7 - TensorFlow CNN with Aleatoric uncertainty binary classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 7 - TensorFlow CNN with Aleatoric Uncertainty"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We've seen how easy it is to train a convolutional neural network in TensorFlow 2.0 `keras`, but let's now try to add on some aleatoric uncertainty by having that CNN parameterize a probability distribution, rather than directly trying to guess the output \"probabilities\"."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.0.0\n"
     ]
    }
   ],
   "source": [
    "from __future__ import absolute_import, division, print_function, unicode_literals\n",
    "\n",
    "import os\n",
    "import pickle\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras import datasets, layers, models\n",
    "import tensorflow_probability as tfp\n",
    "\n",
    "print(tf.__version__)\n",
    "tf.get_logger().setLevel('ERROR')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For this exercise, we'll focus again on the CIFAR-10 data set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "(train_images, train_labels), (test_images, test_labels) = datasets.cifar10.load_data()\n",
    "\n",
    "# Normalize pixel values to be between 0 and 1\n",
    "train_images, test_images = train_images / 255.0, test_images / 255.0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To simplify things, we'll limit the classification task to just two classes - birds and planes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "chosen_labels = [0, 2]\n",
    "\n",
    "train_images = np.array(train_images[[i for (i, label) in enumerate(train_labels) if label in chosen_labels]])\n",
    "train_labels = np.array([label for label in train_labels if label in chosen_labels]) // 2\n",
    "test_images = np.array(test_images[[i for (i, label) in enumerate(test_labels) if label in chosen_labels]])\n",
    "test_labels = np.array([label for label in test_labels if label in chosen_labels]) // 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 25 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "class_names = ['airplane', 'bird']\n",
    "\n",
    "plt.figure(figsize=(10, 10))\n",
    "for i in range(25):\n",
    "    plt.subplot(5, 5, i+1)\n",
    "    plt.xticks([])\n",
    "    plt.yticks([])\n",
    "    plt.grid(False)\n",
    "    plt.imshow(test_images[i], cmap=plt.cm.binary)\n",
    "    plt.xlabel(class_names[test_labels[i][0]])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### It's a bird, it's a plane..."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll start by building our convolutional network as before."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = models.Sequential()\n",
    "model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(32, 32, 3)))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Conv2D(64, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Conv2D(64, (3, 3), activation='relu'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_6 (Conv2D)            (None, 30, 30, 32)        896       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 15, 15, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_7 (Conv2D)            (None, 13, 13, 64)        18496     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_5 (MaxPooling2 (None, 6, 6, 64)          0         \n",
      "_________________________________________________________________\n",
      "conv2d_8 (Conv2D)            (None, 4, 4, 64)          36928     \n",
      "=================================================================\n",
      "Total params: 56,320\n",
      "Trainable params: 56,320\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, though, instead of outputting a final `layers.Dense(10, activation='softmax')`, we'll instead scale the number of outputs down to the number of parameters a probability distribution needs in order to describe our new number of classes (2). Fortunately, TensorFlow Probability comes with such a helper function."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfp.layers.IndependentBernoulli(200).params_size()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We're using the Bernoulli distribution (sometimes called the coin-flip distribution) in order to describe predictions of our two classes. You probably already knew that the Bernoulli distribution takes a single parameter, but TFP saw fit to remind us anyway."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.add(layers.Flatten())\n",
    "model.add(layers.Dense(64, activation='relu'))\n",
    "model.add(layers.Dense(1))\n",
    "model.add(\n",
    "    tfp.layers.DistributionLambda(\n",
    "        lambda t: tfp.distributions.Bernoulli(logits=t)\n",
    "    )\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_6 (Conv2D)            (None, 30, 30, 32)        896       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 15, 15, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_7 (Conv2D)            (None, 13, 13, 64)        18496     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_5 (MaxPooling2 (None, 6, 6, 64)          0         \n",
      "_________________________________________________________________\n",
      "conv2d_8 (Conv2D)            (None, 4, 4, 64)          36928     \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 64)                65600     \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 1)                 65        \n",
      "_________________________________________________________________\n",
      "distribution_lambda_2 (Distr ((None, 1), (None, 1))    0         \n",
      "=================================================================\n",
      "Total params: 121,985\n",
      "Trainable params: 121,985\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that, because our dense layer lost 9 of its activations, we actually have _fewer_ trainable parameters to learn. The Bernoulli distribution at the end doesn't wind up costing anything in parameter space -- it only consumes the outputs of other layers.\n",
    "\n",
    "Let's get to training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "checkpoint_path = 'saved_models/aleatoric_cnn.ckpt'\n",
    "\n",
    "model.compile(optimizer=tf.optimizers.Adam(learning_rate=0.001),\n",
    "              loss=lambda y, p_y: -p_y.log_prob(y),  # Use the negative-log likelihood directly as the model loss.\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Run this cell to train the model from scratch.\n",
    "\n",
    "# cp_callback = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_path,\n",
    "#                                                  save_weights_only=True,\n",
    "#                                                  verbose=1)\n",
    "\n",
    "# history = model.fit(train_images, train_labels, epochs=20, \n",
    "#                     validation_data=(test_images, test_labels), callbacks=[cp_callback])\n",
    "\n",
    "# history_dict = history.history\n",
    "\n",
    "# with open(os.path.join(os.getcwd(), 'saved_models', 'aleatoric_cnn_history.pkl'), 'wb') as file:\n",
    "#     pickle.dump(history_dict, file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Run this cell instead to load pre-trained model weights\n",
    "model.load_weights(checkpoint_path)\n",
    "\n",
    "with open(os.path.join(os.getcwd(), 'saved_models', 'aleatoric_cnn_history.pkl'), 'rb') as file:\n",
    "    history_dict = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2000/1 - 1s - loss: 0.6163 - accuracy: 0.8795\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history_dict['accuracy'], label='accuracy')\n",
    "plt.plot(history_dict['val_accuracy'], label = 'val_accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.ylim([0.5, 1])\n",
    "plt.legend(loc='lower right')\n",
    "\n",
    "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So a bit overfitting going on here -- remember, we didn't allow for uncertainty over the model weights, only in the nature of the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2000/1 - 0s - loss: 0.6163 - accuracy: 0.8835\n"
     ]
    }
   ],
   "source": [
    "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our first image in the test set is a plane (label zero) -- let's see if we got it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0]], dtype=int32)"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.predict(test_images[0:1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It looks like it. That extra index in the numpy array is kind of annoying -- we can clean it up with `np.squeeze()` and `np.ndarray.tolist()`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The third image was a plane, but it looked a lot like it could be a bird. How was that one?\n",
    "\n",
    "Recall that our model is actually passing a distribution whenever it is called on an input image. Let's use the distribution directly to answer our question."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [],
   "source": [
    "third_prediction = model(test_images[2:3]).sample(1000).numpy().squeeze()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0,\n",
       "       0, 1, 0, 1, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 0, 0, 1, 0,\n",
       "       0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 0, 1,\n",
       "       0, 1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 0,\n",
       "       0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "       1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 1,\n",
       "       0, 0, 1, 1, 0, 1, 1, 0, 1, 0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0,\n",
       "       1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0,\n",
       "       0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 1, 0,\n",
       "       0, 1, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 1, 0,\n",
       "       1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0,\n",
       "       0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1,\n",
       "       1, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0,\n",
       "       1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 1,\n",
       "       1, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 0, 1, 1, 1, 0, 1, 0,\n",
       "       0, 0, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1,\n",
       "       0, 0, 0, 0, 1, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0,\n",
       "       0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 1, 1, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0,\n",
       "       1, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0,\n",
       "       1, 1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 0, 1, 0, 1, 0, 0, 0,\n",
       "       1, 0, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1,\n",
       "       0, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1,\n",
       "       1, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0,\n",
       "       0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0,\n",
       "       0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0,\n",
       "       1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0,\n",
       "       0, 1, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1,\n",
       "       1, 1, 0, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1,\n",
       "       0, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0,\n",
       "       1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 1,\n",
       "       0, 1, 0, 1, 1, 1, 0, 0, 1, 0, 1, 1, 0, 1, 0, 1, 0, 0, 0, 0, 1, 0,\n",
       "       0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0,\n",
       "       0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0, 0, 0, 1,\n",
       "       1, 0, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0,\n",
       "       0, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0,\n",
       "       0, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1, 0, 0, 1,\n",
       "       1, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1,\n",
       "       0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1,\n",
       "       0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 0, 1,\n",
       "       0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0,\n",
       "       0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0,\n",
       "       0, 1, 1, 1, 0, 0, 0, 0, 0, 0], dtype=int32)"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "third_prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.337"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "third_prediction.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The model is quite uncertain, which means that it's picked up on the fact that it's a fuzzy image that could potentially go either way."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Examining uncertainty up-close"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From a high number of samples, an input image that causes a low mean prediction will have a high probability of being classified as a plane, and an image that causes a high mean prediction will have a high probability of being classified as a bird.\n",
    "\n",
    "Let's find the images that are most in the middle. These would be the inputs that the model is most uncertain about from an aleatoric standpoint -- it hasn't been able to learn features that would distinguish this image consistently. Following our intuition about high and low mean outputs, we should want the input images that have a mean close to 0.5. We can highlight these by subtracting 0.5 from all means, taking the absolute value, and finding the minimum entry.\n",
    "\n",
    "This cell takes about 20 seconds to run."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_samples = 1_000\n",
    "\n",
    "preds = [abs(pd.DataFrame(model(test_images[i:i + 1]).sample(n_samples).numpy().squeeze()).mean() - 0.5) \n",
    "         for i in range(len(test_images))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0    0.392\n",
       " dtype: float64, 0    0.085\n",
       " dtype: float64, 0    0.17\n",
       " dtype: float64, 0    0.498\n",
       " dtype: float64, 0    0.499\n",
       " dtype: float64, 0    0.499\n",
       " dtype: float64, 0    0.427\n",
       " dtype: float64, 0    0.5\n",
       " dtype: float64, 0    0.5\n",
       " dtype: float64, 0    0.219\n",
       " dtype: float64]"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds[0:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "min_ind = np.array(preds).argmin()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at sample model predictions for this minimum index. Remember, we're interested in finding images that have a roughly equal number of zeros ane ones in their sample predicted outputs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 1, 0, 0, 1, 1, 1, 0, 1, 1, 0, 1, 0, 0, 1, 0, 0, 1, 1, 0, 0, 1,\n",
       "       0, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 0, 0, 1,\n",
       "       0, 1, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 0, 1, 0, 0, 0, 1, 1, 1, 1, 0,\n",
       "       0, 1, 1, 0, 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1, 0, 0, 1, 0, 1,\n",
       "       0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0, 1, 0,\n",
       "       1, 0, 1, 0, 0, 1, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0, 1,\n",
       "       0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0,\n",
       "       1, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0, 0, 1, 1, 0,\n",
       "       1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 0, 0, 1, 0, 1, 1, 1, 0,\n",
       "       0, 0, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 1,\n",
       "       1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0,\n",
       "       1, 0, 1, 0, 1, 1, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1,\n",
       "       0, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 1,\n",
       "       0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1,\n",
       "       0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 1, 0, 0,\n",
       "       0, 1, 1, 1, 0, 0, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1,\n",
       "       1, 0, 0, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0,\n",
       "       0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 0,\n",
       "       1, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1,\n",
       "       1, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 0,\n",
       "       1, 1, 0, 1, 0, 0, 0, 1, 1, 1, 1, 0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0,\n",
       "       0, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 0, 0, 1,\n",
       "       1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 1, 1,\n",
       "       0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,\n",
       "       0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 1, 0, 0, 1, 0, 1,\n",
       "       0, 1, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1,\n",
       "       1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1,\n",
       "       0, 0, 1, 1, 1, 1, 0, 1, 0, 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1,\n",
       "       1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1,\n",
       "       0, 1, 1, 0, 0, 0, 0, 1, 0, 1, 1, 1, 0, 1, 0, 0, 1, 0, 1, 1, 1, 1,\n",
       "       1, 0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1,\n",
       "       1, 0, 1, 1, 0, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1,\n",
       "       1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 0, 0, 1, 0, 1, 1,\n",
       "       0, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 1, 0, 1, 0, 1, 0, 0,\n",
       "       1, 0, 0, 0, 1, 0, 1, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 1, 1, 1,\n",
       "       1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 1,\n",
       "       1, 1, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 1, 0,\n",
       "       0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1,\n",
       "       0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 0, 1,\n",
       "       0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 1, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1,\n",
       "       0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 0, 0, 1,\n",
       "       1, 1, 1, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 0,\n",
       "       0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0,\n",
       "       0, 1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 0, 0, 0,\n",
       "       0, 1, 1, 1, 0, 0, 0, 0, 1, 0], dtype=int32)"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model(test_images[min_ind:min_ind+1]).sample(1000).numpy().squeeze()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.482"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model(test_images[min_ind:min_ind+1]).sample(1000).numpy().squeeze().mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Excellent! This image is almost right in between the two classes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x1566a0a90>"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(test_images[min_ind], cmap=plt.cm.binary)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Throwing a wrench in the plan"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We've seen that the aleatoric model successfully conveys uncertainty for images that could either be classified as a bird or a plane. How does it do for an image that isn't either?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "(train_images, train_labels), (test_images, test_labels) = datasets.cifar10.load_data()\n",
    "\n",
    "# Normalize pixel values to be between 0 and 1\n",
    "train_images, test_images = train_images / 255.0, test_images / 255.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x156c16438>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD5CAYAAADhukOtAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAesElEQVR4nO2da4ycZ5Xn/6fu3V19c9tuXxPHzoUYhjhZK5sRLAuDmMkipIA0iuADygc0Hq0GaZFmPkSMtLDSfmBWC4hPrMwSTWbFctkJEdEMA4SIVWbQTIgTEufi2LGN77e2232tqq7b2Q9VHtnZ5/90x91dbXj+P8ly9XPqed9T7/ueequef51zzN0hhPjdJ7PWDggheoOCXYhEULALkQgKdiESQcEuRCIo2IVIhNxyJpvZwwC+ASAL4H+6+1diz89kzDOZ8PuLmd2MB9QSExRvZk+djYa3Gt1X5HXF/IhtMyaX5rPh41sq8FOdJecEAFqtFrVV6w1qa7fJvrJZOid2rNpt7ge7phbbJuNm1ejYrmLbdHK2izn+uth5nq0uoFZvBj2xm9XZzSwL4AiAjwE4A+BFAJ9x9zfZnFwu68NDA0FbhjgPAEZPJr9w2MXWmcXPSuyjTotstN5q0jn5fJ7ailkegK3IaalHgmzLcDE4fvf2DXTO0GCZ2qampqjt0Mlz1DZbDY8PDw/TOYU8P5/zc3PU1tffz7dZKFAbox15g/NIQMfedJotfkG22uHzuWOMv65t68Ln7KlfvoGJ6fmgl8v5GP8ggKPuftzd6wC+B+CRZWxPCLGKLCfYtwI4fd3fZ7pjQohbkGV9Z18KZrYPwD4AyGRu+tuyEGKZLOfOfhbA9uv+3tYduwF33+/ue919780twgkhVoLlBPuLAO4yszvMrADg0wCeWRm3hBArzU1/jHf3ppl9HsBP0VkWf8Ld34jNKQ/04wP/9t8EbacuXKbzTp8+FRzP5/hK9yLiFbVY5P2PfQ0pZiKyVuTTTCaihMRWhLMZ/rotH17tHtpyF51z1727qW1mkp+XgY0nqa2ZCasuA5GV83K5RG0v/uqfqa1WI0v/AGq1WnA8du24xXQyfj5bzlfcGxHFZuNgX3B8nChXADBXqQfH223u+7K+s7v7jwH8eDnbEEL0Bv2CTohEULALkQgKdiESQcEuRCIo2IVIhFX/Bd0NO8vmMDo8ErRt2cR/aTt1ZTI4PjcXllWAuOTVjmWN5XgyRsbC742ZiOSyYZwnoMSSKlqRxIk7dtxJbb+35/3B8dt27aRzBgZGqS0XkageiEiYGQsfx4WFBTqnHJHldu8JS7YAcP4clwB/8bNng+MXzv5/v//6V8hpXtwWSebK5/jJ3rYh/LrdIwk5WXLsY4k63CSE+F1CwS5EIijYhUgEBbsQiaBgFyIRbros1c3QVyz4ji3jQdsnPv4xOm+uGi7b88+/+jWdMzHBEzhiibaZSBmpof5wwsLt27mS8MEP/Ttqi1TOiiY0bNp6G7Vt2LopOD48xMtB9ZWGqK0Z8aPZ5MkdRo5yLlKDrtHgq899A7y81NRVfq7ffPVgcPwnz/yIzjl3lq/u1xu8JJiB+79zS1iFAoAtw+EEoLlKJBmKrO7/5MUjuDJTWfGyVEKI3yIU7EIkgoJdiERQsAuRCAp2IRJBwS5EIvQ0ESabAUZIWa1fH+Tl6z7ykY8Exx/dvD04DgDPPf9Lapua4V1OhtbxpJBt45uD4+/dfS+ds3XbFmrrK/NOLLEElFyB226/PSzL1aMdZrjEk4t0rSG5LgCAYiZstEjyTKnEN1geCsueAGDGZa17dr+P7IvXu3vt1Zeo7efP/pzamgsVastGkqVqjXA9uUY7kszVCG8vJqXrzi5EIijYhUgEBbsQiaBgFyIRFOxCJIKCXYhEWJb0ZmYnAMwCaAFouvve2PP7ijm8f2e4JtvpC+E6cwDwf3/+0+D4Qw/9ezrn/ffdR20zkXZBIyNcxhksDwbH142HM80AoBCpq+YR7WpkhEuA5T6eAVYgslxljstCiMg1pRLfV6frV5giyW5rRTLlyhEpMhup79bu5/7XR8Ky4s577qZzLkdaXo2tG6O25iyvr1eK1KCbr8yHDc4l1mIpLEWybENgZXT2j7g7PzpCiFsCfYwXIhGWG+wO4Gdm9pKZ7VsJh4QQq8NyP8Z/0N3PmtlGAM+a2Vvu/vz1T+i+CewDgHJfrMWyEGI1Wdad3d3Pdv+/BOBpAA8GnrPf3fe6+95Soac/xRdCXMdNB7uZDZjZ4LXHAP4QwOsr5ZgQYmVZzq12HMDT1mmzlAPwv939J7EJ7bajWgkX7LttHUmHAzC3EM4KOnbyCJ2z5Y73Utv2TduordjHs6EG+sJyx8j6dXTO6EhYrgOAFslcAoChIT5v3XBkm0SiKmYipzrL5ZpChtvyeb5NmuQVKTjZbvPj0R+RMKsLkcKXpA1YO9Jea2CAX4t77r+f+3GVt/qqXjlPbaOD4a+301dn6Zx6nbyuSHbdTQe7ux8HwMVsIcQthaQ3IRJBwS5EIijYhUgEBbsQiaBgFyIRevorl7Yb5ohkUGvxoocs06iwhUto5VHev2zdKM8oc+dS09BgWHorRCSoRp33BiuT3nEAMBCRAC3P5au56ZngeK0ali8BoL+fS01NflqQAZe8cixLLdJob3Y27DsA5CM9+HIRebCUC/vRv2kjnVMscnltbIxnvR0+GO4rBwB//+u3qG2+Hn7dsf5wW4lcl40cC93ZhUgEBbsQiaBgFyIRFOxCJIKCXYhE6OlqvANokOVYQyTXfeSO4PD49nvolIEBnjjRTxJaAKDVjtRIGwivkFub10DziG10lK+2DkRW6meqvJ7c1PQ02R5XJ4Yjflyd5LUBa1Vey28DSQ7KWOz+wo/VpYkJauuLtHJiNflykdX9DRv5Sn2+EKnJF0muMeev+4dPPx0cP3CEV3vzO8OtyCKdvHRnFyIVFOxCJIKCXYhEULALkQgKdiESQcEuRCL0XHprkkST4bHb6LyNt70nOF4qcnmqkOfSWzbHX3Yh0qbHEdY1+vp5TbixMV6fLhepx0ZbAgFYaPCklo0bw3XQpian6JzfHD5MbaUil6jyGa7zWDOcXJOJHHtEZM92JFHq6uRVasuSYzw0MkznxFpUbdjA68xV5uaordTHr8c8qa/39z8Ntz0DgBcPhxNrKjWeeKU7uxCJoGAXIhEU7EIkgoJdiERQsAuRCAp2IRJhUenNzJ4A8AkAl9z9fd2xdQC+D2AHgBMAHnV3rn90yWSy6B8ISx633R2W1wBgdCycldVuR9oWlYoRR/i8WFugRi0seeXX8UyoYiRLql6rcT8iGWD9kdZQfbnw6/7lvzxD55w9/hq17dixm9p27bqb2uYKYf+LZe779BTP5kOGn892g8tNIJKdRVpN9RX4vrKRrL3RSH26U2fOUtuOu8LH8ZEclz2fqoWP1eGjb9M5S7mz/zWAh98x9jiA59z9LgDPdf8WQtzCLBrs3X7r70xqfgTAk93HTwL45Ar7JYRYYW72O/u4u19rS3kBnY6uQohbmGUv0Lm7I1JixMz2mdkBMzuwEKmhLoRYXW422C+a2WYA6P5/iT3R3fe7+15331skJYKEEKvPzQb7MwAe6z5+DMCPVsYdIcRqsRTp7bsAPgxgvZmdAfAlAF8B8AMz+xyAkwAeXcrO8oUiNu3YFbQNj62n85pEWimWynSOc+UKZlx6i2VQDRTDMlo2w98zqxVelNFbPHvNI9JbaYEXWJydCe/vyqWLdM7OHZuoLZ/nx2qhwrO80A6fz0Ih0taqzTP9pq7wwpeIZA/miuFz02gs0Dkjg1werEcyDvsjRUJj2Y9ZC19Xw5G2XGdOnQiPnztH5ywa7O7+GWL66GJzhRC3DvoFnRCJoGAXIhEU7EIkgoJdiERQsAuRCD0tOJnN5zGyIdxHKxfJ8Mlkwm4WiRQGAJVIwcaY9Far8l/5lUgGWzbSN2wu0petHOk5V434n52PSV7hbK4td/KswnJEDmtFsgDb5VFqa+TD8pWD7ysXySg7+tbr1JZp8WKa67eFpd7xTeFeaQCQiVwfzVhW5AKX5YYi2X4Xzl0Iju+4fQeds+f++4Lj//AsL1KpO7sQiaBgFyIRFOxCJIKCXYhEULALkQgKdiESoafSGwBYJpyhlDGeuZQhWWURhQSlEpd4PFKoshTJpMsSebAakddyWX6IMzkuHc7OT3A/LJLSR9hGsg0BYID0GgOAYkQStSy/VzTIObswwbPvpi+dobY8ePbglTORIosevg78vbx33MICz4jLRIqVViv8Oliocv9ZBt7UNM/A3LwlXBwqH6kZoTu7EImgYBciERTsQiSCgl2IRFCwC5EIPV2NNzPkWL2wyMo6Wx1lCTIA0BdZYbbIyn+sjph5Mzg+fZnXRytH6pl5pFCeRxIuarM8SWZubjY4PhRpTZSLJPKUh8PtugDg3G9OUFuV1GrbuJH7gcj5jF2qJ8/xle7xe0lCTkTQqEcSWmpNnijVbISvj8UYI+fmypUrdE5lJrxS34iUa9edXYhEULALkQgKdiESQcEuRCIo2IVIBAW7EImwlPZPTwD4BIBL7v6+7tiXAfwJgGvZGl909x8vZYdMYWu2ePJBlsh1jUgrnnyDv7RYK57+fl5XbeJiOInj0gSX3hotrvHUJ2g/TORjCSgRWbHRDksvkdwf5Iu8zdCRN96gtqP/+DzfZiF8Hxn46B/ROSPrt1Nb+TKvu3ff3t+ntmEqHXJpM09qDQLA9HxY2lyMQmSbs7PhbV4k1xsAHDt6JDheiSTcLOXO/tcAHg6Mf93d93T/LSnQhRBrx6LB7u7PA4h01RNC/DawnO/snzezg2b2hJnxz75CiFuCmw32bwLYBWAPgPMAvsqeaGb7zOyAmR2ozPPkfiHE6nJTwe7uF9295e5tAN8C8GDkufvdfa+77+0f4AtLQojV5aaC3cyub6fxKQC8XYcQ4pZgKdLbdwF8GMB6MzsD4EsAPmxmewA4gBMA/nQpO8tYBqW+sMxTjrTHYYlGsVph7TavMUbKowEA5udnqO3S5cvB8cEyzwyr1rmPFZKhBgBjY+upzSKy3EK9FhyfneQZVCOjfF9vvnaI2qZI2yIAKJPWXCePn6Jz3vfAA9TWP8yvj9YcXz+uTYXP2dWJ83ROPpK9FpO2WOstAGg0eDZalWzz/AXu49Fjx4PjsZhYNNjd/TOB4W8vNk8IcWuhX9AJkQgKdiESQcEuRCIo2IVIBAW7EInQ04KTmUwGpb7wD2uGykN0XmU+nN3WWOByBiIFBY8dPUpttYh0MTI6Ehxft4FLV9UKl2qGR3j2XT0i/0xOhuUkALh07mxwPB+RhWItg+YjmYUvnz5NbRuJnHdPpPBlK8tT8wbJsQeA+swUtU1NhSXHSo2/LkzytkvI82Klrcg5a0eOf5a0RGtErsULRJaLSXy6swuRCAp2IRJBwS5EIijYhUgEBbsQiaBgFyIReiq9ORzeCmejtetcCukvhaWJ+Vme2daK9OtqNfm8kREuDZVKYYkqY/w9c2SEy3LtFpdJ5iOFDVsVLjWBFLicvMSLWzbqr1Abkz0BYHznbmrbdeeu4Pi2rVvpnHadS1eFSB+4IsmkBIAM6Yl2MXI8+oa4JFqM1GRoNbn/1Wo4GxEAFmphic0jsufVyXCmX8wH3dmFSAQFuxCJoGAXIhEU7EIkgoJdiETo6Wp8h/BqMWvxBAAwMidWTC7C0BBPuoFFEh2IkpDP80SSwcFIgk+Fr7gXmiVqm6tyH9dv3hYc37KZr4IvtHgCik9NU9tQRLkYGQnX5fvN6y/SOaVIQk49w48H2nwF+uL5cMJIJcdrDd57H1+Nb5BWTUA82eXKlUifFQ8f/wvneY2/KXJemuQaBXRnFyIZFOxCJIKCXYhEULALkQgKdiESQcEuRCIspf3TdgB/A2AcHd1sv7t/w8zWAfg+gB3otIB61N0jxbsAd6DRDMsTrTYvGpcn7Y7KEQltMlZHjCskqNfnqW1kXbgOWqlYpHOKBX6I+/ojSTIT3MmFGm+FVJ0lMlSW1zPLFvh7fpnnmCA3xF/3+rGw9Pb2y7+kc+ameYLP5jvupbZCnkuHL7/6WnB8w9330TnI8nPWirTz6i/38U3ORqTldtj/DH9ZyJFaeFbnk5ZyZ28C+HN33w3gIQB/Zma7ATwO4Dl3vwvAc92/hRC3KIsGu7ufd/eXu49nARwCsBXAIwCe7D7tSQCfXC0nhRDL5119ZzezHQDuB/ACgHF3v/bzpAvofMwXQtyiLDnYzawM4CkAX3D3G35r6O4O8jtYM9tnZgfM7MD8PP8+LIRYXZYU7GaWRyfQv+PuP+wOXzSzzV37ZgDB0h/uvt/d97r73oGByGqPEGJVWTTYzczQ6cd+yN2/dp3pGQCPdR8/BuBHK++eEGKlWErW2wcAfBbAa2Z2rVjZFwF8BcAPzOxzAE4CeHSxDXm7jUYt/FG+tsDbJPX1hSWNgUiW1OQ0z9aamuIST38/rzGWI5KMRSSS/v5CxMY/6bQaXOKpV7lt8nK43dFslWd5jQyVuR9VXget3eZ11fo2bwmOrxvfTOfMz/PXdelcOHsNAK5G2mEdOXYi7Meuu/n2Jnh9ukxEt+0rcynSnWejVebnguMWyebrJ7pcNSbXcVMHd/8nAGwTH11svhDi1kC/oBMiERTsQiSCgl2IRFCwC5EICnYhEqGnBSeb9QVcPnU8bORJbzALyxalPp71ZhE9bGGBS0a5HM9OyhC5I7av2RkuAdarYckFAIqRApwxeXDdWLgI5OwU93FyhhdRzLT5/aC/xItAThAJsBbZXnFsE7VdjcilZ67w41hbCMt5F0+8SedcGOFZhdtuv5PaZq7wTMtjhw9TW6NWCY5fPn+azvm9HeHz/MIRLpXqzi5EIijYhUgEBbsQiaBgFyIRFOxCJIKCXYhE6Kn01m7VMTN5Jmwj/dwAIJcLZxqVhxp0TmWOy0nW4NJbbYZLF7VKuAfYTKT/VzHSB250KFzAEgDyfXxe/0Ck75mFfRwd4/3Lpqd5RtzcFJeTKhWebTZ9kWSOkeKKANAsRjIOh/m8Vp5LVOsGw8fR57gkeuXcOWrbNL6D2i5c5H689C8vUNs97wnLeY0mv77Hx8IZk7kcv3/rzi5EIijYhUgEBbsQiaBgFyIRFOxCJEJPV+PRbsMa4aSFuQmSIAPggodrcQ1v4Cvn9Rav+TUQWemu1vg2K6QU9kyk3t2O22+ntpHRcIskAKhF6o8ZFxNoss5ApEXVQClyPCItjc6f5z5eJaZSiW/PjSf/FLJc8ahGVtbXD4WVizJpnwQAM5MXqe3c6aPUdvjwEWprRGosTp4/FRwfGokoKAvh49GKtDbTnV2IRFCwC5EICnYhEkHBLkQiKNiFSAQFuxCJsKj0ZmbbAfwNOi2ZHcB+d/+GmX0ZwJ8AmOg+9Yvu/uPFtpclili7xWWcy0TumJ3mSRrj23ZSW6HAa9fVM/z9r9EKJyZkI/Xi6nWuk7XavN3RQKTO3OxVLjW1GmEfJ6bCNeEAoBBpo7WwwKXIQoH7ODIS3majzbfXqPHEj3OnwvIUAMxOTlDb9hHSfisibXozXBMOAC6dO0ZtC5EuxeUibwNWuRpOKBrbuI3OmSPJS7FWZEvR2ZsA/tzdXzazQQAvmdmzXdvX3f2/L2EbQog1Zim93s4DON99PGtmhwBsXW3HhBAry7v6zm5mOwDcD+Bacu7nzeygmT1hZqMr7JsQYgVZcrCbWRnAUwC+4O4zAL4JYBeAPejc+b9K5u0zswNmdqBW59/JhBCry5KC3czy6AT6d9z9hwDg7hfdveXubQDfAvBgaK6773f3ve6+txRZCBJCrC6LBrt12p18G8Ahd//adeObr3vapwC8vvLuCSFWiqWsxn8AwGcBvGZmr3THvgjgM2a2Bx057gSAP13KDtuk1Fyk+xMK5C3JapHaaRMnqa0yyWu49Y1uprbyYDnsB2lPBQALkWynXIYffo9IXsUs/4Q02wz7EpNkshE/CpHssIUaf92tZlhynJoitekANBr83nP0rUPUNshdRD4b3matyY9vrhWRB2cmqS1b5+e66FxmLQ+F203lW/z6ztXD2zMWYFjaavw/AQhdKotq6kKIWwf9gk6IRFCwC5EICnYhEkHBLkQiKNiFSITeFpwEqAaUiWhDmQxp4RN5q5qeOE9tuSx/2X2D/Fe/eSJRzUWynYp5Xuhx8govotgmRTYBoM0VL7RJoc1qlctC1SqXhbIZnq3VIBl2AFAj8lVllu/r/BmeUdaa55lto/3cxwzJSOzPxVpoRYTgFs9i9Da3DQ/zQpvIhPdXj8i2l2bDx74Zkd50ZxciERTsQiSCgl2IRFCwC5EICnYhEkHBLkQi9FZ6c8BJLyqLFHo0JstFss2KkWwtmnoHoLHAZTRvh53PF7iM03S+r+m5cN87AKhUuC1WLDGXC7/uWFHM2Vn+mg1cXusbDGdrAUCBFBCtzHG58cJveGbbxqFINhfJbAOALJFZGw0uAbZbEfmKyMBdT6il1eI+FrNhmXKqxn28OB8+Lw1Jb0IIBbsQiaBgFyIRFOxCJIKCXYhEULALkQg9z3oz9v4SyXprNlmPNS6D5PL8fazV4AUFLcP9sExYeisPcAlqPiKfuPNedbVIllqjyqWyXC58SlskG24xGuB+VCamqK1VCWeAnXr7NTqn3Mcvx2KWS0qVSD+CYil8PlstogEDaEWkt0KB24bKMfk4ksVITBev8mNfr4cnuaQ3IYSCXYhEULALkQgKdiESQcEuRCIsuhpvZiUAzwModp//t+7+JTO7A8D3AIwBeAnAZ92dL3MDgAFGV1X5anEmE07iiHQ0WmT1mc9sRFbq5+amg+Nj68fpnIxxxeDcWV4nb7A8QG2vv87b6jHlYv36DXROocBruOWz/FidOslbbDXmwm2SbIG3TxoditSSA082akXaLrG+YrHXHFuNj11zmUhxwFJkf8cvhK+ry7NcMWAxEXNwKXf2BQB/4O73odOe+WEzewjAXwH4urvfCeAqgM8tYVtCiDVi0WD3DtfyLfPdfw7gDwD8bXf8SQCfXBUPhRArwlL7s2e7HVwvAXgWwDEAU+7/Wu/4DICtq+OiEGIlWFKwu3vL3fcA2AbgQQDvWeoOzGyfmR0wswO1yC+dhBCry7tajXf3KQC/APD7AEbM7NoC3zYAZ8mc/e6+1933lgqxKh9CiNVk0WA3sw1mNtJ93AfgYwAOoRP0f9x92mMAfrRaTgohls9SEmE2A3jSzLLovDn8wN3/zszeBPA9M/uvAH4N4NtL2SGr4xarQZch7XHc+dcCKk0AaDR5UkIsyWRmaiY4fnWSy0nl8hC1tRa4/2cuc1nryFtvUVu7HZYOSzmuyZSHx6jNc1wy2jAWkRw3httoVfu4PFW5fIb7UeBttEZGeCISu8CrkWOPSJJMLhe5Pzb5vLn5SNuraWKzSG09krBlEe1t0WB394MA7g+MH0fn+7sQ4rcA/YJOiERQsAuRCAp2IRJBwS5EIijYhUgE80h7ohXfmdkEgGua0noAl3u2c478uBH5cSO/bX7c7u7BFMeeBvsNOzY74O5712Tn8kN+JOiHPsYLkQgKdiESYS2Dff8a7vt65MeNyI8b+Z3xY82+swsheos+xguRCGsS7Gb2sJkdNrOjZvb4WvjQ9eOEmb1mZq+Y2YEe7vcJM7tkZq9fN7bOzJ41s7e7/4fTxlbfjy+b2dnuMXnFzD7eAz+2m9kvzOxNM3vDzP5Td7ynxyTiR0+PiZmVzOxXZvZq14//0h2/w8xe6MbN982MpySGcPee/gOQRaes1U4ABQCvAtjdaz+6vpwAsH4N9vshAA8AeP26sf8G4PHu48cB/NUa+fFlAH/R4+OxGcAD3ceDAI4A2N3rYxLxo6fHBJ0aseXu4zyAFwA8BOAHAD7dHf8fAP7ju9nuWtzZHwRw1N2Pe6f09PcAPLIGfqwZ7v48gHcmwT+CTuFOoEcFPIkfPcfdz7v7y93Hs+gUR9mKHh+TiB89xTuseJHXtQj2rQBOX/f3WhardAA/M7OXzGzfGvlwjXF3v1ZI/gIAXhli9fm8mR3sfsxf9a8T12NmO9Cpn/AC1vCYvMMPoMfHZDWKvKa+QPdBd38AwH8A8Gdm9qG1dgjovLODtjdYdb4JYBc6PQLOA/hqr3ZsZmUATwH4grvfUBaol8ck4EfPj4kvo8grYy2C/SyA7df9TYtVrjbufrb7/yUAT2NtK+9cNLPNAND9/9JaOOHuF7sXWhvAt9CjY2JmeXQC7Dvu/sPucM+PSciPtTom3X2/6yKvjLUI9hcB3NVdWSwA+DSAZ3rthJkNmNngtccA/hAA76u0+jyDTuFOYA0LeF4Lri6fQg+OiZkZOjUMD7n7164z9fSYMD96fUxWrchrr1YY37Ha+HF0VjqPAfjLNfJhJzpKwKsA3uilHwC+i87HwQY6370+h07PvOcAvA3g5wDWrZEf/wvAawAOohNsm3vgxwfR+Yh+EMAr3X8f7/UxifjR02MC4P3oFHE9iM4by3++7pr9FYCjAP4PgOK72a5+QSdEIqS+QCdEMijYhUgEBbsQiaBgFyIRFOxCJIKCXYhEULALkQgKdiES4f8BGkQPQLCyB30AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(test_images[1000], cmap=plt.cm.binary)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1], dtype=int32)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model(test_images[1000:1001]).sample(1000).numpy().squeeze()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Apparently, our model is certain that this puppy is a bird. This is an important point -- aleatoric uncertainty doesn't allow us to convey uncertainty about data far away from what the model has trained on; it only lets a model indicate the variance present in the data itself. In this case, it does nothing to protect us from foreign image classes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Takeaways\n",
    "\n",
    "- Aleatoric uncertainty can be expressed in any model by directing an output layer to parameterize a distribution describing predictions for an input, rather than predicting the input directly.\n",
    "- TensorFlow Probability lets us do this cleanly and transparently with the `DistributionLambda` layer and the `params_size` helper methods.\n",
    "- Aleatoric uncertainty can uncover edge case inputs that are probabilistically very difficult to predict accuracy with confidence.\n",
    "- Aleatoric uncertainty does _not_ help us guard against inputs outside the context a model was trained on."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}