DL - Practica 1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 159,
   "id": "8cf3be7d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'2.10.0'"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "tf.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "id": "4c1ec241",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'2.10.0'"
      ]
     },
     "execution_count": 160,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "keras.__version__"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "db8d1c8b",
   "metadata": {},
   "source": [
    "## Download the Data\n",
    "\n",
    "For this example we will work with the \"Fashin MNIST\" dataset, composed of thousands of $28 \\times 28$ images."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "id": "d53063b6",
   "metadata": {},
   "outputs": [],
   "source": [
    "fashion_mnist = keras.datasets.fashion_mnist\n",
    "(X_train_full, y_train_full), (X_test, y_test) = fashion_mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "id": "2aac2bd1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(60000, 28, 28)"
      ]
     },
     "execution_count": 162,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_full.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "id": "3baa0642",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 28, 28)"
      ]
     },
     "execution_count": 163,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5f86f2ea",
   "metadata": {},
   "source": [
    "Let's visualize the first image of our training set, so we have a better idea of how the data looks like. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "id": "4f0c3484",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib as mlp\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "sample_image = X_train_full[0]\n",
    "\n",
    "plt.imshow(sample_image, cmap=\"binary\")\n",
    "plt.axis(\"off\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "45f80502",
   "metadata": {},
   "source": [
    "## Define Trainining and Validation Sets\n",
    "\n",
    "Now let's scale the data by dividing by the maximum value (255), so all the values are in the $[0, 1]$ range.\n",
    "\n",
    "We will create a validation set from our training set, composed of the first 5000 thousand images, the rest will be used for training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "id": "537cb86a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# X_train, X_valid, y_train, y_valid = train_test_split(X_train_full, y_train_full)\n",
    "\n",
    "X_valid, X_train = X_train_full[:5000] / 255.0, X_train_full[5000:] / 255.0\n",
    "y_valid, y_train = y_train_full[:5000], y_train_full[5000:]\n",
    "X_test = X_test / 255.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "id": "c41020a1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(55000, 28, 28)\n",
      "(5000, 28, 28)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape)\n",
    "print(X_valid.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dca45535",
   "metadata": {},
   "source": [
    "Let's define the names of our classes/categories."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "id": "e6ce9fcb",
   "metadata": {},
   "outputs": [],
   "source": [
    "class_names = [\"T-shirt/top\", \"Trouser\", \"Pullover\", \"Dress\", \n",
    "               \"Coat\", \"Sandal\", \"Shirt\", \"Sneaker\", \"Bag\", \"Ankle boot\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "id": "569aafcd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Coat'"
      ]
     },
     "execution_count": 168,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class_names[y_train[0]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "id": "24e6585e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([4, 0, 7, 9, 9, 9, 4, 4, 3, 4], dtype=uint8)"
      ]
     },
     "execution_count": 169,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train[0:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "63615d87",
   "metadata": {},
   "source": [
    "## Binary Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "id": "2d2cd406",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 1, 1, ..., 1, 1, 1])"
      ]
     },
     "execution_count": 170,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "y_ankle = y_train_full[0]\n",
    "y_train_bin = np.where(y_train == y_ankle, 0, 1)\n",
    "y_train_bin"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "id": "af8a2010",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_valid_bin = np.where(y_valid == y_ankle, 0, 1)\n",
    "y_test_bin = np.where(y_test == y_ankle, 0, 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4eb175b",
   "metadata": {},
   "source": [
    "## Create the  Model\n",
    "\n",
    "The input layer has 785 units, which represents the 28 x 28 pixels in the image. Let's define two hidden layers, with 300 and 100 neurons, respecitively. The output layer has 10 units, each representing a class; this output layer uses the `softmax`function to use a multinomial approach to handle multiple classes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "id": "610a592e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_3\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input (Flatten)             (None, 784)               0         \n",
      "                                                                 \n",
      " hidden1 (Dense)             (None, 300)               235500    \n",
      "                                                                 \n",
      " hidden2 (Dense)             (None, 100)               30100     \n",
      "                                                                 \n",
      " output (Dense)              (None, 1)                 101       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 265,701\n",
      "Trainable params: 265,701\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = keras.models.Sequential([\n",
    "    keras.layers.Flatten(input_shape=[28, 28], name=\"input\"),\n",
    "    keras.layers.Dense(300, activation=\"sigmoid\", name=\"hidden1\"),\n",
    "    keras.layers.Dense(100, activation=\"sigmoid\", name=\"hidden2\"),\n",
    "    keras.layers.Dense(1, activation=\"sigmoid\", name=\"output\")\n",
    "])\n",
    "model.summary()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2666836d",
   "metadata": {},
   "source": [
    "We can get an array of the layers in our network with the `layers` propery."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "id": "2ef9f71b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<keras.layers.reshaping.flatten.Flatten at 0x2a02b1280>,\n",
       " <keras.layers.core.dense.Dense at 0x2a02b11f0>,\n",
       " <keras.layers.core.dense.Dense at 0x2a3fb24c0>,\n",
       " <keras.layers.core.dense.Dense at 0x2a02b1970>]"
      ]
     },
     "execution_count": 173,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.layers"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0dcb5dd7",
   "metadata": {},
   "source": [
    "We can get an especific layer by index, and access the properties of that layer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "id": "4880756d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'hidden1'"
      ]
     },
     "execution_count": 174,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hidden1 = model.layers[1]\n",
    "hidden1.name"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e3baaf20",
   "metadata": {},
   "source": [
    "We also can get a layer by its name with the `get_layer` method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "id": "4218d259",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 175,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.get_layer(\"hidden1\") is hidden1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "26693938",
   "metadata": {},
   "source": [
    "The `get_weights` method returns the weight parameters and bias terms of a layer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "id": "48f0caf8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.06646881, -0.01292914,  0.04986201, ..., -0.04686245,\n",
       "         0.00351156, -0.02708911],\n",
       "       [-0.05719316,  0.02932094,  0.04294334, ...,  0.03775851,\n",
       "        -0.0699632 ,  0.03168429],\n",
       "       [-0.05547386, -0.0236292 ,  0.06560589, ..., -0.04187467,\n",
       "         0.06099258,  0.02143674],\n",
       "       ...,\n",
       "       [ 0.00425687,  0.05048473,  0.04456843, ..., -0.05111232,\n",
       "         0.02372888,  0.04002428],\n",
       "       [ 0.00693998, -0.00932447,  0.01112469, ...,  0.06034316,\n",
       "        -0.064955  ,  0.0357217 ],\n",
       "       [ 0.03471807,  0.03846084,  0.00690506, ...,  0.00571021,\n",
       "        -0.01373942, -0.0207839 ]], dtype=float32)"
      ]
     },
     "execution_count": 176,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weights, biases = hidden1.get_weights()\n",
    "weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "id": "87aef5ac",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(784, 300)\n",
      "(300,)\n"
     ]
    }
   ],
   "source": [
    "print(weights.shape)\n",
    "print(biases.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79f61f7e",
   "metadata": {},
   "source": [
    "## Compile the Model\n",
    "\n",
    "In this step we specify the loss function, optimizers, and metrics for our network. For this example, we are using the categorical crossentroy loss function, since we are dealing with a multiclass categorical target variable. The `sgd` refers to the \"Stochastic Gradiend Descent\" algorithm, and we will use the accuracy for measuring the performance of our network. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "id": "1ecb200e",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"binary_crossentropy\",\n",
    "              optimizer=\"sgd\",\n",
    "              metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b877884d",
   "metadata": {},
   "source": [
    "## Adding Callbacks\n",
    "\n",
    "Callback are helpful to execute periodic specific tasks, for example, at the end of one epoch. In this case, we are adding an \"Early Stopping\" callback, indicating that if the validation score doesn't improve in 10 epochs, the training will stop and we will get the best solution computed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "id": "a6ab0e72",
   "metadata": {},
   "outputs": [],
   "source": [
    "early_stopping_cb = keras.callbacks.EarlyStopping(patience=10, restore_best_weights=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6c739548",
   "metadata": {},
   "source": [
    "You can also define your own callbacks, here we are defining a callback that prints the ration between the validation score and the training score."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "id": "d5ba10e0",
   "metadata": {},
   "outputs": [],
   "source": [
    "class MyCallback(keras.callbacks.Callback):\n",
    "    def on_epoch_end(self, epoch, logs):\n",
    "        print(f\"\\nval/train: {logs['val_loss'] / logs['loss']}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e6cef0ff",
   "metadata": {},
   "source": [
    "## Training and Evaluating\n",
    "\n",
    "To train the network we need to specify the training and validation sets, as well as the number of epochs, callbacks ,and all the necessary parameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "id": "9409a00b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "1719/1719 [==============================] - 2s 908us/step - loss: 0.2695 - accuracy: 0.8982 - val_loss: 0.1875 - val_accuracy: 0.8988\n",
      "Epoch 2/100\n",
      "1719/1719 [==============================] - 1s 872us/step - loss: 0.1249 - accuracy: 0.9511 - val_loss: 0.0901 - val_accuracy: 0.9696\n",
      "Epoch 3/100\n",
      "1719/1719 [==============================] - 1s 872us/step - loss: 0.0763 - accuracy: 0.9745 - val_loss: 0.0696 - val_accuracy: 0.9722\n",
      "Epoch 4/100\n",
      "1719/1719 [==============================] - 1s 872us/step - loss: 0.0643 - accuracy: 0.9762 - val_loss: 0.0622 - val_accuracy: 0.9736\n",
      "Epoch 5/100\n",
      "1719/1719 [==============================] - 2s 873us/step - loss: 0.0590 - accuracy: 0.9774 - val_loss: 0.0582 - val_accuracy: 0.9748\n",
      "Epoch 6/100\n",
      "1719/1719 [==============================] - 1s 872us/step - loss: 0.0560 - accuracy: 0.9783 - val_loss: 0.0555 - val_accuracy: 0.9760\n",
      "Epoch 7/100\n",
      "1719/1719 [==============================] - 2s 880us/step - loss: 0.0538 - accuracy: 0.9792 - val_loss: 0.0536 - val_accuracy: 0.9770\n",
      "Epoch 8/100\n",
      "1719/1719 [==============================] - 2s 883us/step - loss: 0.0523 - accuracy: 0.9798 - val_loss: 0.0519 - val_accuracy: 0.9784\n",
      "Epoch 9/100\n",
      "1719/1719 [==============================] - 1s 872us/step - loss: 0.0510 - accuracy: 0.9802 - val_loss: 0.0508 - val_accuracy: 0.9786\n",
      "Epoch 10/100\n",
      "1719/1719 [==============================] - 2s 880us/step - loss: 0.0499 - accuracy: 0.9807 - val_loss: 0.0493 - val_accuracy: 0.9796\n",
      "Epoch 11/100\n",
      "1719/1719 [==============================] - 2s 876us/step - loss: 0.0489 - accuracy: 0.9810 - val_loss: 0.0485 - val_accuracy: 0.9790\n",
      "Epoch 12/100\n",
      "1719/1719 [==============================] - 2s 875us/step - loss: 0.0480 - accuracy: 0.9814 - val_loss: 0.0471 - val_accuracy: 0.9806\n",
      "Epoch 13/100\n",
      "1719/1719 [==============================] - 1s 872us/step - loss: 0.0472 - accuracy: 0.9815 - val_loss: 0.0462 - val_accuracy: 0.9804\n",
      "Epoch 14/100\n",
      "1719/1719 [==============================] - 2s 886us/step - loss: 0.0465 - accuracy: 0.9821 - val_loss: 0.0450 - val_accuracy: 0.9818\n",
      "Epoch 15/100\n",
      "1719/1719 [==============================] - 2s 901us/step - loss: 0.0458 - accuracy: 0.9823 - val_loss: 0.0443 - val_accuracy: 0.9816\n",
      "Epoch 16/100\n",
      "1719/1719 [==============================] - 2s 894us/step - loss: 0.0451 - accuracy: 0.9825 - val_loss: 0.0437 - val_accuracy: 0.9814\n",
      "Epoch 17/100\n",
      "1719/1719 [==============================] - 2s 910us/step - loss: 0.0444 - accuracy: 0.9830 - val_loss: 0.0426 - val_accuracy: 0.9832\n",
      "Epoch 18/100\n",
      "1719/1719 [==============================] - 2s 881us/step - loss: 0.0439 - accuracy: 0.9833 - val_loss: 0.0419 - val_accuracy: 0.9834\n",
      "Epoch 19/100\n",
      "1719/1719 [==============================] - 2s 878us/step - loss: 0.0433 - accuracy: 0.9837 - val_loss: 0.0413 - val_accuracy: 0.9832\n",
      "Epoch 20/100\n",
      "1719/1719 [==============================] - 2s 877us/step - loss: 0.0428 - accuracy: 0.9838 - val_loss: 0.0407 - val_accuracy: 0.9838\n",
      "Epoch 21/100\n",
      "1719/1719 [==============================] - 2s 895us/step - loss: 0.0423 - accuracy: 0.9840 - val_loss: 0.0403 - val_accuracy: 0.9838\n",
      "Epoch 22/100\n",
      "1719/1719 [==============================] - 2s 872us/step - loss: 0.0419 - accuracy: 0.9842 - val_loss: 0.0396 - val_accuracy: 0.9846\n",
      "Epoch 23/100\n",
      "1719/1719 [==============================] - 2s 873us/step - loss: 0.0414 - accuracy: 0.9846 - val_loss: 0.0391 - val_accuracy: 0.9850\n",
      "Epoch 24/100\n",
      "1719/1719 [==============================] - 2s 878us/step - loss: 0.0409 - accuracy: 0.9847 - val_loss: 0.0394 - val_accuracy: 0.9834\n",
      "Epoch 25/100\n",
      "1719/1719 [==============================] - 2s 873us/step - loss: 0.0405 - accuracy: 0.9850 - val_loss: 0.0384 - val_accuracy: 0.9838\n",
      "Epoch 26/100\n",
      "1719/1719 [==============================] - 1s 871us/step - loss: 0.0402 - accuracy: 0.9848 - val_loss: 0.0378 - val_accuracy: 0.9852\n",
      "Epoch 27/100\n",
      "1719/1719 [==============================] - 2s 878us/step - loss: 0.0398 - accuracy: 0.9851 - val_loss: 0.0377 - val_accuracy: 0.9840\n",
      "Epoch 28/100\n",
      "1719/1719 [==============================] - 2s 878us/step - loss: 0.0395 - accuracy: 0.9854 - val_loss: 0.0371 - val_accuracy: 0.9860\n",
      "Epoch 29/100\n",
      "1719/1719 [==============================] - 2s 875us/step - loss: 0.0392 - accuracy: 0.9854 - val_loss: 0.0368 - val_accuracy: 0.9856\n",
      "Epoch 30/100\n",
      "1719/1719 [==============================] - 2s 880us/step - loss: 0.0388 - accuracy: 0.9857 - val_loss: 0.0367 - val_accuracy: 0.9856\n",
      "Epoch 31/100\n",
      "1719/1719 [==============================] - 2s 874us/step - loss: 0.0385 - accuracy: 0.9858 - val_loss: 0.0363 - val_accuracy: 0.9860\n",
      "Epoch 32/100\n",
      "1719/1719 [==============================] - 1s 872us/step - loss: 0.0382 - accuracy: 0.9856 - val_loss: 0.0361 - val_accuracy: 0.9858\n",
      "Epoch 33/100\n",
      "1719/1719 [==============================] - 2s 875us/step - loss: 0.0380 - accuracy: 0.9859 - val_loss: 0.0358 - val_accuracy: 0.9862\n",
      "Epoch 34/100\n",
      "1719/1719 [==============================] - 2s 876us/step - loss: 0.0377 - accuracy: 0.9857 - val_loss: 0.0357 - val_accuracy: 0.9866\n",
      "Epoch 35/100\n",
      "1719/1719 [==============================] - 2s 877us/step - loss: 0.0374 - accuracy: 0.9863 - val_loss: 0.0352 - val_accuracy: 0.9866\n",
      "Epoch 36/100\n",
      "1719/1719 [==============================] - 2s 873us/step - loss: 0.0372 - accuracy: 0.9865 - val_loss: 0.0354 - val_accuracy: 0.9858\n",
      "Epoch 37/100\n",
      "1719/1719 [==============================] - 1s 869us/step - loss: 0.0370 - accuracy: 0.9865 - val_loss: 0.0349 - val_accuracy: 0.9862\n",
      "Epoch 38/100\n",
      "1719/1719 [==============================] - 1s 859us/step - loss: 0.0367 - accuracy: 0.9866 - val_loss: 0.0346 - val_accuracy: 0.9872\n",
      "Epoch 39/100\n",
      "1719/1719 [==============================] - 1s 859us/step - loss: 0.0365 - accuracy: 0.9867 - val_loss: 0.0346 - val_accuracy: 0.9862\n",
      "Epoch 40/100\n",
      "1719/1719 [==============================] - 1s 855us/step - loss: 0.0363 - accuracy: 0.9865 - val_loss: 0.0343 - val_accuracy: 0.9870\n",
      "Epoch 41/100\n",
      "1719/1719 [==============================] - 1s 851us/step - loss: 0.0360 - accuracy: 0.9867 - val_loss: 0.0346 - val_accuracy: 0.9864\n",
      "Epoch 42/100\n",
      "1719/1719 [==============================] - 1s 840us/step - loss: 0.0359 - accuracy: 0.9868 - val_loss: 0.0340 - val_accuracy: 0.9874\n",
      "Epoch 43/100\n",
      "1719/1719 [==============================] - 1s 841us/step - loss: 0.0356 - accuracy: 0.9870 - val_loss: 0.0342 - val_accuracy: 0.9870\n",
      "Epoch 44/100\n",
      "1719/1719 [==============================] - 1s 841us/step - loss: 0.0355 - accuracy: 0.9872 - val_loss: 0.0338 - val_accuracy: 0.9868\n",
      "Epoch 45/100\n",
      "1719/1719 [==============================] - 1s 853us/step - loss: 0.0353 - accuracy: 0.9872 - val_loss: 0.0344 - val_accuracy: 0.9866\n",
      "Epoch 46/100\n",
      "1719/1719 [==============================] - 1s 849us/step - loss: 0.0351 - accuracy: 0.9870 - val_loss: 0.0334 - val_accuracy: 0.9874\n",
      "Epoch 47/100\n",
      "1719/1719 [==============================] - 1s 859us/step - loss: 0.0349 - accuracy: 0.9871 - val_loss: 0.0333 - val_accuracy: 0.9878\n",
      "Epoch 48/100\n",
      "1719/1719 [==============================] - 1s 856us/step - loss: 0.0348 - accuracy: 0.9872 - val_loss: 0.0334 - val_accuracy: 0.9868\n",
      "Epoch 49/100\n",
      "1719/1719 [==============================] - 1s 868us/step - loss: 0.0346 - accuracy: 0.9872 - val_loss: 0.0330 - val_accuracy: 0.9880\n",
      "Epoch 50/100\n",
      "1719/1719 [==============================] - 1s 869us/step - loss: 0.0344 - accuracy: 0.9874 - val_loss: 0.0335 - val_accuracy: 0.9870\n",
      "Epoch 51/100\n",
      "1719/1719 [==============================] - 1s 865us/step - loss: 0.0342 - accuracy: 0.9876 - val_loss: 0.0328 - val_accuracy: 0.9880\n",
      "Epoch 52/100\n",
      "1719/1719 [==============================] - 1s 871us/step - loss: 0.0341 - accuracy: 0.9875 - val_loss: 0.0329 - val_accuracy: 0.9876\n",
      "Epoch 53/100\n",
      "1719/1719 [==============================] - 1s 864us/step - loss: 0.0339 - accuracy: 0.9874 - val_loss: 0.0330 - val_accuracy: 0.9876\n",
      "Epoch 54/100\n",
      "1719/1719 [==============================] - 2s 879us/step - loss: 0.0337 - accuracy: 0.9873 - val_loss: 0.0328 - val_accuracy: 0.9878\n",
      "Epoch 55/100\n",
      "1719/1719 [==============================] - 2s 874us/step - loss: 0.0336 - accuracy: 0.9875 - val_loss: 0.0327 - val_accuracy: 0.9878\n",
      "Epoch 56/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1719/1719 [==============================] - 1s 866us/step - loss: 0.0334 - accuracy: 0.9877 - val_loss: 0.0325 - val_accuracy: 0.9878\n",
      "Epoch 57/100\n",
      "1719/1719 [==============================] - 1s 866us/step - loss: 0.0333 - accuracy: 0.9878 - val_loss: 0.0330 - val_accuracy: 0.9876\n",
      "Epoch 58/100\n",
      "1719/1719 [==============================] - 1s 867us/step - loss: 0.0332 - accuracy: 0.9877 - val_loss: 0.0327 - val_accuracy: 0.9876\n",
      "Epoch 59/100\n",
      "1719/1719 [==============================] - 1s 868us/step - loss: 0.0329 - accuracy: 0.9883 - val_loss: 0.0337 - val_accuracy: 0.9864\n",
      "Epoch 60/100\n",
      "1719/1719 [==============================] - 1s 865us/step - loss: 0.0330 - accuracy: 0.9880 - val_loss: 0.0322 - val_accuracy: 0.9880\n",
      "Epoch 61/100\n",
      "1719/1719 [==============================] - 1s 868us/step - loss: 0.0328 - accuracy: 0.9880 - val_loss: 0.0323 - val_accuracy: 0.9874\n",
      "Epoch 62/100\n",
      "1719/1719 [==============================] - 2s 881us/step - loss: 0.0327 - accuracy: 0.9879 - val_loss: 0.0322 - val_accuracy: 0.9876\n",
      "Epoch 63/100\n",
      "1719/1719 [==============================] - 1s 869us/step - loss: 0.0326 - accuracy: 0.9880 - val_loss: 0.0323 - val_accuracy: 0.9878\n",
      "Epoch 64/100\n",
      "1719/1719 [==============================] - 1s 869us/step - loss: 0.0325 - accuracy: 0.9879 - val_loss: 0.0319 - val_accuracy: 0.9878\n",
      "Epoch 65/100\n",
      "1719/1719 [==============================] - 1s 868us/step - loss: 0.0323 - accuracy: 0.9880 - val_loss: 0.0318 - val_accuracy: 0.9884\n",
      "Epoch 66/100\n",
      "1719/1719 [==============================] - 1s 869us/step - loss: 0.0322 - accuracy: 0.9881 - val_loss: 0.0319 - val_accuracy: 0.9882\n",
      "Epoch 67/100\n",
      "1719/1719 [==============================] - 1s 868us/step - loss: 0.0321 - accuracy: 0.9880 - val_loss: 0.0323 - val_accuracy: 0.9872\n",
      "Epoch 68/100\n",
      "1719/1719 [==============================] - 1s 867us/step - loss: 0.0319 - accuracy: 0.9885 - val_loss: 0.0320 - val_accuracy: 0.9876\n",
      "Epoch 69/100\n",
      "1719/1719 [==============================] - 1s 869us/step - loss: 0.0318 - accuracy: 0.9882 - val_loss: 0.0326 - val_accuracy: 0.9876\n",
      "Epoch 70/100\n",
      "1719/1719 [==============================] - 1s 867us/step - loss: 0.0318 - accuracy: 0.9884 - val_loss: 0.0317 - val_accuracy: 0.9884\n",
      "Epoch 71/100\n",
      "1719/1719 [==============================] - 1s 868us/step - loss: 0.0316 - accuracy: 0.9883 - val_loss: 0.0321 - val_accuracy: 0.9872\n",
      "Epoch 72/100\n",
      "1719/1719 [==============================] - 2s 894us/step - loss: 0.0315 - accuracy: 0.9884 - val_loss: 0.0315 - val_accuracy: 0.9882\n",
      "Epoch 73/100\n",
      "1719/1719 [==============================] - 2s 879us/step - loss: 0.0315 - accuracy: 0.9883 - val_loss: 0.0314 - val_accuracy: 0.9882\n",
      "Epoch 74/100\n",
      "1719/1719 [==============================] - 1s 871us/step - loss: 0.0313 - accuracy: 0.9886 - val_loss: 0.0314 - val_accuracy: 0.9882\n",
      "Epoch 75/100\n",
      "1719/1719 [==============================] - 1s 868us/step - loss: 0.0313 - accuracy: 0.9885 - val_loss: 0.0313 - val_accuracy: 0.9882\n",
      "Epoch 76/100\n",
      "1719/1719 [==============================] - 1s 868us/step - loss: 0.0312 - accuracy: 0.9884 - val_loss: 0.0313 - val_accuracy: 0.9882\n",
      "Epoch 77/100\n",
      "1719/1719 [==============================] - 1s 869us/step - loss: 0.0311 - accuracy: 0.9887 - val_loss: 0.0319 - val_accuracy: 0.9874\n",
      "Epoch 78/100\n",
      "1719/1719 [==============================] - 1s 868us/step - loss: 0.0310 - accuracy: 0.9887 - val_loss: 0.0319 - val_accuracy: 0.9876\n",
      "Epoch 79/100\n",
      "1719/1719 [==============================] - 2s 875us/step - loss: 0.0310 - accuracy: 0.9884 - val_loss: 0.0319 - val_accuracy: 0.9876\n",
      "Epoch 80/100\n",
      "1719/1719 [==============================] - 2s 929us/step - loss: 0.0308 - accuracy: 0.9887 - val_loss: 0.0314 - val_accuracy: 0.9878\n",
      "Epoch 81/100\n",
      "1719/1719 [==============================] - 2s 893us/step - loss: 0.0307 - accuracy: 0.9887 - val_loss: 0.0312 - val_accuracy: 0.9884\n",
      "Epoch 82/100\n",
      "1719/1719 [==============================] - 2s 900us/step - loss: 0.0306 - accuracy: 0.9886 - val_loss: 0.0315 - val_accuracy: 0.9876\n",
      "Epoch 83/100\n",
      "1719/1719 [==============================] - 2s 887us/step - loss: 0.0305 - accuracy: 0.9887 - val_loss: 0.0312 - val_accuracy: 0.9886\n",
      "Epoch 84/100\n",
      "1719/1719 [==============================] - 2s 882us/step - loss: 0.0305 - accuracy: 0.9886 - val_loss: 0.0312 - val_accuracy: 0.9882\n",
      "Epoch 85/100\n",
      "1719/1719 [==============================] - 2s 891us/step - loss: 0.0304 - accuracy: 0.9887 - val_loss: 0.0310 - val_accuracy: 0.9886\n",
      "Epoch 86/100\n",
      "1719/1719 [==============================] - 2s 891us/step - loss: 0.0303 - accuracy: 0.9887 - val_loss: 0.0310 - val_accuracy: 0.9884\n",
      "Epoch 87/100\n",
      "1719/1719 [==============================] - 2s 896us/step - loss: 0.0302 - accuracy: 0.9887 - val_loss: 0.0309 - val_accuracy: 0.9884\n",
      "Epoch 88/100\n",
      "1719/1719 [==============================] - 2s 898us/step - loss: 0.0301 - accuracy: 0.9889 - val_loss: 0.0310 - val_accuracy: 0.9886\n",
      "Epoch 89/100\n",
      "1719/1719 [==============================] - 2s 911us/step - loss: 0.0301 - accuracy: 0.9889 - val_loss: 0.0315 - val_accuracy: 0.9874\n",
      "Epoch 90/100\n",
      "1719/1719 [==============================] - 2s 890us/step - loss: 0.0299 - accuracy: 0.9891 - val_loss: 0.0311 - val_accuracy: 0.9878\n",
      "Epoch 91/100\n",
      "1719/1719 [==============================] - 2s 894us/step - loss: 0.0299 - accuracy: 0.9890 - val_loss: 0.0308 - val_accuracy: 0.9886\n",
      "Epoch 92/100\n",
      "1719/1719 [==============================] - 2s 883us/step - loss: 0.0298 - accuracy: 0.9889 - val_loss: 0.0307 - val_accuracy: 0.9882\n",
      "Epoch 93/100\n",
      "1719/1719 [==============================] - 2s 878us/step - loss: 0.0298 - accuracy: 0.9890 - val_loss: 0.0310 - val_accuracy: 0.9880\n",
      "Epoch 94/100\n",
      "1719/1719 [==============================] - 2s 903us/step - loss: 0.0296 - accuracy: 0.9894 - val_loss: 0.0307 - val_accuracy: 0.9886\n",
      "Epoch 95/100\n",
      "1719/1719 [==============================] - 2s 904us/step - loss: 0.0296 - accuracy: 0.9891 - val_loss: 0.0310 - val_accuracy: 0.9888\n",
      "Epoch 96/100\n",
      "1719/1719 [==============================] - 2s 886us/step - loss: 0.0295 - accuracy: 0.9890 - val_loss: 0.0309 - val_accuracy: 0.9882\n",
      "Epoch 97/100\n",
      "1719/1719 [==============================] - 2s 879us/step - loss: 0.0294 - accuracy: 0.9891 - val_loss: 0.0312 - val_accuracy: 0.9888\n",
      "Epoch 98/100\n",
      "1719/1719 [==============================] - 2s 883us/step - loss: 0.0293 - accuracy: 0.9893 - val_loss: 0.0314 - val_accuracy: 0.9874\n",
      "Epoch 99/100\n",
      "1719/1719 [==============================] - 2s 875us/step - loss: 0.0293 - accuracy: 0.9893 - val_loss: 0.0313 - val_accuracy: 0.9874\n",
      "Epoch 100/100\n",
      "1719/1719 [==============================] - 2s 900us/step - loss: 0.0293 - accuracy: 0.9891 - val_loss: 0.0307 - val_accuracy: 0.9880\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X_train, \n",
    "                    y_train_bin, \n",
    "                    epochs=100, \n",
    "                    validation_data=(X_valid, y_valid_bin),\n",
    "                    callbacks=[early_stopping_cb]\n",
    "                   )"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b537a308",
   "metadata": {},
   "source": [
    "With the `epoch` property we get a list with all the epochs that were completed. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "id": "ed64ae54",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "100"
      ]
     },
     "execution_count": 182,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(history.epoch)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4472128b",
   "metadata": {},
   "source": [
    "The `steps` parameter refers to the number number of steps to complete an epoch, `batch_size * steps = n`. By default the batch size is 32."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "id": "acb39dbf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'verbose': 1, 'epochs': 100, 'steps': 1719}"
      ]
     },
     "execution_count": 183,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history.params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "id": "ca7d36ad",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'loss': [0.2695257067680359,\n",
       "  0.1248958483338356,\n",
       "  0.07634817808866501,\n",
       "  0.06427951902151108,\n",
       "  0.05904070660471916,\n",
       "  0.05596290901303291,\n",
       "  0.053811512887477875,\n",
       "  0.05229172855615616,\n",
       "  0.05100465938448906,\n",
       "  0.04994119703769684,\n",
       "  0.04892474412918091,\n",
       "  0.04803336411714554,\n",
       "  0.04724544659256935,\n",
       "  0.04647666960954666,\n",
       "  0.04576048627495766,\n",
       "  0.04510057717561722,\n",
       "  0.04441598057746887,\n",
       "  0.04387957230210304,\n",
       "  0.043290115892887115,\n",
       "  0.042792901396751404,\n",
       "  0.042272184044122696,\n",
       "  0.041858453303575516,\n",
       "  0.041392870247364044,\n",
       "  0.04090234264731407,\n",
       "  0.040541451424360275,\n",
       "  0.040205348283052444,\n",
       "  0.039828650653362274,\n",
       "  0.03950323909521103,\n",
       "  0.03917328640818596,\n",
       "  0.03882341831922531,\n",
       "  0.03853675723075867,\n",
       "  0.03822055086493492,\n",
       "  0.038023605942726135,\n",
       "  0.037690967321395874,\n",
       "  0.03744324669241905,\n",
       "  0.03720756620168686,\n",
       "  0.03695417195558548,\n",
       "  0.0366806834936142,\n",
       "  0.03647635877132416,\n",
       "  0.036282915621995926,\n",
       "  0.03596533089876175,\n",
       "  0.035904910415410995,\n",
       "  0.035638827830553055,\n",
       "  0.03546031937003136,\n",
       "  0.03526468575000763,\n",
       "  0.03505067527294159,\n",
       "  0.03493110090494156,\n",
       "  0.03477976843714714,\n",
       "  0.034612592309713364,\n",
       "  0.034381378442049026,\n",
       "  0.034175533801317215,\n",
       "  0.034062620252370834,\n",
       "  0.03394034877419472,\n",
       "  0.0337410494685173,\n",
       "  0.03363463282585144,\n",
       "  0.03343001753091812,\n",
       "  0.03331517055630684,\n",
       "  0.03319168463349342,\n",
       "  0.032946426421403885,\n",
       "  0.0329732745885849,\n",
       "  0.03279561176896095,\n",
       "  0.03272617980837822,\n",
       "  0.032585978507995605,\n",
       "  0.03247179463505745,\n",
       "  0.03228034824132919,\n",
       "  0.032238204032182693,\n",
       "  0.03208518773317337,\n",
       "  0.031911615282297134,\n",
       "  0.031830083578825,\n",
       "  0.031760092824697495,\n",
       "  0.031630951911211014,\n",
       "  0.03153686597943306,\n",
       "  0.03148450329899788,\n",
       "  0.03133995831012726,\n",
       "  0.031262535601854324,\n",
       "  0.031153613701462746,\n",
       "  0.031128915026783943,\n",
       "  0.030976733192801476,\n",
       "  0.03095061518251896,\n",
       "  0.030770018696784973,\n",
       "  0.030730461701750755,\n",
       "  0.03061543218791485,\n",
       "  0.030546464025974274,\n",
       "  0.030500788241624832,\n",
       "  0.03036491945385933,\n",
       "  0.030313076451420784,\n",
       "  0.030222691595554352,\n",
       "  0.03014216385781765,\n",
       "  0.03007403202354908,\n",
       "  0.029946012422442436,\n",
       "  0.029904894530773163,\n",
       "  0.029788577929139137,\n",
       "  0.02977198176085949,\n",
       "  0.02959716133773327,\n",
       "  0.029586568474769592,\n",
       "  0.02949339523911476,\n",
       "  0.02941412664949894,\n",
       "  0.02926352247595787,\n",
       "  0.029307816177606583,\n",
       "  0.029290560632944107],\n",
       " 'accuracy': [0.8981636166572571,\n",
       "  0.951090931892395,\n",
       "  0.97447270154953,\n",
       "  0.9761818051338196,\n",
       "  0.9774181842803955,\n",
       "  0.9783272743225098,\n",
       "  0.9791818261146545,\n",
       "  0.9797818064689636,\n",
       "  0.9802181720733643,\n",
       "  0.9807090759277344,\n",
       "  0.9810181856155396,\n",
       "  0.9814363718032837,\n",
       "  0.9814545512199402,\n",
       "  0.9820908904075623,\n",
       "  0.982272744178772,\n",
       "  0.9824908971786499,\n",
       "  0.983018159866333,\n",
       "  0.9833454489707947,\n",
       "  0.9836909174919128,\n",
       "  0.9837818145751953,\n",
       "  0.9839818477630615,\n",
       "  0.9842000007629395,\n",
       "  0.9846363663673401,\n",
       "  0.9846545457839966,\n",
       "  0.9850181937217712,\n",
       "  0.9848363399505615,\n",
       "  0.9851454496383667,\n",
       "  0.9853818416595459,\n",
       "  0.9853636622428894,\n",
       "  0.9856727123260498,\n",
       "  0.985836386680603,\n",
       "  0.9855999946594238,\n",
       "  0.985945463180542,\n",
       "  0.9857272505760193,\n",
       "  0.9862909317016602,\n",
       "  0.9865090847015381,\n",
       "  0.9864727258682251,\n",
       "  0.986618161201477,\n",
       "  0.9866909384727478,\n",
       "  0.9865454435348511,\n",
       "  0.9867454767227173,\n",
       "  0.9868181943893433,\n",
       "  0.9870363473892212,\n",
       "  0.9872182011604309,\n",
       "  0.9871636629104614,\n",
       "  0.9869999885559082,\n",
       "  0.9871454834938049,\n",
       "  0.9871818423271179,\n",
       "  0.9872363805770874,\n",
       "  0.9873999953269958,\n",
       "  0.9876000285148621,\n",
       "  0.9874545335769653,\n",
       "  0.9874363541603088,\n",
       "  0.9873454570770264,\n",
       "  0.9875272512435913,\n",
       "  0.987672746181488,\n",
       "  0.987781822681427,\n",
       "  0.987745463848114,\n",
       "  0.9882727265357971,\n",
       "  0.9879817962646484,\n",
       "  0.9879817962646484,\n",
       "  0.9878727197647095,\n",
       "  0.9879636168479919,\n",
       "  0.987854540348053,\n",
       "  0.9880181550979614,\n",
       "  0.9881272912025452,\n",
       "  0.9880181550979614,\n",
       "  0.9884545207023621,\n",
       "  0.9881818294525146,\n",
       "  0.9883636236190796,\n",
       "  0.9882545471191406,\n",
       "  0.9883636236190796,\n",
       "  0.9883454442024231,\n",
       "  0.9885818362236023,\n",
       "  0.9885091185569763,\n",
       "  0.9884363412857056,\n",
       "  0.9886909127235413,\n",
       "  0.9887090921401978,\n",
       "  0.9883999824523926,\n",
       "  0.9887454509735107,\n",
       "  0.9887454509735107,\n",
       "  0.9886000156402588,\n",
       "  0.9887090921401978,\n",
       "  0.9885818362236023,\n",
       "  0.9886727333068848,\n",
       "  0.9886909127235413,\n",
       "  0.9886909127235413,\n",
       "  0.9888727068901062,\n",
       "  0.9888727068901062,\n",
       "  0.9890545606613159,\n",
       "  0.9890000224113464,\n",
       "  0.988945484161377,\n",
       "  0.9889636635780334,\n",
       "  0.9894182085990906,\n",
       "  0.9891454577445984,\n",
       "  0.9890363812446594,\n",
       "  0.9890909194946289,\n",
       "  0.9892727136611938,\n",
       "  0.9892908930778503,\n",
       "  0.9891454577445984],\n",
       " 'val_loss': [0.18747682869434357,\n",
       "  0.09005577117204666,\n",
       "  0.0696358010172844,\n",
       "  0.062248002737760544,\n",
       "  0.05819746106863022,\n",
       "  0.055548712611198425,\n",
       "  0.05359289050102234,\n",
       "  0.05193748697638512,\n",
       "  0.05075442045927048,\n",
       "  0.049267709255218506,\n",
       "  0.04847145825624466,\n",
       "  0.04712915048003197,\n",
       "  0.0462062805891037,\n",
       "  0.04502633959054947,\n",
       "  0.04431537911295891,\n",
       "  0.04374188929796219,\n",
       "  0.042614128440618515,\n",
       "  0.04194486141204834,\n",
       "  0.041313719004392624,\n",
       "  0.04065830633044243,\n",
       "  0.040269847959280014,\n",
       "  0.039564963430166245,\n",
       "  0.03907685726881027,\n",
       "  0.039414845407009125,\n",
       "  0.03837061673402786,\n",
       "  0.03780388459563255,\n",
       "  0.03772414103150368,\n",
       "  0.03707388415932655,\n",
       "  0.03681352362036705,\n",
       "  0.03666991740465164,\n",
       "  0.03631455823779106,\n",
       "  0.03611858934164047,\n",
       "  0.03579263016581535,\n",
       "  0.0356891006231308,\n",
       "  0.03519444912672043,\n",
       "  0.035372983664274216,\n",
       "  0.034896090626716614,\n",
       "  0.034599464386701584,\n",
       "  0.03458399698138237,\n",
       "  0.034308213740587234,\n",
       "  0.03458008915185928,\n",
       "  0.033957112580537796,\n",
       "  0.03420272469520569,\n",
       "  0.03377291560173035,\n",
       "  0.03435410186648369,\n",
       "  0.03341996297240257,\n",
       "  0.033254146575927734,\n",
       "  0.033420708030462265,\n",
       "  0.033042460680007935,\n",
       "  0.0335419699549675,\n",
       "  0.03284173458814621,\n",
       "  0.0328514538705349,\n",
       "  0.03296709433197975,\n",
       "  0.03275001421570778,\n",
       "  0.032731663435697556,\n",
       "  0.03245707228779793,\n",
       "  0.033018819987773895,\n",
       "  0.03273606672883034,\n",
       "  0.033668402582407,\n",
       "  0.032233260571956635,\n",
       "  0.032280176877975464,\n",
       "  0.03223203867673874,\n",
       "  0.03234684467315674,\n",
       "  0.03190525993704796,\n",
       "  0.0318441241979599,\n",
       "  0.03193214163184166,\n",
       "  0.032328538596630096,\n",
       "  0.03200498968362808,\n",
       "  0.032556310296058655,\n",
       "  0.03165417164564133,\n",
       "  0.03209212049841881,\n",
       "  0.03146107494831085,\n",
       "  0.031411461532115936,\n",
       "  0.03139358013868332,\n",
       "  0.03132281452417374,\n",
       "  0.0312621183693409,\n",
       "  0.031943224370479584,\n",
       "  0.031931787729263306,\n",
       "  0.03194207325577736,\n",
       "  0.031448230147361755,\n",
       "  0.03116503916680813,\n",
       "  0.031491369009017944,\n",
       "  0.03115103207528591,\n",
       "  0.031152738258242607,\n",
       "  0.030970245599746704,\n",
       "  0.030981088057160378,\n",
       "  0.030897356569767,\n",
       "  0.031019818037748337,\n",
       "  0.03152761608362198,\n",
       "  0.031107159331440926,\n",
       "  0.030811157077550888,\n",
       "  0.03073747642338276,\n",
       "  0.030964169651269913,\n",
       "  0.030660754069685936,\n",
       "  0.03104613907635212,\n",
       "  0.030879922211170197,\n",
       "  0.031208841130137444,\n",
       "  0.03143481910228729,\n",
       "  0.03131946921348572,\n",
       "  0.03074372559785843],\n",
       " 'val_accuracy': [0.8988000154495239,\n",
       "  0.9696000218391418,\n",
       "  0.9721999764442444,\n",
       "  0.9735999703407288,\n",
       "  0.9747999906539917,\n",
       "  0.9760000109672546,\n",
       "  0.9769999980926514,\n",
       "  0.9783999919891357,\n",
       "  0.978600025177002,\n",
       "  0.9796000123023987,\n",
       "  0.9789999723434448,\n",
       "  0.9805999994277954,\n",
       "  0.980400025844574,\n",
       "  0.9818000197410583,\n",
       "  0.9815999865531921,\n",
       "  0.9814000129699707,\n",
       "  0.9832000136375427,\n",
       "  0.9833999872207642,\n",
       "  0.9832000136375427,\n",
       "  0.9837999939918518,\n",
       "  0.9837999939918518,\n",
       "  0.9846000075340271,\n",
       "  0.9850000143051147,\n",
       "  0.9833999872207642,\n",
       "  0.9837999939918518,\n",
       "  0.9851999878883362,\n",
       "  0.984000027179718,\n",
       "  0.9860000014305115,\n",
       "  0.9855999946594238,\n",
       "  0.9855999946594238,\n",
       "  0.9860000014305115,\n",
       "  0.98580002784729,\n",
       "  0.9861999750137329,\n",
       "  0.9865999817848206,\n",
       "  0.9865999817848206,\n",
       "  0.98580002784729,\n",
       "  0.9861999750137329,\n",
       "  0.9872000217437744,\n",
       "  0.9861999750137329,\n",
       "  0.9869999885559082,\n",
       "  0.9864000082015991,\n",
       "  0.9873999953269958,\n",
       "  0.9869999885559082,\n",
       "  0.9868000149726868,\n",
       "  0.9865999817848206,\n",
       "  0.9873999953269958,\n",
       "  0.9878000020980835,\n",
       "  0.9868000149726868,\n",
       "  0.9879999756813049,\n",
       "  0.9869999885559082,\n",
       "  0.9879999756813049,\n",
       "  0.9876000285148621,\n",
       "  0.9876000285148621,\n",
       "  0.9878000020980835,\n",
       "  0.9878000020980835,\n",
       "  0.9878000020980835,\n",
       "  0.9876000285148621,\n",
       "  0.9876000285148621,\n",
       "  0.9864000082015991,\n",
       "  0.9879999756813049,\n",
       "  0.9873999953269958,\n",
       "  0.9876000285148621,\n",
       "  0.9878000020980835,\n",
       "  0.9878000020980835,\n",
       "  0.9883999824523926,\n",
       "  0.9882000088691711,\n",
       "  0.9872000217437744,\n",
       "  0.9876000285148621,\n",
       "  0.9876000285148621,\n",
       "  0.9883999824523926,\n",
       "  0.9872000217437744,\n",
       "  0.9882000088691711,\n",
       "  0.9882000088691711,\n",
       "  0.9882000088691711,\n",
       "  0.9882000088691711,\n",
       "  0.9882000088691711,\n",
       "  0.9873999953269958,\n",
       "  0.9876000285148621,\n",
       "  0.9876000285148621,\n",
       "  0.9878000020980835,\n",
       "  0.9883999824523926,\n",
       "  0.9876000285148621,\n",
       "  0.9886000156402588,\n",
       "  0.9882000088691711,\n",
       "  0.9886000156402588,\n",
       "  0.9883999824523926,\n",
       "  0.9883999824523926,\n",
       "  0.9886000156402588,\n",
       "  0.9873999953269958,\n",
       "  0.9878000020980835,\n",
       "  0.9886000156402588,\n",
       "  0.9882000088691711,\n",
       "  0.9879999756813049,\n",
       "  0.9886000156402588,\n",
       "  0.9887999892234802,\n",
       "  0.9882000088691711,\n",
       "  0.9887999892234802,\n",
       "  0.9873999953269958,\n",
       "  0.9873999953269958,\n",
       "  0.9879999756813049]}"
      ]
     },
     "execution_count": 184,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history.history"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9b7dd3ba",
   "metadata": {},
   "source": [
    "We can use the `history` property to obtain the score and loss of the training and validation sets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "id": "c9672ea3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function matplotlib.pyplot.show(close=None, block=None)>"
      ]
     },
     "execution_count": 185,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "pd.DataFrame(history.history).plot(figsize=(8, 5))\n",
    "plt.grid(True)\n",
    "plt.gca().set_ylim(0, 1)\n",
    "plt.show"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6edb15f7",
   "metadata": {},
   "source": [
    "We use the `evaluate` method to measure the performance of a given data set. Below we display the loss and accuracy of the test set.\n",
    "\n",
    "Both, the accuracy and the loss are similar to the validation set, which is a sign that we aren't overfitting the training data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "id": "8f8ad911",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 0s 535us/step - loss: 0.0318 - accuracy: 0.9881\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.031798988580703735, 0.988099992275238]"
      ]
     },
     "execution_count": 186,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(X_test, y_test_bin)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "id": "4c7f21a3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9"
      ]
     },
     "execution_count": 187,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_ankle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "id": "f414f193",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([9, 2, 1, 1, 6, 1, 4, 6, 5, 7], dtype=uint8)"
      ]
     },
     "execution_count": 188,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "id": "c79d0d8d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 1, 1, 1, 1, 1, 1, 1, 1])"
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test_bin[:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "462e5156",
   "metadata": {},
   "source": [
    "The `predict` method returns the output of a forward step through the network with the given input. For this example, since we are using the `softmax` activation function; we get the probabilities to belong to each class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "id": "40eb816e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 19ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[0.33],\n",
       "       [1.  ],\n",
       "       [1.  ],\n",
       "       [1.  ],\n",
       "       [1.  ],\n",
       "       [1.  ],\n",
       "       [1.  ],\n",
       "       [1.  ],\n",
       "       [1.  ],\n",
       "       [1.  ]], dtype=float32)"
      ]
     },
     "execution_count": 199,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_new = X_test[:10]\n",
    "y_proba = model.predict(X_new)\n",
    "y_proba.round(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "id": "668aa2a1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAGFCAYAAAASI+9IAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8o6BhiAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAKqklEQVR4nO3cu47WhRbG4TXDYdAR5JBIEI02JBTGaOMV2Ghh4o3YehNegI22NoYetfNIrIynIIVi4giOGJgTjDLMbnbelllrb2dP3M9T8843fHwzP/7NWtjd3d0tAKiqxf/1NwDAwSEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAEAc/l9/A8DBsrOz094sLvb/f7mwsNDeTG1vb7c3S0tL7c21a9fam6qqCxcujHZ/B08KAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQrqfwj7e7u7stmch30l19+aW+qqj7//PP25pVXXmlvlpeX25uDbnLxdOLSpUuj3Ztvvvlf/k7mPCkAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhIN48G+T43YTH3/88Wh35cqV9mZlZaW9eeONN9qbg+63335rby5fvtzeHD9+vL05aDwpABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAISDePwj7ezstDeHD/d/HL788sv25vvvv29vqqrOnj3b3ly7dq29ef3119ubU6dOtTf37t1rb6qqnnnmmfbm1q1b7c3a2lp7c/78+fbmoPGkAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABAO4nHgPXjwoL2ZHLfb3Nxsb95///32Zmlpqb2pmh2QW19fb292d3cP7Kaq6ttvv21vnnrqqfZmcuRvcojxoPGkAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAEC4kroPJtcgFxYWRq81uSg6ea3JZnpB8tChQ6Nd19tvv93enD17tr05duxYe1NVdf369fZmcll18ne6f/9+ezP9jC8vL7c3k8u0d+7caW+2t7fbm6rZhd7J+7AXnhQACFEAIEQBgBAFAEIUAAhRACBEAYAQBQBCFAAIUQAgRAGAEAUA4v/6IN5+HaqbHv6aWFzcn85Pjtvt12G7qqr33nuvvblx40Z78+KLL7Y3k+NxVVW3b99ub06fPt3enDlzpr35/fff25uNjY32pmr+/nVNfj9sbW2NXuvatWvtzQsvvDB6rYfxpABAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQ/9cH8fbrUN2DBw/2ZVM1Ozo3eR/287jdu+++29788MMP7c3TTz/d3ty6dau9mRxaq6q6e/due3P+/Pn2Zn19vb2ZfIYeffTR9qaq6t69e+3Nfh2/nLp8+XJ74yAeAH87UQAgRAGAEAUAQhQACFEAIEQBgBAFAEIUAAhRACBEAYAQBQDiwB3Emx6Cm5gcvJoc1lpc7Ld3stlPKysr7c2lS5dGrzU5BHfhwoX2ZmNjo73Z3t5ubyZH9Kqqjhw50t5MPuNbW1vtzcT0M760tLQvr7W8vNzeTI/offrpp6Pd3+Fg/+YBYF+JAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABB7Poi3s7PT/uKHDh1qbw76Ibjpwauu1dXV0e6nn35qb65evdre/Prrr+3N0aNH25uqqhMnTrQ3t2/fbm/W1tbam7/++qu9mRzRq5r9PE0+D/fv329vTp482d5MPw+T30WTQ5aPPPJIezP53qqqHnvssfbmm2++aW+ee+65h/6Zg/0bGIB9JQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAsecrqZMLjRM3b94c7a5fv97ebG5u7svm7t277c2PP/7Y3lRVbW1ttTeHD+/5YxDHjx9vbx48eNDeVFXduXOnvZm855P3YfJ+T65vVlUtLS21N3/++Wd7c+7cufZmcmF28t5VVZ06daq92djYaG/++OOP9mZy7bSq6saNG+3N5PvbC08KAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCANG/ANbw0UcftTcrKyuj15ocM1tdXW1vdnZ22pvJMcHJ36dqdqhucixscsBrd3e3vamq2t7ebm8mR9MmB/sm793kM1RVtby83N5MDrSdPHmyvZn8LO2nyedhcbH/f+bJIcaq2eHC6e+Ih/GkAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABB7vqj0wQcftL/4O++8095cvHixvamqOnfuXHszOR43OZp29OjR9mZ6NG1ydG7yPkwOeE0OjFVVra+vtzeT92FyzGxhYaG9mf7bTo4Q3rx5s7357rvv2pvJ52H6PkxMDgNubm62N8eOHWtvqmbf3xNPPDF6rYfxpABAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQez6I99JLL7W/+BdffNHefP311+1NVdUnn3wy2nUdOXKkvZkcnDt9+nR7M909/vjj7c3kANrkSF1V1a1bt9qbq1evtjdbW1vtzdraWnszOaJXVfXVV1+1N88//3x78+yzz7Y3H374YXuzvb3d3lTNDyt2HT6851+P8eSTT45e68SJE+3N5FDkXnhSACBEAYAQBQBCFAAIUQAgRAGAEAUAQhQACFEAIEQBgBAFAEIUAIiF3emVsgNmY2Ojvbly5Up7Mzm09tlnn7U3q6ur7U3V7EDb5uZmezP52EwPwU0OoE0OA168eLG9efnll9ubV199tb2pqjp27Nhotx9ee+219ubnn38evdaZM2fam8nBuckhy8kRvaqqpaWl9uatt95qb5aXlx/6ZzwpABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABD/mCupAPznPCkAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAIQoABCiAECIAgAhCgCEKAAQogBAiAIAIQoAhCgAEKIAQIgCACEKAMS/ANJT3ENkQDZRAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(X_test[0], cmap=\"binary\")\n",
    "plt.axis(\"off\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f81bc80f",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9b29c339",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}