX68FiwviiICYkNUUJooCCJgoRNakA4hFEGk15Tz+2M3YQkpC2T7+TzPPuzszs6cDJs5uffOnCuqijHGGJOWLL4OwBhjjH+zRGGMMSZdliiMMcakyxKFMcaYdFmiMMYYky5LFMYYY9JlicIYY0y6LFGYoCMiu0XkrIicEpGDIjJZRPKkWKeeiPwiIidF5LiIzBaRiinWySci74pIjHNbO5zLN3r3JzLGtyxRmGDVTlXzABFAdeCZpDdEpC4wH5gJFAFKA+uA30WkjHOdHMBCoBLQCsgH1AWOALU9FbSIZPPUto25WpYoTFBT1YPAPBwJI8kbwGeq+j9VPamq/6rq88Ay4GXnOn2AEkBHVY1W1URVPaSqr6rq3NT2JSKVRORnEflXRP4WkWedr08WkZEu6zUSkViX5d0i8pSIrAdOO59PS7Ht/4nIGOfz/CIyUUQOiMg+ERkpIlmv8VAZkyZLFCaoiUgx4C5gu3M5F1AP+DaV1b8BmjufNwN+UtVTbu4nL7AA+AlHK+VWHC0Sd/UA2gAFgKlAa+c2cSaBrsAU57qTgXjnPqoDLYCBV7AvY66IJQoTrGaIyElgL3AIeMn5+g04vvcHUvnMASBp/KFgGuukpS1wUFXfUtVzzpbK8iv4/BhV3auqZ1V1D7AG6Oh8rwlwRlWXiUghoDXwH1U9raqHgHeA7lewL2OuiCUKE6zuVtW8QCOgAhcTwFEgESicymcKA/84nx9JY520FAd2XFWkDntTLE/B0coA6MnF1kRJIDtwQESOicgx4CPg5mvYtzHpskRhgpqqLsbRVTPauXwa+BPoksrqXbnYXbQAaCkiud3c1V6gTBrvnQZyuSzfklqoKZa/BRo5u846cjFR7AXOAzeqagHnI5+qVnIzTmOumCUKEwreBZqLSDXn8tNAXxEZJiJ5ReR652BzXeAV5zqf4zgpTxeRCiKSRUQKisizItI6lX3MAQqLyH9E5Drndus434vCMeZwg4jcAvwno4BV9TDwK/AJsEtVNztfP4Djiq23nJfvZhGRsiLS8CqOizFusURhgp7zpPsZ8KJz+TegJXAPjnGIPTgGhe9Q1W3Odc7jGND+C/gZOAGswNGFddnYg6qexDEQ3g44CGwDGjvf/hzH5be7cZzkv3Yz9CnOGKakeL0PkAOIxtGVNo0r6yYz5oqITVxkjDEmPdaiMMYYky6PJQoRmSQih0RkYxrvi4iMEZHtIrJeRGp4KhZjjDFXz5Mtisk4Sh+k5S6gnPMxCPjQg7EYY4y5Sh5LFKq6BPg3nVU64CijoKq6DCggIjYgZ4wxfsaXBciKculNRrHO1y67G1ZEBuFodZA7d+6aFSpU8EqAxhjjj/49fYFjZ+LcWvf0v2e5cCYOEvf/o6o3Xc3+AqJSpaqOB8YDREZG6qpVq3wckTEm0E1ZHsPMqH2+DuOq/LPrX/IBdUrfkOr7SVezigjbF+/l/MkLbJrTe8/V7s+XiWIfjrIHSYo5XzPGhBBfnbCX73L0jKd1svVndUrfQIeIovSsU+Ky9/btO8GDD/5At26VuPfeqjC4LgAiva96f75MFLOAoSIyFagDHHfedWqM8RPeOIn76oSd3sk2EKkqEyas4fHHfyYuLoE2bcpl2rY9lihE5CscBdludNbefwlHMTNUdRwwF0cVzO3AGaC/p2IxJhRlxkneGyfxYDth+8KOHf9y//2zWbRoN40bl+Ljj9tRtmzm/Z95LFGoao8M3lfgIU/t35hQkF4yyIyTvJ3EA8OGDYdYvfoA48e3ZeDAGohIpm4/4Ep42GC2CQXutgYySgZ2kg9eGzceYs2aA/Tp46h1eeTIGQoWzJXm+iKyWlUjr2ZfAXHVkzHBKq2E4G5rwP7iDz0XLiQwatRSRo1aSqFCeejatRJhYdnSTRLXyhKFMR6UUcsgrYRgCcCkZvnyWAYMmMWmTYfp1asq77zTkrAwz5/GLVEYk4lSJoaMWgaWEIy79u07QYMGn1CoUB7mzOlBmza3eW3fliiMccPVjhlYIjDXauvWI9x2W0GKFs3H1193pmnTMuTLd51XY7BEYYxTZlxBZInBZJZjx87x5JM/M2HCGn79tR933lmSjh3DfRKLJQoT0lyTQ3rJwBKA8aZZs7bw4IM/cPDgKZ54oh61ahXxaTyWKEzISK3F4JocLBkYfzBw4CwmTlxLlSo3M3NmdyIjfZskwBKFCQFJCSK1FoMlB+MPXIv4RUYWoWTJ/Dz11B3kyJHVx5E5WKIwQW3K8hie/X4DYEnB+Ke9e4/zwAM/0L17JXr3rsYDD1zVPXEeZYnCBJW0Lk8d1bGKJQjjVxITlY8+WsVTTy0gIUHp2NF/59mxRGGCRsrWQ9K/1oow/mbbtiMMHDibJUv20KxZGcaPb0vp0tf7Oqw0WaIwQcE1SVjrwfi76OjDrF//N5Mmtadfv4hML+KX2SxRmICSUW0kSxLGX61bd5CoqIP07RtBhw4V2LmzJNdfn9PXYbnFEoXxS1daLM+6mIy/On8+npEjl/Daa79TuHAeunWrTFhYtoBJEmCJwvgRd25+s4RgAsmff+5lwIBZbN78D336VOPtt1t4pYhfZgu8iE3QSK+AniUEE+j27TtBw4aTueWWPMyd25O77sq8qUm9zRKF8Qm7QskEq82bDxMefhNFi+bjm2+60LRpafLm9W4Rv8xmicJ4nV2hZILR0aNnGT58Pp98EsWSJf1o0KAkd9/tv/dGXAlLFMZrUpbSsCRhgsX3329myJC5HD58mmeeuYNatYr6OqRMZYnCeEx6YxDWxWSCxX33zeSTT6KIiLiFH37oSY0ahX0dUqazRGEyXVpF+CxBmGDhWsTv9tuLUa7cDTz+eD2yZ/ePIn6ZzRKFuWbWcjChZM+eYwwePIeePavQp081Bg2q6euQPM4ShbkmdvWSCRWJicqHH67k6acXoqp06VLR1yF5jSUKc1VsYNqEki1b/mHgwNn89lsMLVqU5aOP2lKqVAFfh+U1liiMW6x7yYSyLVuOsGnTISZP7kCfPtX8vohfZpOkQZlAERkZqatWrfJ1GCElte4lwBKECWpr1x4gKuog/ftXB+DYsXMUKBDm46iunoisVtWrmhXJWhQmTda9ZELRuXPxjBixmDfe+J2iRfPRo0cVwsKyBXSSuFaWKEyqbApRE4p+/z2GAQNmsWXLEfr3j+CttwKziF9msyNgLmGtCBOq9u07QePGn1K0aD7mzetFixZlfR2S37BEYZJZK8KEoujow1Ss6CjiN316Vxo3Lk2ePDl8HZZfyeLrAIz/SLqqaVTHKnw9uK4lCRPU/v33LP36zaBSpQ9YsmQPAO3albckkQprUYQ418teow+coE7pGyxBmKA3fXo0Dz00lyNHzvLccw2oXTu4ivhlNksUISxlV1PFwvnoEGG/MCa49es3g08/XUeNGoX56adeRETc4uuQ/J4lihBlc0KYUOJaxK9eveKEh9/I8OH1yJbNet/d4dFEISKtgP8BWYEJqvpaivdLAJ8CBZzrPK2qcz0ZUyhLbU5qSxIm2O3adZRBg+bQq1cV+vaNCIkifpnNY+lURLICY4G7gIpADxFJWUXreeAbVa0OdAc+8FQ8oS6pBeFaesOShAlmCQmJjBmznMqVP2TZslgCrAiFX/Fki6I2sF1VdwKIyFSgAxDtso4C+ZzP8wP7PRhPyLJuJhNqNm8+zIABs/jzz1juuutWxo1rS4kS+X0dVsDyZKIoCux1WY4F6qRY52Vgvog8DOQGmqW2IREZBAwCKFHCTnLuspvnTKjavv1ftmw5wuefd+Tee6uEXBG/zObrwewewGRVfUtE6gKfi0hlVU10XUlVxwPjwVEU0AdxBhy7ec6EmtWr97Nu3d/cd1912rUrz65dj5Av33W+DisoeDJR7AOKuywXc77magDQCkBV/xSRMOBG4JAH4wpq1oowoebs2TheeWUxo0f/QfHi+enZ01HEz5JE5vFkolgJlBOR0jgSRHegZ4p1YoCmwGQRCQfCgMMejCmoWSvChJolS/YwcOAstm37lwEDqjN6tBXx8wSPHVFVjReRocA8HJe+TlLVTSIyAlilqrOA4cDHIvIojoHtfhpoE2T4AWtFmFC0b98Jmjb9jOLF87FgQW+aNi3j65CClk1cFMBSJghrRZhQsGHD31SpUgiAOXO20rhxKXLntvpMGbGJi0KQdTOZUPPPP2d49NF5fPHFehYv7sedd5akbdvbfB1WSLBEEWCsm8mEGlXl22+jGTp0LkePnuOllxpSp47VJPMmSxQBZmbUvuQqr9aKMKGgb98ZfP75eiIji7BwYfvkbifjPZYoAlDFwvn4enBdX4dhjMe4FvFr2LAkVasW4j//ud2K+PmIHfUAMmV5THKXkzHBaufOozRr9jmTJ0cBMGBADR5/3Cq9+pId+QDhOnhtc0aYYJSQkMi77y6jSpUPWblyH1myWNkNf2FdTwHAivqZYBcdfZj77pvJ8uX7aNOmHOPGtaVYsXwZf9B4hSWKAOA6l7UlCROMdu06yo4dR5ky5R66d69sRfz8jCUKP5c0LmFzWZtgs3LlPqKiDnL//TVp0+Y2du4cRt68Vp/JH1mi8FMp75ewcQkTLM6ciePFFxfxzjvLKFkyP717VyMsLJslCT9micIP2V3XJlj9+utuBg6cxY4dRxk8uCavv97MivgFAPsf8kM2JmGCUWzsCZo3/5ySJfPzyy99aNy4tK9DMm6yROFnbEzCBJt16w5SrdotFCuWj5kzu9OoUSly5cru67DMFbD7KPzElOUxdPvoT7tXwgSNw4dP07PndCIiPmLx4t0AtG5dzpJEALIWhZ+wGk4mWKgqU6duZNiwnzh+/ByvvNKIunWLZ/xB47csUfgRq+FkgkHv3t/z5ZcbqFOnKBMntqdSpZt9HZK5Rm4nChHJpapnPBmMMSYwJSYqIo4ifo0bl6JmzcIMG1aHrFmtdzsYZPi/KCL1RCQa+Mu5XE1EPvB4ZCHEiv2ZQLZ9+780bfoZn3xysYjfo4/WtSQRRNz5n3wHaAkcAVDVdcCdngwqlFixPxOo4uMTGT36D6pU+ZC1aw+QI0dWX4dkPMStridV3Zui9kqCZ8IJLVbszwSqjRsP0b//TFat2k+HDuX54IM2FCmS19dhGQ9xJ1HsFZF6gIpIduARYLNnwwp+liRMIIuJOc6ePceYOrUTXbtWsiJ+Qc6dRPEA8D+gKLAPmA8M8WRQocDuvjaBZvnyWNat+5tBg2rSunU5du58hDx5cvg6LOMF7oxRlFfVe1W1kKrerKq9gHBPBxbM7O5rE0hOn77AY4/No27dibzxxu+cPx8PYEkihLiTKN5z8zXjpqTWhA1eG3/3yy+7qFp1HO+8s4wHHohkzZrBXHed3X4VatL8HxeRukA94CYReczlrXyAXd5wFZJKhyfdgW2tCePPYmNP0LLlF5QuXYDFi/tx550lfR2S8ZH0/jTIAeRxruN6OcMJoLMngwpGqZUON8YfrV17gOrVC1OsWD5mz+5Bw4YlyZnT6jOFsjQThaouBhaLyGRV3ePFmIKOXeFkAsHff59i2LCf+OabTfz6a18aNixFq1a3+jos4wfc6Ww8IyJvApWAsKQXVbWJx6IKMnaFk/FnqsqXX27gkUd+4tSpC4wc2Zh69ayIn7nInUTxJfA10BbHpbJ9gcOeDCoY2ZiE8Vc9e37H1KkbqVu3GBMntic8/CZfh2T8jDuJoqCqThSRR1y6o1Z6OrBg4XoprDH+wrWIX4sWZahbtxgPPVTL6jOZVLmTKOKc/x4QkTbAfsDOem6wOk7GH23deoT7759Nnz5VGTCgBv37V/d1SMbPuZMoRopIfmA4jvsn8gH/8WhUQcLGJow/iY9P5O23/+Sll34lLCybXclk3JZholDVOc6nx4HGACJS35NBBRMbmzD+YP36v7nvvpmsXn2Ajh0rMHZsawoXtiJ+xj3p3XCXFeiKo8bTT6q6UUTaAs8COQFrr6bDxiaMP4mNPcHevSf49tsudOoUbkX8zBVJb+RqIjAQKAiMEZEvgNHAG6rqVpIQkVYiskVEtovI02ms01VEokVkk4hMudIfwF9ZmQ7ja3/8sZdx41YBOIv4DaNz54qWJMwVS6/rKRKoqqqJIhIGHATKquoRdzbsbJGMBZoDscBKEZmlqtEu65QDngHqq+pREQmqyXWt28n4wqlTF3juuYW8994Kypa9gf79I7juumzkzm1F/MzVSa9FcUFVEwFU9Ryw090k4VQb2K6qO1X1AjAV6JBinfuBsap61LmfQ1ewfWNMCvPn76By5Q94770VPPRQLdasGWRF/Mw1S+8bVEFE1jufC1DWuSyAqmrVDLZdFNjrshwL1Emxzm0AIvI7jkKDL6vqTyk3JCKDgEEAJUrYX+jGpGbv3uO0aTOFsmWvZ8mS/txxh/2umMyRXqLwxpwT2YByQCOgGLBERKqo6jHXlVR1PDAeIDIyUr0Q11VzrRBbsXA+X4djQsDq1fupWbMIxYvnZ+7cnjRoUJKwMGtFmMyTZteTqu5J7+HGtvcBrgVjijlfcxULzFLVOFXdBWzFkTgClmuSsIFs40kHD56iS5dviYz8mMWLdwPQvHlZSxIm03nyG7USKCcipXEkiO5AzxTrzAB6AJ+IyI04uqJ2ejAmr6hYOB9fD67r6zBMkFJVPvtsHY8+Oo8zZ+IYNaqJFfEzHuWxRKGq8SIyFJiHY/xhkqpuEpERwCpVneV8r4WIRAMJwBNXOGBuTMjp3n0633yzifr1izNhQnsqVLjR1yGZIOdWohCRnEAJVd1yJRtX1bnA3BSvvejyXIHHnI+AZzfZGU9xLeLXuvWtNGhQgiFDapEli90TYTwvw1KRItIOiAJ+ci5HiMgsTwcWaKwAoPGUv/76hzvv/ISJE9cC0LdvBEOH1rYkYbzGnZrCL+O4J+IYgKpGAaU9GFPAsRnsjCfExSUwatRSqlUbR3T0YfLksRvmjG+4VWZcVY+nuO3fry9R9ZakS2GX7/oXsCRhMk9U1EH6959JVNRBOneuyHvv3cUtt+TxdVgmRLmTKDaJSE8gq7PkxjDgD8+GFRiSLoWtU/oGOkQUtSRhMs3Bg6c4ePAU06d35Z57vHFLkzFpcydRPAw8B5wHpuC4UmmkJ4MKBK4D13YprMkMv/0Ww/r1fzNkSC1atbqVHTuGkSuXzRlhfM+dRFFBVZ/DkSyMk1WHNZnl5MnzPPPMQsaOXUm5cjcwYEB1rrsumyUJ4zfcGcx+S0Q2i8irIlLZ4xEFEKsOa67VvHnbqVz5Qz74YCWPPFKHNWsGWxE/43cyTBSq2hjHzHaHgY9EZIOIPO/xyPxYUreTMddi797jtG37FblyZee33+7j3Xdb2ZVNxi+506JAVQ+q6hjgARz3VLyYwUeCmnU7maulqqxY4fj+FC+enx9/vJe1awdbCQ7j19y54S5cRF4WkQ3AeziueCrm8cj8nHU7mSt14MBJOnX6hjp1JiQX8WvWrIwV8TN+z51v6CTga6Clqu73cDzGBB1VZfLkKB57bD7nzsXz+uvNqF/f/sgwgSPDRKGqdu2nMdega9dpTJsWTYMGJZgwoT233VbQ1yEZc0XSTBQi8o2qdnV2Obneie3uDHfGhKyEhEREhCxZhHbtbqNJk1IMHhxp9ZlMQEqvRfGI89+23ggkUFiFWJORzZsPM2DALPr3j+D++2vSp081X4dkzDVJb4a7A86nQ1KZ3W6Id8LzP3bFk0lLXFwCI0cuISLiI7ZsOUL+/GG+DsmYTOHO5bHNU3ntrswOJJDYFU8mpbVrDxAZ+TEvvLCIjh0rsHnzQ3TtWsnXYRmTKdIbo3gQR8uhjIisd3krL/C7pwPzR9btZNLy99+n+eefM8yY0Y0OHSr4OhxjMlV6YxRTgB+B/wJPu7x+UlVD7rZkm5jIpLRkyR42bPibhx6qTatWt7J9+8PkzGn1mUzwSa/rSVV1N/AQcNLlgYiE3J/USWMTNueEOXHiPEOG/EDDhpMZM2YF58/HA1iSMEEroxZFW2A1jstjXa/rU6CMB+PyG0mTEyXNO2FJIrTNnbuNwYPnsH//SR577HZGjGhsRfxM0EvzG66qbZ3/hvS0p0lJomLhfNblFOL27j1Ohw5TKV++INOmdaFOnZCvZGNCRIZ/ColIfSBKVU+LSC+gBvCuqsZ4PDo/UbFwPpucKESpKsuX7+P224tRvHh+5s/vRf36JciRI6uvQzPGa9y5PPZD4IyIVAOGAzuAzz0alTF+YP/+k9x999fUrTsxuYhf48alLUmYkONOoohXVQU6AO+r6lgcl8gaE5RUlQkT1lCx4ljmz9/B6NHNrYifCWnujMKdFJFngN5AAxHJAtjlHSZode78Ld99t5mGDUsyYUJ7br015C7yM+YS7rQougHngftU9SCOuSje9GhUfsJmsgsdCQmJJCY6al/efXd5xo1rwy+/9LUkYQzuTYV6EPgSyC8ibYFzqvqZxyPzA1bXKTRs3HiI+vUnMXHiGgB6965mlV6NceHODHddgRVAF6ArsFxEOns6MF9zLddh904EpwsXEnjllV+pUeMjduw4yvXX5/R1SMb4JXfGKJ4DaqnqIQARuQlYAEzzZGC+ZOU6gt/q1fvp128mGzceomfPKrz7bktuuim3r8Myxi+5kyiyJCUJpyO4N7YRsKxcR/A7cuQsx46dY/bsHrRte5uvwzHGr7mTKH4SkXnAV87lbsBcz4XkW9blFLwWLdrFhg2HGDasDi1alGXbtocJC7PyG8ZkxJ3B7CeAj4Cqzsd4VX3K04H5ig1gB5/jx88xePBsmjT5jA8/XJVcxM+ShDHuSW8+inLAaKAssAF4XFX3eSswX7LWRPCYPXsLDzzwAwcPnuLxx+vyyitWxM+YK5Xeb8wk4DNgCdAOeA+4xxtBGZMZ9u49TqdO31Chwo3MmNGNWrWslWjM1UgvUeRV1Y+dz7eIyBpvBGTMtVBV/vwzlnr1ijuL+PWmXr3iVp/JmGuQ3hhFmIhUF5EaIlIDyJliOUMi0kpEtojIdhF5Op31OomIikjklf4AmcnuxA5ssbEnaN9+KvXrT0ou4teoUSlLEsZco/RaFAeAt12WD7osK9AkvQ2LSFZgLNAciAVWisgsVY1OsV5e4BFg+ZWFnvlsIDswJSYqH3+8miee+Jn4+ETefrsFd9xhY0zGZJb0Ji5qfI3brg1sV9WdACIyFUcF2ugU670KvA48cY37uyZ2WWzg6tTpG2bM+IsmTUrz8cftKFPmel+HZExQ8eSNc0WBvS7Lsc7Xkjm7sIqr6g/pbUhEBonIKhFZdfjw4cyPFGtNBJr4+ItF/Dp1Cufjj9uxYEFvSxLGeIDP7rB2lit/G8dkSOlS1fGqGqmqkTfddJPHYrLWRGBYv/5v6tadyMcfrwagV6+qDBxYAxEr4meMJ3gyUewDirssF3O+liQvUBn4VUR2A7cDs3wxoG2D2IHh/Pl4XnppETVrjmfPnmNWm8kYL3FnzmwB7gXKqOoIESkB3KKqKzL46EqgnIiUxpEgugM9k95U1ePAjS77+RXHTX2rrvinuEbW7eT/Vq7cR79+M4mOPkzv3lV5552WFCyYy9dhGRMS3LlF9QMgEcdVTiOAk8B0oFZ6H1LVeBEZCswDsgKTVHWTiIwAVqnqrGuKPJNZt5N/O3r0HKdOXWDu3J7cdVc5X4djTEhxJ1HUUdUaIrIWQFWPikgOdzauqnNJUUBQVV9MY91G7mwzs7le7WT8yy+/7GLDhr955JHbadGiLFu3DrXyG8b4gDtjFHHOeyIUkuejSPRoVF5k3U7+59ixc9x//yyaNv2Mjz5anVzEz5KEMb7hTqIYA3wP3Cwi/wf8BozyaFReYvdO+J+ZM/+iYsWxTJoUxZNP1mP16kGWIIzxsQx/A1X1SxFZDTQFBLhbVTd7PDIPs1ns/E9MzHG6dPmW8PCbmDWrB5GRRXwdkjEG9656KgGcAWa7vqaqMZ4MzJNck4TNYudbqspvv8XQoEFJSpTIz4IFfbj99mJWn8kYP+JO19MPwBznvwuBncCPngzK02yqU/8QE3OcNm2mcOedk5OL+N15Z0lLEsb4GXe6nqq4LjvLbgzxWEQeZuMSvpeYqIwbt4qnnlqAqjJmTCsr4meMH7viUUJVXSMidTwRjKfZuIR/uOeer5k5cwvNm5dh/Ph2lCpVwNchGWPS4c4YxWMui1mAGsB+j0XkQdbl5Dvx8YlkySJkySJ061aJDh3K069fhNVnMiYAuDNGkdflcR2OsYoOngzKk6zLyfvWrTtInToTGD/eUcSvR48q9O9f3ZKEMQEi3RaF80a7vKr6uJfiMUHk3Ll4Ro5cwuuv/84NN+Tkllvy+DokY8xVSDNRiEg2Z72m+t4MyASHFSv20bfvDP766x/69q3G22+35IYbcvo6LGPMVUivRbECx3hElIjMAr4FTie9qarfeTg2E8BOnDjP2bNx/PTTvbRseauvwzHGXAN3rnoKA47gqB6rOO7OVsAShbnE/Pk72LTpEI8+WpdmzcqwZYsV8TMmGKQ3mH2z84qnjcAG57+bnP9u9EJsmcomJ/Kco0fP0r//TFq2/IKJE9daET9jgkx6v8lZgTw4WhApqWfC8RyrEusZ3323mYcemsvhw6d55pk7ePHFhpYgjAky6f1GH1DVEV6LxAvs0tjMFRNznO7dp1G58s3MnduT6tUL+zokY4wHpJco7CJ3cxlVZcmSPTRsWIoSJfLzyy99qVOnKNmzW30mY4JVemMUTb0WhYfZ+ETm2LPnGHfd9SWNGn2aXMTvjjtKWJIwJsil2aJQ1aA5s9r4xLVJTFQ++GAlTz+9AID33ruLBg1K+jgqY4y3BP2oo1WLvXZ33zcUGL8AABkBSURBVD2V2bO30rJlWT76qC0lS1oRP2NCSdAnCmtNXJ24uASyZs1ClixCjx6V6dy5Ir17V7X6TMaEIHeKAgY8a01cmTVrDlC79gTGjVsFOIr49elTzZKEMSEqqBOFDWJfmbNn43jmmQXUrv0xBw+eonjxfL4OyRjjB4K668m6ndy3bFksffvOYOvWI9x3XwSjR7fg+uutiJ8xJsgTBVi3k7tOn75AXFwCP//cm2bNyvg6HGOMHwn6RGHS9tNP29m06RDDh9ejadMy/PXXUHLksHsijDGXCtoxChufSNuRI2fo23cGd931JZ9+uo4LFxIALEkYY1IVtInCxicup6pMmxZNxYofMGXKBp5/vgErV95vCcIYk66g7nqy8YlLxcQcp2fP6VStWoj583tRrdotvg7JGBMAgjpRGEcrYtGi3TRpUpqSJQvw66/9qF27KNmyBW1j0hiTyexsEcR27TpKixZf0LTpZ8lF/OrVK25JwhhzRYLyjBHqA9kJCYn873/LqFz5Q5Yvj+XDD9tYET9jzFULyq6nUB/I7tBhKj/8sI3WrcsxblwbihfP7+uQjDEBLOgSRahWi3Ut4te7d1V69KhMz55VrD6TMeaaebTrSURaicgWEdkuIk+n8v5jIhItIutFZKGIXHP/SCi2Jlat2k9k5Md8+OFKALp1q8y991qlV2NM5vBYohCRrMBY4C6gItBDRCqmWG0tEKmqVYFpwBuZse9QaU2cPRvHU0/9TJ06Ezh8+LTNE2GM8QhPdj3VBrar6k4AEZkKdACik1ZQ1UUu6y8DenkwnqDy55976dt3Btu2/cvAgdV5880WFCgQ5uuwjDFByJOJoiiw12U5FqiTzvoDgB9Te0NEBgGDAEqUCP6WgjvOno0nMVFZsKA3TZtaET9jjOf4xWC2iPQCIoGGqb2vquOB8QCRkZHqxdD8yty529i06RBPPFGfJk1Ks3nzQ2TPbuU3jDGe5cnB7H1AcZflYs7XLiEizYDngPaqev5adhis90/8888ZevX6jjZtpvDllxuSi/hZkjDGeIMnE8VKoJyIlBaRHEB3YJbrCiJSHfgIR5I4dK07DLYrnlSVqVM3Eh4+lm++2cRLLzVkxQor4meM8S6PdT2paryIDAXmAVmBSaq6SURGAKtUdRbwJpAH+NZ5KWeMqra/lv0G0xVPMTHH6dt3BtWqFWLixPZUqVLI1yEZY0KQR8coVHUuMDfFay+6PG/myf0HIlVl4cJdNGtWhpIlC7B4cT9q1SpC1qxBWW3FGBMA7OzjR3bs+JemTT+jefPPk4v43X57MUsSxhifCpozUCAPZCckJPL2239SpcqHrF59gI8+amtF/IwxfsMvLo/NDIE8kN2u3Vf8+ON22ra9jQ8/bEOxYvl8HZIxxiQLikQRiIUAL1xIIFs2RxG/fv0i6N27Kt27V7b6TMYYvxPwXU9Tlsfw7PcbgMBpTaxYsY+aNcfzwQeOIn5du1aiRw+r9GqM8U8BnyiSupxGdazi962JM2fiGD58HnXrTuTo0bOULXu9r0MyxpgMBUXXUyB0Of32Wwx9+85g586jDB5ck9dfb0b+/FbEzxjj/4IiUQQCx8RCwqJFfWnUqJSvwzHGGLdZovCg2bO3sHnzPzz5ZH0aNy5NdPRDZMsW8L19xpgQY2ctDzh8+DQ9e06nffupfPXVxuQifpYkjDGBKKDPXP52k52qMmXKBsLDxzJtWjQjRjRi+fKBVsTPGBPQArrryd9usouJOU7//jOpXv0WJk5sT6VKN/s6JGOMuWYBnSjA91c8JSYqP/+8g5Ytb6VkyQIsXdqfmjULW30mY0zQsLPZNdi27QhNmnxKq1ZfsmTJHgBq1y5qScIYE1TsjHYV4uMTefPN36ladRxRUQeZOLE9DRr4930cxhhztQK268m1vpO3tW07hXnzdtChQ3k++KANRYrk9XoMxv/FxcURGxvLuXPnfB2KCSFhYWEUK1aM7NmzZ9o2AzZReHsg+/z5eLJnz0qWLMLAgTW4777qdOlS0eozmTTFxsaSN29eSpUqZd8T4xWqypEjR4iNjaV06dKZtt2A7nry1kD2smWx1KgxnrFjVwDQuXNFunatZL/8Jl3nzp2jYMGC9j0xXiMiFCxYMNNbsQGdKDzt9OkLPProT9SrN5GTJ89TrlxBX4dkAowlCeNtnvjOBWzXk6ctXbqHvn1nsGvXMYYMieS//21GvnzX+TosY4zxOmtRpCE+PpHs2bOyeHE/xo5tY0nCBKSsWbMSERFB5cqVadeuHceOHUt+b9OmTTRp0oTy5ctTrlw5Xn31VVQ1+f0ff/yRyMhIKlasSPXq1Rk+fLgvfoR0rV27lgEDBvg6jDSdP3+ebt26ceutt1KnTh12796d6nrvvPMOlSpVonLlyvTo0SO566hfv36ULl2aiIgIIiIiiIqKAmDOnDm8+OKL3voxHIMfgfSoWbOmfrlsj5Z8ao52HfeHZqbvv9+so0YtSV6Oi0vI1O2b0BIdHe3rEDR37tzJz/v06aMjR45UVdUzZ85omTJldN68eaqqevr0aW3VqpW+//77qqq6YcMGLVOmjG7evFlVVePj4/WDDz7I1Nji4uKueRudO3fWqKgor+7zSowdO1YHDx6sqqpfffWVdu3a9bJ1YmNjtVSpUnrmzBlVVe3SpYt+8sknqqrat29f/fbbby/7TGJiokZEROjp06dT3W9q3z1glV7leTcgu54y+4qnv/8+xcMP/8i330ZTo0Zhhg+vR44cWa2In8k0r8zeRPT+E5m6zYpF8vFSu0pur1+3bl3Wr18PwJQpU6hfvz4tWrQAIFeuXLz//vs0atSIhx56iDfeeIPnnnuOChUqAI6WyYMPPnjZNk+dOsXDDz/MqlWrEBFeeuklOnXqRJ48eTh16hQA06ZNY86cOUyePJl+/foRFhbG2rVrqV+/Pt999x1RUVEUKFAAgHLlyvHbb7+RJUsWHnjgAWJiYgB49913qV+//iX7PnnyJOvXr6datWoArFixgkceeYRz586RM2dOPvnkE8qXL8/kyZP57rvvOHXqFAkJCcydO5eHH36YjRs3EhcXx8svv0yHDh3YvXs3vXv35vTp0wC8//771KtXz+3jm5qZM2fy8ssvA9C5c2eGDh2Kql42jhAfH8/Zs2fJnj07Z86coUiRIuluV0Ro1KgRc+bMoWvXrtcUozsCMlFA5lzxpKp88cV6/vOfeZw6dYH/+78mPPFEPbJntyJ+JrgkJCSwcOHC5G6aTZs2UbNmzUvWKVu2LKdOneLEiRNs3LjRra6mV199lfz587Nhg2M64qNHj2b4mdjYWP744w+yZs1KQkIC33//Pf3792f58uWULFmSQoUK0bNnTx599FHuuOMOYmJiaNmyJZs3b75kO6tWraJy5crJyxUqVGDp0qVky5aNBQsW8OyzzzJ9+nQA1qxZw/r167nhhht49tlnadKkCZMmTeLYsWPUrl2bZs2acfPNN/Pzzz8TFhbGtm3b6NGjB6tWrbos/gYNGnDy5MnLXh89ejTNmjW75LV9+/ZRvHhxALJly0b+/Pk5cuQIN954Y/I6RYsW5fHHH6dEiRLkzJmTFi1aJCdwgOeee44RI0bQtGlTXnvtNa67ztENHhkZydKlSy1ReFpMzHEGDpxNZGQRJk5sT4UKN2b8IWOuwpX85Z+Zzp49S0REBPv27SM8PJzmzZtn6vYXLFjA1KlTk5evvz7j6X27dOlC1qyOP8a6devGiBEj6N+/P1OnTqVbt27J242Ojk7+zIkTJzh16hR58uRJfu3AgQPcdNNNycvHjx+nb9++bNu2DREhLi4u+b3mzZtzww2Om3Pnz5/PrFmzGD16NOC4jDkmJoYiRYowdOhQoqKiyJo1K1u3bk01/qVLl2b4M16Jo0ePMnPmTHbt2kWBAgXo0qULX3zxBb169eK///0vt9xyCxcuXGDQoEG8/vrryWMTN998M/v378/UWNIScn0riYnKjz9uA6BkyQL8/vt9LFnSz5KECUo5c+YkKiqKPXv2oKqMHTsWgIoVK7J69epL1t25cyd58uQhX758VKpU6bL3r4Rr10rKa/pz586d/Lxu3bps376dw4cPM2PGDO655x4AEhMTWbZsGVFRUURFRbFv375LkkTSz+a67RdeeIHGjRuzceNGZs+efcl7rvtUVaZPn5687ZiYGMLDw3nnnXcoVKgQ69atY9WqVVy4cCHVn61BgwbJg8uujwULFly2btGiRdm7dy/g6F46fvw4BQteepn9ggULKF26NDfddBPZs2fnnnvu4Y8//gCgcOHCiAjXXXcd/fv3Z8WKFZcc15w5c6YaY2YLuETx7+kLVz0HxdatR2jUaDKtW09h8eLdAERGFrEifibo5cqVizFjxvDWW28RHx/Pvffey2+//ZZ8cjt79izDhg3jySefBOCJJ55g1KhRyX9VJyYmMm7cuMu227x58+TkAxe7ngoVKsTmzZtJTEzk+++/TzMuEaFjx4489thjhIeHJ59EW7RowXvvvZe8XtLVPq7Cw8PZvn178vLx48cpWtQxbjl58uQ099myZUvee++95Cu81q5dm/z5woULkyVLFj7//HMSEhJS/fzSpUuTk4zrI2W3E0D79u359NNPAcdYTZMmTS4bnyhRogTLli3jzJkzqCoLFy4kPDwccLSawJHcZsyYcUlX29atWy9Z9qSAO0MeO+NoTl7JQHZ8fCKvv/4bVat+yIYNh/jkkw7ceWdJT4VojF+qXr06VatW5auvviJnzpzMnDmTkSNHUr58eapUqUKtWrUYOnQoAFWrVuXdd9+lR48ehIeHU7lyZXbu3HnZNp9//nmOHj1K5cqVqVatGosWLQLgtddeo23bttSrV4/ChQunG1e3bt344osvkrudAMaMGcOqVauoWrUqFStWTDVJVahQgePHjyePFzz55JM888wzVK9enfj4+DT398ILLxAXF0fVqlWpVKkSL7zwAgBDhgzh008/pVq1avz111+XtEKu1oABAzhy5Ai33norb7/9Nq+99hoA+/fvp3Xr1gDUqVOHzp07U6NGDapUqUJiYiKDBg0C4N5776VKlSpUqVKFf/75h+effz5524sWLaJNmzbXHKM7JCmrBoobSoZr82cn8fXgum5/pmXLL5g/fwf33BPO2LGtueWWPBl/yJhrtHnz5uS/DI1nvPPOO+TNm5eBAwf6OhSv+vvvv+nZsycLFy5M9f3UvnsislpVI69mfwHXonDXuXPxJCQkAjBoUA2mTevC9OldLUkYE0QefPDB5KuAQklMTAxvvfWW1/YXlFc9/f57DAMGzGLIkFoMG1aHTp0q+jokY4wHhIWF0bt3b1+H4XW1atXy6v6CqkVx6tQFhg37kQYNPuHcuXjCw+1KJuNbgda1awKfJ75zQdOiWLx4N337ziAm5jhDh9Zm1Kim5MmTw9dhmRAWFhbGkSNHrNS48Rp1zkcRFhaWqdsNmkQBkCtXdpYu7U/9+jYtqfG9YsWKERsby+HDh30digkhSTPcZaaAThTffbeZv/76h2efbUDDhqXYsOFBuyfC+I3s2bNn6ixjxviKR8+qItJKRLaIyHYReTqV968Tka+d7y8XkVLubPfs8fN07vwNnTp9w/ff/8WFC44bYyxJGGNM5vPYfRQikhXYCjQHYoGVQA9VjXZZZwhQVVUfEJHuQEdV7ZbqBp2y5S+tWeIHkiVBefnlRgwfXteK+BljTAau5T4KT3Y91Qa2q+pOABGZCnQAol3W6QC87Hw+DXhfRETTyV4JJy5wa7WCzPy6M+XL21VNxhjjaZ5MFEWBvS7LsUCdtNZR1XgROQ4UBP5xXUlEBgGDnIvnt6x7cGOFCpfXxg9BN5LiWIUwOxYX2bG4yI7FReWv9oMBMZitquOB8QAisupqm0/Bxo7FRXYsLrJjcZEdi4tE5PLJNdzkydHffUBxl+ViztdSXUdEsgH5gSMejMkYY8wV8mSiWAmUE5HSIpID6A7MSrHOLKCv83ln4Jf0xieMMcZ4n8e6npxjDkOBeUBWYJKqbhKRETgm+Z4FTAQ+F5HtwL84kklGxnsq5gBkx+IiOxYX2bG4yI7FRVd9LAKuzLgxxhjvsjvUjDHGpMsShTHGmHT5baLwVPmPQOTGsXhMRKJFZL2ILBSRoJ3nNaNj4bJeJxFREQnaSyPdORYi0tX53dgkIlO8HaO3uPE7UkJEFonIWufvSWtfxOlpIjJJRA6JyMY03hcRGeM8TutFpIZbG1ZVv3vgGPzeAZQBcgDrgIop1hkCjHM+7w587eu4fXgsGgO5nM8fDOVj4VwvL7AEWAZE+jpuH34vygFrgeudyzf7Om4fHovxwIPO5xWB3b6O20PH4k6gBrAxjfdbAz8CAtwOLHdnu/7aokgu/6GqF4Ck8h+uOgCfOp9PA5pKcBb9z/BYqOoiVT3jXFyG456VYOTO9wLgVeB14Jw3g/Myd47F/cBYVT0KoKqHvByjt7hzLBTI53yeH9jvxfi8RlWX4LiCNC0dgM/UYRlQQEQKZ7Rdf00UqZX/KJrWOqoaDySV/wg27hwLVwNw/MUQjDI8Fs6mdHFV/cGbgfmAO9+L24DbROR3EVkmIq28Fp13uXMsXgZ6iUgsMBd42Duh+Z0rPZ8AAVLCw7hHRHoBkUBDX8fiCyKSBXgb6OfjUPxFNhzdT41wtDKXiEgVVT3m06h8owcwWVXfEpG6OO7fqqyqib4OLBD4a4vCyn9c5M6xQESaAc8B7VX1vJdi87aMjkVeoDLwq4jsxtEHOytIB7Td+V7EArNUNU5Vd+Eo+1/OS/F5kzvHYgDwDYCq/gmE4SgYGGrcOp+k5K+Jwsp/XJThsRCR6sBHOJJEsPZDQwbHQlWPq+qNqlpKVUvhGK9pr6pXXQzNj7nzOzIDR2sCEbkRR1fUTm8G6SXuHIsYoCmAiITjSBShOEftLKCP8+qn24Hjqnogow/5ZdeTeq78R8Bx81i8CeQBvnWO58eoanufBe0hbh6LkODmsZgHtBCRaCABeEJVg67V7eaxGA58LCKP4hjY7heMf1iKyFc4/ji40Tke8xKQHUBVx+EYn2kNbAfOAP3d2m4QHitjjDGZyF+7nowxxvgJSxTGGGPSZYnCGGNMuixRGGOMSZclCmOMMemyRGH8kogkiEiUy6NUOuueyoT9TRaRXc59rXHevXul25ggIhWdz59N8d4f1xqjcztJx2WjiMwWkQIZrB8RrJVSjffY5bHGL4nIKVXNk9nrprONycAcVZ0mIi2A0apa9Rq2d80xZbRdEfkU2Kqq/5fO+v1wVNAdmtmxmNBhLQoTEEQkj3OujTUiskFELqsaKyKFRWSJy1/cDZyvtxCRP52f/VZEMjqBLwFudX72Mee2NorIf5yv5RaRH0RknfP1bs7XfxWRSBF5DcjpjONL53unnP9OFZE2LjFPFpHOIpJVRN4UkZXOeQIGu3FY/sRZ0E1Eajt/xrUi8oeIlHfepTwC6OaMpZsz9kkissK5bmrVd425lK/rp9vDHqk9cNxJHOV8fI+jikA+53s34rizNKlFfMr573DgOefzrDhqP92I48Sf2/n6U8CLqexvMtDZ+bwLsByoCWwAcuO4830TUB3oBHzs8tn8zn9/xTn/RVJMLuskxdgR+NT5PAeOSp45gUHA887XrwNWAaVTifOUy8/3LdDKuZwPyOZ83gyY7nzeD3jf5fOjgF7O5wVw1H/K7ev/b3v498MvS3gYA5xV1YikBRHJDowSkTuBRBx/SRcCDrp8ZiUwybnuDFWNEpGGOCaq+d1Z3iQHjr/EU/OmiDyPowbQABy1gb5X1dPOGL4DGgA/AW+JyOs4uquWXsHP9SPwPxG5DmgFLFHVs87urqoi0tm5Xn4cBfx2pfh8ThGJcv78m4GfXdb/VETK4ShRkT2N/bcA2ovI487lMKCEc1vGpMoShQkU9wI3ATVVNU4c1WHDXFdQ1SXORNIGmCwibwNHgZ9VtYcb+3hCVaclLYhI09RWUtWt4pj3ojUwUkQWquoId34IVT0nIr8CLYFuOCbZAceMYw+r6rwMNnFWVSNEJBeO2kYPAWNwTNa0SFU7Ogf+f03j8wJ0UtUt7sRrDNgYhQkc+YFDziTRGLhsXnBxzBX+t6p+DEzAMSXkMqC+iCSNOeQWkdvc3OdS4G4RySUiuXF0Gy0VkSLAGVX9AkdBxtTmHY5ztmxS8zWOYmxJrRNwnPQfTPqMiNzm3Geq1DGj4TBguFwss59ULrqfy6oncXTBJZkHPCzO5pU4Kg8bky5LFCZQfAlEisgGoA/wVyrrNALWichaHH+t/09VD+M4cX4lIutxdDtVcGeHqroGx9jFChxjFhNUdS1QBVjh7AJ6CRiZysfHA+uTBrNTmI9jcqkF6pi6ExyJLRpYIyIbcZSNT7fF74xlPY5Jed4A/uv82V0/twiomDSYjaPlkd0Z2ybnsjHpsstjjTHGpMtaFMYYY9JlicIYY0y6LFEYY4xJlyUKY4wx6bJEYYwxJl2WKIwxxqTLEoUxxph0/T//cKJxXQwP7gAAAABJRU5ErkJggg==\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":["# CLAHE\n","We will try now to use the dataset in which the images are preprocesses with CLAHE. We will test the performance with the best network previuosly found and we will compare the results."],"metadata":{"id":"Eyo2gJ5yhR61"}},{"cell_type":"markdown","source":["## Load dataset from Google Drive\n","The dataset is stored in Google Drive in a zip file, we unzip it locally for better performance."],"metadata":{"id":"auFZf6kYhXuj"}},{"cell_type":"code","source":["drive.mount('/content/drive', force_remount=True)\n","!unzip -q '/content/drive/My Drive/Baldi_Cioffo/SplittedDataset_CLAHE.zip' -d '/content'"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1641669597754,"user_tz":-60,"elapsed":29984,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"7b313451-fa5f-461f-c0fd-c3e21145aaba","id":"Vjx3Cey8hXul"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Mounted at /content/drive\n"]}]},{"cell_type":"markdown","source":["## Set parameters"],"metadata":{"id":"VHh3wjhVhXum"}},{"cell_type":"code","source":["DATASET_PATH = \"/content/SplittedDataset_CLAHE\"\n","TRAINING_PATH = DATASET_PATH + '/training_set'\n","TRAINING_ALL_PATH = TRAINING_PATH + '/all'\n","TRAINING_HEM_PATH = TRAINING_PATH + '/hem'\n","VALIDATION_PATH = DATASET_PATH + '/validation_set'\n","TEST_PATH = DATASET_PATH + '/test_set'"],"metadata":{"id":"SlArazhmhXun"},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":["## Load dataset"],"metadata":{"id":"qmzpmcoQiEks"}},{"cell_type":"code","source":["set_seed()\n","\n","train_dataset = load_training_set(IMAGE_HEIGHT, IMAGE_WIDTH, BATCH_SIZE) # with default values\n","validation_dataset = load_validation_set(IMAGE_HEIGHT, IMAGE_WIDTH, BATCH_SIZE) # with default values\n","test_dataset = load_test_set(IMAGE_HEIGHT, IMAGE_WIDTH, BATCH_SIZE)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"wISYLHUXiHnJ","executionInfo":{"status":"ok","timestamp":1641669599365,"user_tz":-60,"elapsed":614,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"af653be2-193e-40d0-d3b1-24438245dff9"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Found 6396 files belonging to 2 classes.\n","Found 2132 files belonging to 2 classes.\n","Found 2133 files belonging to 2 classes.\n"]}]},{"cell_type":"markdown","source":["## Experiment increasing MaxPooling with two dropout"],"metadata":{"id":"wo0z-B6Gh4R3"}},{"cell_type":"code","source":["inputs = keras.Input(shape=(IMAGE_HEIGHT, IMAGE_WIDTH, 3))\n","x = layers.Rescaling(1./255)(inputs)\n","x = layers.Conv2D(filters=32, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=64, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(2,2))(x)\n","x = layers.Conv2D(filters=128, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(3,3))(x)\n","x = layers.Conv2D(filters=256, padding='same', kernel_size=(3,3), activation=\"relu\")(x)\n","x = layers.MaxPooling2D(pool_size=(5,5))(x)\n","x = layers.Flatten()(x)\n","x = layers.Dropout(0.5)(x)\n","x = layers.Dense(256, activation='relu') (x)\n","x = layers.Dropout(0.5)(x)\n","outputs = layers.Dense(1, activation='sigmoid')(x)\n","\n","model = keras.Model(inputs=inputs, outputs=outputs)\n","compile_model(model)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"tE6zg_fvhz2d","executionInfo":{"status":"ok","timestamp":1641391115020,"user_tz":-60,"elapsed":404,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"fd6d4e61-65f0-4705-a220-72bf4333c129"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Model: \"model\"\n","_________________________________________________________________\n"," Layer (type)                Output Shape              Param #   \n","=================================================================\n"," input_1 (InputLayer)        [(None, 300, 300, 3)]     0         \n","                                                                 \n"," rescaling (Rescaling)       (None, 300, 300, 3)       0         \n","                                                                 \n"," conv2d (Conv2D)             (None, 300, 300, 32)      896       \n","                                                                 \n"," max_pooling2d (MaxPooling2D  (None, 150, 150, 32)     0         \n"," )                                                               \n","                                                                 \n"," conv2d_1 (Conv2D)           (None, 150, 150, 64)      18496     \n","                                                                 \n"," max_pooling2d_1 (MaxPooling  (None, 75, 75, 64)       0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_2 (Conv2D)           (None, 75, 75, 128)       73856     \n","                                                                 \n"," max_pooling2d_2 (MaxPooling  (None, 25, 25, 128)      0         \n"," 2D)                                                             \n","                                                                 \n"," conv2d_3 (Conv2D)           (None, 25, 25, 256)       295168    \n","                                                                 \n"," max_pooling2d_3 (MaxPooling  (None, 5, 5, 256)        0         \n"," 2D)                                                             \n","                                                                 \n"," flatten (Flatten)           (None, 6400)              0         \n","                                                                 \n"," dropout (Dropout)           (None, 6400)              0         \n","                                                                 \n"," dense (Dense)               (None, 256)               1638656   \n","                                                                 \n"," dropout_1 (Dropout)         (None, 256)               0         \n","                                                                 \n"," dense_1 (Dense)             (None, 1)                 257       \n","                                                                 \n","=================================================================\n","Total params: 2,027,329\n","Trainable params: 2,027,329\n","Non-trainable params: 0\n","_________________________________________________________________\n"]}]},{"cell_type":"code","source":["run_model(model, \"OneDense256_2dropout_CLAHE\", epochs=50)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"m2weRVdXiQhz","executionInfo":{"status":"ok","timestamp":1641392678580,"user_tz":-60,"elapsed":1535583,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"cb09b1c6-a3df-4670-9d76-a72dcc5a4ab0"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Epoch 1/50\n","200/200 [==============================] - ETA: 0s - loss: 0.6946 - accuracy: 0.5073\n","Epoch 00001: val_loss improved from inf to 0.68611, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 57s 229ms/step - loss: 0.6946 - accuracy: 0.5073 - val_loss: 0.6861 - val_accuracy: 0.6764\n","Epoch 2/50\n","200/200 [==============================] - ETA: 0s - loss: 0.6550 - accuracy: 0.6123\n","Epoch 00002: val_loss improved from 0.68611 to 0.58582, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.6550 - accuracy: 0.6123 - val_loss: 0.5858 - val_accuracy: 0.7430\n","Epoch 3/50\n","200/200 [==============================] - ETA: 0s - loss: 0.6022 - accuracy: 0.6765\n","Epoch 00003: val_loss improved from 0.58582 to 0.52053, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.6022 - accuracy: 0.6765 - val_loss: 0.5205 - val_accuracy: 0.7725\n","Epoch 4/50\n","200/200 [==============================] - ETA: 0s - loss: 0.5586 - accuracy: 0.7140\n","Epoch 00004: val_loss improved from 0.52053 to 0.46401, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.5586 - accuracy: 0.7140 - val_loss: 0.4640 - val_accuracy: 0.7941\n","Epoch 5/50\n","200/200 [==============================] - ETA: 0s - loss: 0.5350 - accuracy: 0.7341\n","Epoch 00005: val_loss improved from 0.46401 to 0.44847, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 43s 214ms/step - loss: 0.5350 - accuracy: 0.7341 - val_loss: 0.4485 - val_accuracy: 0.7969\n","Epoch 6/50\n","200/200 [==============================] - ETA: 0s - loss: 0.5059 - accuracy: 0.7505\n","Epoch 00006: val_loss improved from 0.44847 to 0.43227, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.5059 - accuracy: 0.7505 - val_loss: 0.4323 - val_accuracy: 0.8166\n","Epoch 7/50\n","200/200 [==============================] - ETA: 0s - loss: 0.4782 - accuracy: 0.7710\n","Epoch 00007: val_loss did not improve from 0.43227\n","200/200 [==============================] - 43s 214ms/step - loss: 0.4782 - accuracy: 0.7710 - val_loss: 0.7775 - val_accuracy: 0.5713\n","Epoch 8/50\n","200/200 [==============================] - ETA: 0s - loss: 0.4674 - accuracy: 0.7842\n","Epoch 00008: val_loss did not improve from 0.43227\n","200/200 [==============================] - 43s 215ms/step - loss: 0.4674 - accuracy: 0.7842 - val_loss: 0.4532 - val_accuracy: 0.7964\n","Epoch 9/50\n","200/200 [==============================] - ETA: 0s - loss: 0.4433 - accuracy: 0.7933\n","Epoch 00009: val_loss did not improve from 0.43227\n","200/200 [==============================] - 43s 214ms/step - loss: 0.4433 - accuracy: 0.7933 - val_loss: 0.5202 - val_accuracy: 0.7655\n","Epoch 10/50\n","200/200 [==============================] - ETA: 0s - loss: 0.4368 - accuracy: 0.8028\n","Epoch 00010: val_loss improved from 0.43227 to 0.38211, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.4368 - accuracy: 0.8028 - val_loss: 0.3821 - val_accuracy: 0.8340\n","Epoch 11/50\n","200/200 [==============================] - ETA: 0s - loss: 0.4140 - accuracy: 0.8150\n","Epoch 00011: val_loss did not improve from 0.38211\n","200/200 [==============================] - 43s 214ms/step - loss: 0.4140 - accuracy: 0.8150 - val_loss: 0.3976 - val_accuracy: 0.8288\n","Epoch 12/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3970 - accuracy: 0.8261\n","Epoch 00012: val_loss did not improve from 0.38211\n","200/200 [==============================] - 43s 214ms/step - loss: 0.3970 - accuracy: 0.8261 - val_loss: 0.4474 - val_accuracy: 0.8044\n","Epoch 13/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3975 - accuracy: 0.8213\n","Epoch 00013: val_loss improved from 0.38211 to 0.35108, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3975 - accuracy: 0.8213 - val_loss: 0.3511 - val_accuracy: 0.8518\n","Epoch 14/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3730 - accuracy: 0.8418\n","Epoch 00014: val_loss did not improve from 0.35108\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3730 - accuracy: 0.8418 - val_loss: 0.3587 - val_accuracy: 0.8513\n","Epoch 15/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3795 - accuracy: 0.8332\n","Epoch 00015: val_loss improved from 0.35108 to 0.34629, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3795 - accuracy: 0.8332 - val_loss: 0.3463 - val_accuracy: 0.8583\n","Epoch 16/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3560 - accuracy: 0.8407\n","Epoch 00016: val_loss improved from 0.34629 to 0.33105, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 44s 216ms/step - loss: 0.3560 - accuracy: 0.8407 - val_loss: 0.3311 - val_accuracy: 0.8724\n","Epoch 17/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3627 - accuracy: 0.8404\n","Epoch 00017: val_loss improved from 0.33105 to 0.33091, saving model to /content/models/OneDense256_2dropout_CLAHE.h5\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3627 - accuracy: 0.8404 - val_loss: 0.3309 - val_accuracy: 0.8677\n","Epoch 18/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3405 - accuracy: 0.8521\n","Epoch 00018: val_loss did not improve from 0.33091\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3405 - accuracy: 0.8521 - val_loss: 0.3724 - val_accuracy: 0.8415\n","Epoch 19/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3476 - accuracy: 0.8485\n","Epoch 00019: val_loss did not improve from 0.33091\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3476 - accuracy: 0.8485 - val_loss: 0.3360 - val_accuracy: 0.8644\n","Epoch 20/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3263 - accuracy: 0.8580\n","Epoch 00020: val_loss did not improve from 0.33091\n","200/200 [==============================] - 43s 216ms/step - loss: 0.3263 - accuracy: 0.8580 - val_loss: 0.3336 - val_accuracy: 0.8593\n","Epoch 21/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3274 - accuracy: 0.8560\n","Epoch 00021: val_loss did not improve from 0.33091\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3274 - accuracy: 0.8560 - val_loss: 0.3313 - val_accuracy: 0.8743\n","Epoch 22/50\n","200/200 [==============================] - ETA: 0s - loss: 0.3149 - accuracy: 0.8632\n","Epoch 00022: val_loss did not improve from 0.33091\n","200/200 [==============================] - 43s 215ms/step - loss: 0.3149 - accuracy: 0.8632 - val_loss: 0.3459 - val_accuracy: 0.8560\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"}},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":["evaluate_model(tf.keras.models.load_model(GLOBAL_MODELS_FOLDER + '/OneDense256_2dropout_CLAHE.h5'))"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":741},"id":"9bwgGV03H-BM","executionInfo":{"status":"ok","timestamp":1641669615451,"user_tz":-60,"elapsed":10749,"user":{"displayName":"DANIELE CIOFFO","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s64","userId":"10062146559806518037"}},"outputId":"7d21585b-3f12-4981-8f92-752532be5cb2"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Classification report: \n","              precision    recall  f1-score   support\n","\n","           0     0.8382    0.7443    0.7884       696\n","           1     0.8825    0.9304    0.9058      1437\n","\n","    accuracy                         0.8697      2133\n","   macro avg     0.8603    0.8373    0.8471      2133\n","weighted avg     0.8680    0.8697    0.8675      2133\n","\n"]},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 2 Axes>"]},"metadata":{"needs_background":"light"}},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"needs_background":"light"}}]}]}