growing_datasets/Lab - Growing Datasets.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "a6f14f0e",
   "metadata": {},
   "source": [
    "# Lab -- Growing Dataset with Active (Machine) Learning\n",
    "\n",
    "Intended to accompany the lecture on Growing or Compressing Datasets, this notebook contains exercises to build labeled training datasets for binary classification through active learning.\n",
    "\n",
    "You may find it helpful to first install the following dependencies:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b8e1a8c4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: faiss-cpu in /root/anaconda3/lib/python3.11/site-packages (1.8.0.post1)\n",
      "Requirement already satisfied: pyarrow in /root/anaconda3/lib/python3.11/site-packages (17.0.0)\n",
      "Requirement already satisfied: pandas in /root/anaconda3/lib/python3.11/site-packages (2.1.4)\n",
      "Requirement already satisfied: matplotlib in /root/anaconda3/lib/python3.11/site-packages (3.8.0)\n",
      "Requirement already satisfied: scikit-learn in /root/anaconda3/lib/python3.11/site-packages (1.2.2)\n",
      "Requirement already satisfied: torch in /root/anaconda3/lib/python3.11/site-packages (2.4.0)\n",
      "Requirement already satisfied: tqdm in /root/anaconda3/lib/python3.11/site-packages (4.66.4)\n",
      "Requirement already satisfied: numpy<2.0,>=1.0 in /root/anaconda3/lib/python3.11/site-packages (from faiss-cpu) (1.26.4)\n",
      "Requirement already satisfied: packaging in /root/anaconda3/lib/python3.11/site-packages (from faiss-cpu) (23.1)\n",
      "Requirement already satisfied: python-dateutil>=2.8.2 in /root/anaconda3/lib/python3.11/site-packages (from pandas) (2.8.2)\n",
      "Requirement already satisfied: pytz>=2020.1 in /root/anaconda3/lib/python3.11/site-packages (from pandas) (2023.3.post1)\n",
      "Requirement already satisfied: tzdata>=2022.1 in /root/anaconda3/lib/python3.11/site-packages (from pandas) (2023.3)\n",
      "Requirement already satisfied: contourpy>=1.0.1 in /root/anaconda3/lib/python3.11/site-packages (from matplotlib) (1.2.0)\n",
      "Requirement already satisfied: cycler>=0.10 in /root/anaconda3/lib/python3.11/site-packages (from matplotlib) (0.11.0)\n",
      "Requirement already satisfied: fonttools>=4.22.0 in /root/anaconda3/lib/python3.11/site-packages (from matplotlib) (4.25.0)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /root/anaconda3/lib/python3.11/site-packages (from matplotlib) (1.4.4)\n",
      "Requirement already satisfied: pillow>=6.2.0 in /root/anaconda3/lib/python3.11/site-packages (from matplotlib) (10.2.0)\n",
      "Requirement already satisfied: pyparsing>=2.3.1 in /root/anaconda3/lib/python3.11/site-packages (from matplotlib) (3.0.9)\n",
      "Requirement already satisfied: scipy>=1.3.2 in /root/anaconda3/lib/python3.11/site-packages (from scikit-learn) (1.11.4)\n",
      "Requirement already satisfied: joblib>=1.1.1 in /root/anaconda3/lib/python3.11/site-packages (from scikit-learn) (1.2.0)\n",
      "Requirement already satisfied: threadpoolctl>=2.0.0 in /root/anaconda3/lib/python3.11/site-packages (from scikit-learn) (2.2.0)\n",
      "Requirement already satisfied: filelock in /root/anaconda3/lib/python3.11/site-packages (from torch) (3.13.1)\n",
      "Requirement already satisfied: typing-extensions>=4.8.0 in /root/anaconda3/lib/python3.11/site-packages (from torch) (4.9.0)\n",
      "Requirement already satisfied: sympy in /root/anaconda3/lib/python3.11/site-packages (from torch) (1.12)\n",
      "Requirement already satisfied: networkx in /root/anaconda3/lib/python3.11/site-packages (from torch) (3.1)\n",
      "Requirement already satisfied: jinja2 in /root/anaconda3/lib/python3.11/site-packages (from torch) (3.1.3)\n",
      "Requirement already satisfied: fsspec in /root/anaconda3/lib/python3.11/site-packages (from torch) (2023.10.0)\n",
      "Requirement already satisfied: nvidia-cuda-nvrtc-cu12==12.1.105 in /root/anaconda3/lib/python3.11/site-packages (from torch) (12.1.105)\n",
      "Requirement already satisfied: nvidia-cuda-runtime-cu12==12.1.105 in /root/anaconda3/lib/python3.11/site-packages (from torch) (12.1.105)\n",
      "Requirement already satisfied: nvidia-cuda-cupti-cu12==12.1.105 in /root/anaconda3/lib/python3.11/site-packages (from torch) (12.1.105)\n",
      "Requirement already satisfied: nvidia-cudnn-cu12==9.1.0.70 in /root/anaconda3/lib/python3.11/site-packages (from torch) (9.1.0.70)\n",
      "Requirement already satisfied: nvidia-cublas-cu12==12.1.3.1 in /root/anaconda3/lib/python3.11/site-packages (from torch) (12.1.3.1)\n",
      "Requirement already satisfied: nvidia-cufft-cu12==11.0.2.54 in /root/anaconda3/lib/python3.11/site-packages (from torch) (11.0.2.54)\n",
      "Requirement already satisfied: nvidia-curand-cu12==10.3.2.106 in /root/anaconda3/lib/python3.11/site-packages (from torch) (10.3.2.106)\n",
      "Requirement already satisfied: nvidia-cusolver-cu12==11.4.5.107 in /root/anaconda3/lib/python3.11/site-packages (from torch) (11.4.5.107)\n",
      "Requirement already satisfied: nvidia-cusparse-cu12==12.1.0.106 in /root/anaconda3/lib/python3.11/site-packages (from torch) (12.1.0.106)\n",
      "Requirement already satisfied: nvidia-nccl-cu12==2.20.5 in /root/anaconda3/lib/python3.11/site-packages (from torch) (2.20.5)\n",
      "Requirement already satisfied: nvidia-nvtx-cu12==12.1.105 in /root/anaconda3/lib/python3.11/site-packages (from torch) (12.1.105)\n",
      "Requirement already satisfied: triton==3.0.0 in /root/anaconda3/lib/python3.11/site-packages (from torch) (3.0.0)\n",
      "Requirement already satisfied: nvidia-nvjitlink-cu12 in /root/anaconda3/lib/python3.11/site-packages (from nvidia-cusolver-cu12==11.4.5.107->torch) (12.5.82)\n",
      "Requirement already satisfied: six>=1.5 in /root/anaconda3/lib/python3.11/site-packages (from python-dateutil>=2.8.2->pandas) (1.16.0)\n",
      "Requirement already satisfied: MarkupSafe>=2.0 in /root/anaconda3/lib/python3.11/site-packages (from jinja2->torch) (2.1.3)\n",
      "Requirement already satisfied: mpmath>=0.19 in /root/anaconda3/lib/python3.11/site-packages (from sympy->torch) (1.3.0)\n",
      "\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n",
      "\u001b[0mNote: you may need to restart the kernel to use updated packages.\n"
     ]
    }
   ],
   "source": [
    "%pip install faiss-cpu pyarrow pandas matplotlib scikit-learn torch tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "6096ec1a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_51563/1857601726.py:17: DeprecationWarning: Importing display from IPython.core.display is deprecated since IPython 7.14, please import from IPython display\n",
      "  from IPython.core.display import Image, display, clear_output\n",
      "/tmp/ipykernel_51563/1857601726.py:17: DeprecationWarning: Importing clear_output from IPython.core.display is deprecated since IPython 7.14, please import from IPython display\n",
      "  from IPython.core.display import Image, display, clear_output\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import time\n",
    "import random\n",
    "from glob import glob\n",
    "from datetime import datetime\n",
    "from typing import List, Tuple, Optional, Callable\n",
    "\n",
    "import faiss  # Must import faiss before everything else\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from torch.utils.data import Dataset\n",
    "from tqdm.notebook import tqdm\n",
    "from sklearn import linear_model\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.metrics import average_precision_score\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from IPython.core.display import Image, display, clear_output"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9e6fb5ff",
   "metadata": {},
   "source": [
    "# Section 1: 1D Classifier\n",
    "\n",
    "First, let's go through the toy 1D example from lecture."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "e9c0ea0d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(100,)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Generate random data for two classes\n",
    "np.random.seed(2)\n",
    "x1 = np.random.normal(-2, 1, 100)\n",
    "x2 = np.random.normal(2, 1, 100)\n",
    "boundary = 0\n",
    "seed = np.array([(boundary + 2 + (random.random() * 2)),\n",
    "                 (boundary - 2 - (random.random() * 2))])\n",
    "x = np.concatenate([seed, x1, x2])\n",
    "labeled_0 = np.arange(len(seed))\n",
    "\n",
    "\n",
    "# This is a simple case, so labeling is easy and cheap.\n",
    "# But this isn't the case in practice. A real 'label' function would\n",
    "# request information from humans or an expensive process.\n",
    "def label(example):\n",
    "    return int(example > boundary)\n",
    "\n",
    "\n",
    "# Helper functions for plotting\n",
    "def color(label):\n",
    "    return 'r' if label else 'b'\n",
    "\n",
    "\n",
    "def plot_boundary(ax):\n",
    "    ax.axvline(x=boundary, color='gray', linestyle='--')\n",
    "    \n",
    "    \n",
    "def plot_labeled(ax, x, labeled, labels):\n",
    "    ax.scatter(x, np.zeros(len(x)), c='gray', alpha=0.2)\n",
    "    ax.scatter(x[labeled], np.zeros(len(labeled)), c=[color(l) for l in labels], s=200, alpha=0.5)\n",
    "    ax.axvline(x=boundary, color='gray', linestyle='--')\n",
    "    ax.set_axis_off()\n",
    "x1.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9baa7aec",
   "metadata": {},
   "source": [
    "## Passive Learning (Random Sampling)\n",
    "Passive learning simply samples unlabeled examples at random. It doesn't adapt to the model or the data, so we have to hope that we get examples that help us close in on the decision boundary.\n",
    "\n",
    "In the figures below, gray dots represent unlabeled examples, red & blue dots represent the labeled examples for the two classes, and the dashed gray line is the actual decision boundary."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "59f6eef3",
   "metadata": {},
   "outputs": [],
   "source": [
    "def passive_selection(x, labeled, label_func, n):\n",
    "    candidates = set(range(0, len(x))) - set(labeled)\n",
    "    labeled = np.concatenate([labeled, random.sample(list(candidates), n)])\n",
    "    labels = [label_func(example) for example in x[labeled]]\n",
    "    return labeled, labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "6fb3d635",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 0: 2 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 1: 3 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeQAAABhCAYAAAAHpNImAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAcQUlEQVR4nO2d228kx3XGv773dM+F5JDcC/cWyY69SWBHdoIgcR4C+01ADP8JBoy8xbIB+SnQf2AJMKwkLzYQGAiSIHEMRw9B8mAF8YMBI0KsbCRrdVl5tdoLd0kOOZyZnr53HpqntmaWu2IsLtm7+/0AYrnDvlR1nTrfqVNV00ZVVRUIIYQQcqyYx10AQgghhFCQCSGEkEZAQSaEEEIaAAWZEEIIaQAUZEIIIaQBUJAJIYSQBkBBJoQQQhoABZkQQghpABRkQgghpAFQkAkhhJAGQEEmhBBCGgAFmRBCCGkA9nEXgJDHifF4jJdeegkA8Pzzz6Pdbh9ziQghjwocIRNCCCENgIJMCCGENAAKMiGEENIAKMiEEEJIA6AgE0IIIQ2AgkwIIYQ0AG57IuQQMU0Tnuep3wkh5KAYVVVVx10IQggh5EmHITwhhBDSACjIhBBCSAPgHDIhhwi/OpMQ8uvCETIhhBDSACjIhBBCSAOgIBNCCCENgIJMCCGENAAKMiGEENIAKMiEEEJIA+C2J0IOEdM04TiO+p0QQg4KvzqTEEIIaQAM4QkhhJAGQEEmhBBCGgDnkAk5RKIowosvvggA+Na3voUgCI65RISQRwUKMiGHSFmWkGUZZVkec2kIIY8STFk3hKqqfx4HqgqgFhGi8Th1cPLQ4Aj5mKgq4P33gf/9X+Ddd4Hr14GiAIIA+MQngKefBj73OWBx8bhLejBu3ABefx24cgX41a+ALANct67HU08BzzwDnDx53KUk5IgoCuDtt4E33wTeew+4dauOUrvduoP/5m8Cv/u7AN8GRjQoyMfAtWvAP/0TcOkSMBoBvg+EIWCawOYmcPUq8O//DqyuAl/6EvDss7VQN5HNTeBHPwJ+/nNge7sW4XYbsCxgOgV++lPgJz8BlpaAL3wB+MpXHp0gg5Bfi7feAn74Q+Dy5boTBEH9YxjAzZt1BP6v/wqsrdWd+4tfBPb2rpMnGwryEVJVwH/+J/AP/wBsbABnzwIXLtT9dJ6iAG7fBv7u74A33gC+9rX6+Cbx+uvAD35QBxBra8CZM/vXpSyBrS3glVeAX/4S+OpXgd/+7SMuLCEPm6KojfyVV4DxGDh//v4j4Cyrxfl736sj8699rY5ayRMN55CPkFdfBb7/fSCOgd/5HWBhYX8BA+oR5unTwKc+VQvyyy/XaeGm8PrrwF//NbC+Xtdlefn+dTFNYGWlFuFr14C/+qtamAl5bKgq4J//Gfj7vwc8D/it33pwOtpxasF+6ingZz8D/vIvgZ2dIysuaSYU5CPi3Xfrvuq69x8V74fnARcv1tNQP/gBkKYPtZgHYmOjLstoVE+FWdbBzrNt4NOfrkfLf/M3j6f/MU0TlmXBsix+deaTxH/9F/Av/1JHnidPHryDB0HdwX/xizp1xtWQTzT0GEdAltV9bTj89dLOlgV88pPAf/93Pco+TmQgcPVqvTbloH5HMIy6Lu+9B/z4xw+jhMdLEAR44YUX8MILL3AP8pPC7i7wj/9Y/76y8v8/X6L0n/4UeO21Qy0aebQ4ljnkqqqQpinKsoRpmnBdF8acZ58/xnEcZFl233P04w3DQFVVSJIEAOD7PjzPm/k8jmNUVQXTNGHbNoqigG3bsCxr3/LMly1JEkynU2RZBsdx0Gq1Zu6RpimKokCe53jzTQf/8z8O1tZyTKd1BGzbthpBVVWFoijUv0VRAAAsy1LXtO0KrZaJf/u3Ep/97Bh5PkZZlvB9H91uF77vAwDSNEWe54iiCEmSYDQaIc9z2LaNXq+HVquFqqqQZRmqqkJVVcjzXI3q4jhGkiRqpDedTlFVFXzfRxAE2Npq4Wc/a+P0aRN5niFJClUf2YOb7g3jLcuCbdswDANJkiDPc2RZBsuy0G57+I//cPEnf1Lg5Elzpr3q+trI81y15/3acL59y7LEaDTCdDqFYRhwXVdd03VdZFmGPM9VncqynGl3eYbSdvK3B9lfWZYYj8fq+DAMVdn3s1X9eP2++rG6DWVZpv4FAM/z7qnzeDxWbdpqtWCapjpfyi/toj9fqed+NhzHMUajEcqyRKvVQhiGmEwmiKJI1VWOlXt7ngfP81AUBTY3N5UtBUEAx3Hguu49bSn1lmev9/k0Tfftq5ZlqT4r9rWzs4MoipCmKTzPg+u68DxP2Uie5xiPx6iqCt1uF67rqs9kz7ht1y6x1Wqh3W6ra6dpqu4tfQMA8jwHXn0VncuXgYsX4ZSl6gfS5lVVqToZhgHLslTdpd8bYQirKGC8+irw+78PaDaQ5zmm06myn1arBdu2P9JPfVwe1M8e1v0+ShceNsddhiMX5DiOMRgMMJlMVKXDMMTS0pISlflj8jxXDkCETD9HPz6OY+zs7CiHbNs2giDA6uoqFhcXsb29jdu3byunXRSFckatVgu9Xg+Li4sz5Zkv/61bt3Djxg0MBgNkWQbbttHv97G2tobFxUVEUYTt7W3s7u4iTVP8+McrWF9fhG0PlcCIk7QsC3meI01T1fHludi2DcdxEIYhPM9DVeV47TUL3/veL3DixA2UZYkgCLC8vIxz586h3W5jOBzi+vXruH37NobDIdI0Vc7BdV0EQaCESBw2APWcyz2HIr8DtVCIU/3gg8/gypVP4zd+Y4zhsFJOXa4h4qGfp7fh3frb2Nw8gb/92w380R+NldhJeQXTNNFqtdDpdHDixImZNhQnJe0bBAGuXbuG9fV1RFGkHHm73YbneUo0pKyGYaDdbqPX66Hb7aoRbRRFGA6HyLJMCYi0x7z9xXGMDz74ADs7O0robdvGwsKCEkbdVnd2dtTxcRyrgG55eRndbhdhGCIIAmVDm5ub2N7exmQyQZ7n6norKys4c+YMgiDA+vo6NjY2lLCIwxcBkDaRt1BJuxZFgTRNYZomfN+fseHbt2/j6tWr6poSPBZFgclkgul0qq4jgZ0EdXJdERMAKqhqt9vodrvodrvodDpYWFiA7/tKCMVG9KByOp2q+0kbSB8CgNFohM3NTdXfxP5EtDzPU9fU7UqeR5Zlyv6knGEYotVqwff9GWESJ62C5zzHZ195BYuDAdIbN9T9TNNU15TnJAF2q9VSAZ7eP71eD6033oB19SriU6cwGAxw+/Zt5a/E1/R6PayuruLEiRP39VMfF/Fzd+7cQRRFAOpA4MSJEzh16tSh3/MguvCwaUIZjlSQ4zjGzZs3kaapivLyPMfu7i7iOMbp06cBYOaYoijUQwrDECdPnoRpmuqcpaUlDAYD5Vh2d3cxGAxQFAWCIECn00Gapnj//feVMzZNE0VRoCxLRFGEoiiwsLAAy7Kwu7uLoihUefSGEOf74YcfYjKZwLIsdf2trS2Mx2M1kpDOmOcl3n7bgGHsqtGCbduYTqeIokhF3eLExKFKpCbOQkYdk0mADz+scOqUqUZ86+vr2NzcxMLCAvI8x2AwwHA4RJIkqs6GYSCKIkynU+WoJDoHMDNi1r9tSp6XaZqYTCZ4990KcTzEzs4uXNeF4zgoikIFN7qjFIesX0tGdFWVIc8TXLq0i1OnPkAYhgAAx3EwHo8RxzFarRa63S6KokAURbhy5Yq6jud56Ha7qKoK0+kU77zzDkajkRIeGbVlWabqrJ9r27Z6vuIcb926pQIXEanpdIqtrS3lCHX7E7HMsgydTgemaeLGjRu4dOkSAODLX/4yFhcXla2GYYj33nsP0+kUrVZLPYvJZIKiKOB5HqbTKUajEXzfRxRF2Nrawmg0QhzHsG0bnuchSRJsbGyogEuiePnbrVu3UFUVFvf2l0k9AaDT6SCOY4zHY1iWpTIfVVUpGwaA8XgM0zSVfV+/fh2TyUQFidJ38jxXga9lWciyTNmdYRhqJCh2VZYlkiRRoi5Bw82bN1FVFdbW1uA4Du7cuYP19XWYprkXjFaYTCaqvUT0dnZ2lK1nWQbDMJQgS1ZhMpmoUY4IrGRKRGAlEJCsUJ7n2N7eVn1c7yO7u7vqPn6SoL2zg2mrhel4rPqWLuSGYajMlGSK8jxX2QQJWmPHAba2UL31FjaqCoPBAOvr6yoYkZ/d3V0VZO3npz4u4udu3bql6i/+49q1a0iSBBcuXDi0ex5EFx62IDahDMARziFXewaWpim63S4cx4FhGHAcB91uV4na1taWOsa2bYxGI1RVhZWVFdUZbNtGt9tFkiTKQDqdDqIowmg0guu66Pf7AIAkSdT179y5o4RbRjKe5ykBvSsWdccZDAZKTMRhbW5uIs9zuK6LTqcDz/PQ6XTgui4mk4lyoGVZIgxD3L6dIoocBMHdkbGk2uY7qPxdBA2AGkFPJpO9spcYjRbUaFdSfdPpFOvr6xgMBkiSRN1DHIRejzzPEcexev7zqX49cJFzakEzsbERwrbrc2V0LNGknqaT8+RfSTnqKTvXTXHnTgeTyUSVK8syFbTIsTL6kjbMskzZh+M46HQ62N3dxdbWljo3yzL4vo+lpSWVlpTRrYyi+v2+cvSS8k3TFJPJBO12G77vq3pKVkHsr91u4+bNm9je3sbKygo8z1OOX+q7sbGhbDWOY7z11luIoggrKytKAHq9Hvr9PtI0xebmJsqyVKIsAYZkJyQAkiBoe3sbw+FQPdtWq4UkSeC6LlzXVcILQKXRoyhSx5dlqVKQkrEZjUa4ceMGkiTB0tKSsmsAylZkVLffSFxGodJukt4Ve9fPL4oCw+EQm5ubyk4nkwl2d3cxHo/RbrdVwCXCaVmWquN4PFapf7mX/EgwoI9QJQgVIdbT7RJUuK5bi+NemlamQNI0xcLCAgAouzcMA53pFNZ0iqrdVv1AbEgPVMqyVP1QAqT5PuH5PsqiwNbly5hOp8omRSAWFhZg2zZ831eBhozqDustuuLntra24Lqu8tVix47jYDAYYGtr61DueRBdOMz6NbUMwpEJshhpq9Xa9++tVgvD4RDD4VAdk2UZ4jhW80Ce5yGOY9XpHcfB9va2GuFIqkGOdxxHpZpEaGXEJB1HjE0cs23biOMYjuMoEZTyD4dD5UjlHoIIBgDltOs53BiADcO4O+oU8QWgRql6elh+F4dRlqVyJJZVoShmI3q53nQ6VQ5PHJJkA+Rack89bSnoYjxP/TcDRWECuJumFjEXZyTPWa/Tftesxb5AntejKhFMGWnrIwzLshBFkXJi8iwESfWbpjkz3yYjNmE6naqASMrs+74aeUmmQoKDPM+RJAmCIECSJMo25B7ymX6sbhej0Uil+wzDwGAwQBAE6lhJmQJ12l23f93WJCCSe0mQIBkWGe3KfKvMn+oBQpqmqq4iSKZpqqyN2IzUT+xGFxf5EQHUgzh9DYQIoQRpepvL5yLeWZZhOByq4HgymWA4HKrjxZbiOJ4JzmSELe07HxTOr3Cf73t6OaUeUmc5Xq4ptiY2on/mACjzHFlZKj8k/UpsUs/I6X5HnxYSDNNEtLODoigwHo/h+76avwbuBkW+72M0GsEwjBk/9XERPyfrAeaRbIVkZw7jfh+lC4dZv6aWQTgyQRbD11OHOiIsMlcs58gCDjlGj3jFOUjHkt+lM0oHls6gC5wewcvxcoz+r9xLn4PSzxHk/3qZ6uMzWBZQlubMIg85Vh9FymfC/N/rspvw/bufyTH66H5eAPX/7/e3g0d+JUyzQFWZ95TrQfe832dFYcCy8pk20Re06Y5zvza8e51iZtSnz2Hro3P9fCmPTBno19PbvtxztPozFlsQcdCfgW4XellE5OVad7MONeKgZZ5Qn5sXm5b66IGPLn56H9PrLveXsuoCJNfXy6zblT6toduvHhDNn/sg5oM1vd9K++v/nxfN+aBAriHtPS9wDyrHfH8U/6J/Nl9W/ZlbloXKtlEaBrAn1Pr50i56W8wHK/PPy6gq5HvH6Jk08YG6b9OD+YPW+6PQ22M/X60vZjuMex5EFw6zfk0tg3BkgiyGuV9HBjCTYtWFT3fSEp3Op1LFScnv+uhPOvZ89DwfIcv95FrzYi3XB7Bv4+gdUcpRp3kKBEGKJLH3dWzy+7xz2O/vdXlNLCyM1Ge689RTdTrzIj//t4OuIrQsA4uLQySJe0+5HnTP+32WZS76/eFMm0iHF2ejj4ruNwLSHbe+AlZ34vul1cWxz+8Znk9/SkpT/1xS6mJH+n0EvSzi4ORadwM2eRaZSsHrq7ulPLooSV30MunOX188p9dVyirPYb8Mhn5dXRT14+7nrOXcB6FnbuQc6S/S/vr/9WeuBz7y/OUa0t7zgfKDyjHfH8W/6J/Nl1V/5kVRIGq3kXse3CSZub+cNx9MzAeHM8+rqlAZBsrlZWU7+ohe6ihl1acMDlrvj0Jvj/18tb5g7jDueRBdOMz6NbUMqiwP/Q57uK6LMAxnlu/rTKdT9Ho99Ho9dYzjOPB9X22HSZJkZkVmlmVYXFxUCz3CMFRpOPm7bBGRziQrqmVOR1JnhmGoeTaZownDUM3Ruq6LXq+nDFHuIRRFoY7tdrtq0UanE2J1dYDp9K6IicEDmOlUwKxT0cWnds4uDKNEpzNU5dCj6FarpeYP9ZGIiJrcXxcgfb76QeIsny8v76AsbZimpZ6LGKsuFPcbgd19XiUAC0tLQziOg3a7jTzP0Wq11PYrwzDUYjbZNiMiMb9IS+b+bNtWqSd9ZTEAtUhQ6m+aplpsJSvepT0kOPQ8D1EUwfM8ZRtyD/lMP1a3i06no1ZuV1WFpaUllWaWVd9CFEUz9q/bmoi33EsESdLf7XZb2Z/0lzRNEYahaneZUw7DUG1Rk6kXfX5T6id2I/12PzHUAxx5bnrmSURH0EfyMlXkOA56vZ5K44dhiF6vp44XW/J9fyZbIqugpX3nMyD3pILn+t58hkzvK3K8XFNPF8toST6bOg6S5WW0tBXz0q/EJuW6cj3xO/OiZozHKDwP3tNP720NbCOOY7XoDIBa+R/HMTqdDqqqmvFTHxfxc4Zh3OPjAKhFar1e71DueRBdOMz6NbUMwpEJsmEYapGIrBIsy3qbzO7urlqI1e/31THibAzDwMbGBgzDQKfTUavfPM/DuXPn4HkeRqMRgiBAu91WC8SA2lnL9VdXV1UnESeXJAnG47FKJYoD9DwPS0tLM5Fyv9/H8vKyWqE7Go3UXl9xgP1+X624HY/HCMMQTz89AFAgSe6OAiUIkBW/+qhAn9cVwa33gPawuDhCv38baZqqPZdALTYnT57E0tISPM+bmQKQ+UGphywMqapKdXDducqx81MDZVniwoVttNsFxmNXjeL2SwfLefKvjOykjuNxC0GQ4Pz5bYRhqMolK3glJak7YWlDx3GUDWVZhtFohG63i36/r851HEcteJFFWFIHWRy0tbUFwzCU3cg+2TAM1Upvcb6y5UXsbzwe4/Tp01hcXMTGxoaaa9aFYHl5Wdmq7/u4ePEigiBQi71M08RwOMTW1pba+mSad7d5dTodJQCyBUjqbFkWFhcXlfOUYzzPU+sIZJsRALU+IgiCmVG2rIiWFcmdTgdra2vwPE8tdJEV8GIrsqpbT5lWVaVEVuxW7qOnofXzLctCt9tVz0nu1el00G63MR6PYRiGCqBlMZ4EPu12G+12WwUI+jSDPoUgopemqQrm9PSz9H0pg2maakWttLnrutjZ+2o5sfuqqpAXBTYvXgSSBNbes9CDGJl6kLljeQYSXOp9orh+HcWnPoVTf/iHah+04zhq1bdsrZM1LmEYqoWLB81yfRTi52Shod7P5PelpSX0+/1DuedBdOEw69fUMghHuu3J932cPn36nr1e3W53Zq/X/DGyR1RSfkVRzJzj+746vtfroaoqtQ95Op0iCAKcPXsWCwsLag+rvnpVRs2yqvB++5B938f58+fhuq7ahzwajdQ+5DNnzmBhYWFmH3JRFPjc5zy89VaGK1cWsbZWL1iR6F7fhywLc0TcZEVt7RB9GIaNP/7j21hZ6ahFRa1WCysrKzh79qzahyxbKR7GPuRz5zoYDiu89toJ+P4uyjJX0bwYsVxTRmxyvbvXtDCd9vB7v3cDn//8WZWRkDLJSFHubVkWgiC4Zx/yaFSn7qV9W62W2ocsoxUR4/32Ibuui3a7jTAM992HHMcxfN/HwsKCSuXN29/8PmRZCCMjLFm5KfbkeZ46XgSj3W6j3++rFf/nzp1TNiR9RKZyqqpebLO6uoq1tbV99yGfOnVK3d8w7t2HLKMgsTuxN92GZR+y7BhYWVnBysrKzD5kffQvo84gCP5f+5B7vZ4KrAEoO1laWkKr1UIcx2ofsozu9RX43W73nn3Igr4PWYJPQUQRgFrZL8/hQfuQZTW6HjxvnD+PMydOYHUyQba34l6mOvT5x/vtQ87zHNbe7hD/T/8U3uIi/L17W5albF36Ybfbfaj7kHU/d+fOHdXPHtY+5IPqwsOkCWUAAKP6qFUYD4GqerK+qcu2bVy5YuLb3zaR5yXOnCln0lX6fJk+QpYOXJYGfvlL4POfr/CNb+Qoy1g536P+pi7f9zEet/Diiy5u3sxx4UIOwzj4N3WlaYZr11o4e7bCX/wFsLLiqNE5v6mL39T1KH5TV5qmCN54A+H3vw+r30e5t51On3vXp3Pu+aauJIH5zjswn30Wxp/9Wf02Fs2n6TsHpGwSbDzMUduD+tnDut+T/k1dxyLITyo/+Un9UgXbrl/0cpB2ThLgnXeAp58GvvGN+jWHTUDe9jQa1d9pbR3gBRN5Xr9ko98H/vzP6xfiEPJYUFX1S85/+MP6hd8HfcFEFNWd4plngOeeq18BR55YKMhHSLXP+5B7vfu/D3l9vX4z0mc+82i8D7nfv//7kDc3gVu3avHm+5DJY0nxa7wPeTQC/uAP+D5kAoCCfCxcu1YH05cu1f3R94EwrDNVWVb35aIAVleBL30JePbZ+i1tTWRzE/jRj4Cf/xzY3q5fXNNu1yPmPK/rkmW1r/nCF4CvfKUeQDyuRFGE73znOwCAb37zm3zj05PI5ct1B798GZhO684bBHW0miTAZFJH52trdef+4hfr9yOTJx4K8jFRVcCvflWL8pUrtUgXRd1vP/GJ+ueZZx4d8bpxox4xv/9+Xa80rd/l/NRT9c8zz9RZvMed8XiMl156CQDw/PPPq1XO5AmjKIC33wbefLN+1+itW3WqqNer3z8qHXxvBTshwDG9fpHUwbKIFVALtHz+KLK2Nju/XVWPbl0I+dhYVr1IQhZKPOodnBwJFOSG8Lj108etPoR8LNghyAE4si8GIYQQQsj9oSATQgghDYCCTAghhDQACjIhhBDSALjtiRBCCGkAHCETQgghDYCCTAghhDQA7kMm5BCJogjf/e53AQDPPfccvzqTEHJgKMiEHCJlWarXSMor/Qgh5CAwZU0IIYQ0AAoyIYQQ0gAoyIQQQkgDoCATQgghDYCCTAghhDQACjIhhBDSAPjVmYQQQkgD4AiZEEIIaQAUZEIIIaQB8Ju6CDlE4jjGyy+/DAD4+te/Dt/3j7lEhJBHBQoyIYdInueIokj9TgghB4Upa0IIIaQBUJAJIYSQBkBBJoQQQhoABZkQQghpABRkQgghpAFQkAkhhJAGwK/OJIQQQhoAR8iEEEJIA6AgE0IIIQ2AgkwIIYQ0AAoyIYQQ0gAoyIQQQkgDoCATQgghDYCCTAghhDQACjIhhBDSACjIhBBCSAP4P+ZB5hcIcboXAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 2: 4 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 3: 5 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 4: 6 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 5: 7 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 6: 8 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 7: 9 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "n = 8\n",
    "labeled, labels = passive_selection(x, labeled_0, label, n)\n",
    "for i in range(n):\n",
    "    print(f\"Round {i}: {2+i} labeled examples\")\n",
    "    fig, ax = plt.subplots(1, 1, figsize=(6, 1))\n",
    "    plot_labeled(ax, x, labeled[:2+i], labels[:2+i])\n",
    "    plot_boundary(ax)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "685512fc",
   "metadata": {},
   "source": [
    "## Exercise 1: Active Learning (Max Entropy)\n",
    "Active learning trains a model $A_r$ after every example $\\mathbf{x}$ (or batch of examples) is labeled and leverages that model to select the following example. For instance, we can quantify the model's uncertainty by calculating the entropy for each of the pool of unlabeled examples $\\mathcal{P}_r$ based on the model's predicted probabilities:\n",
    "\n",
    "$$\\phi_{\\text{Ent}}(\\mathbf{x}, A_r) = - \\sum_{\\hat{y}} P(\\hat{y}|\\mathbf{x}; A_r) \\log P(\\hat{y}|\\mathbf{x}; A_r)$$\n",
    "\n",
    "and select the examples with the highest entropy:\n",
    "\n",
    "$$argmax_{\\mathbf{x} \\in \\mathcal{P}_r} \\phi_{\\text{Ent}}(\\mathbf{x}, A_r, \\mathcal{P}_r)$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "bbef1f1d",
   "metadata": {},
   "outputs": [],
   "source": [
    "def select_max_entropy(x, model, indices: List[int], budget: int):\n",
    "    # Get prediction probability for elegible indices\n",
    "    proba = model.predict_proba(x[indices].reshape(-1, 1))\n",
    "\n",
    "    # Calculate information entropy from probabilities\n",
    "    entropy = -1.0 * (np.log(proba) * proba).sum(axis=1)\n",
    "\n",
    "    # Select indices with highest entropy (i.e. MaxEnt)\n",
    "    selected = entropy.argsort(axis=0)[::-1][:budget]\n",
    "\n",
    "    # Convert to original index\n",
    "    selected = np.array(indices)[selected]\n",
    "\n",
    "    return selected"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a294044",
   "metadata": {},
   "source": [
    "In this exercise, you'll leverage this max entropy heuristic to actively select examples to label! You should implement the active learning loop from lecture with a `linear_model.LogisticRegression()` as the classifier. This should effectively perform binary search to find the decision boundary. You just need to fill in the section below marked `YOUR CODE HERE`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "a022fe72",
   "metadata": {},
   "outputs": [],
   "source": [
    "def active_selection(x, labeled, label_func, n):\n",
    "    labels = [label_func(example) for example in x[labeled]]\n",
    "    candidates = set(np.arange(len(x))) - set(labeled)\n",
    "    # YOUR CODE HERE\n",
    "    pass"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bad1afa0",
   "metadata": {},
   "source": [
    "Let's put your active selection process into action! In the figures below, gray dots represent unlabeled examples, red & blue dots represent the labeled examples for the two classes, and the dashed gray line is the actual decision boundary. Rounds 1+ will show the results of your active learning implementation. How does it compare to the passive approach?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "c69e16eb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 0: 2 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 1: 3 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 2: 4 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 3: 5 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 4: 6 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 5: 7 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 6: 8 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 7: 9 labeled examples\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x100 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "n = 8\n",
    "labeled, labels = active_selection(x, labeled_0, label, n)\n",
    "for i in range(n):\n",
    "    print(f\"Round {i}: {2+i} labeled examples\")\n",
    "    fig, ax = plt.subplots(1, 1, figsize=(6, 1))\n",
    "    plot_labeled(ax, x, labeled[:2+i], labels[:2+i])\n",
    "    plot_boundary(ax)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "145c5e27",
   "metadata": {},
   "source": [
    "# Section 2: OpenImages"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a36cd68e",
   "metadata": {},
   "source": [
    "Let's scale up active learning to a real dataset, [OpenImages](https://storage.googleapis.com/openimages/web/index.html)! The OpenImages dataset has over 8 million images from Flickr and captures tens of thousands of classes. To make this manageable for our lab assignment, we will train a `LogisticRegression()` classifier on top of pre-computed embeddings for the available images. Please download and unpack the embeddings.\n",
    "\n",
    "1. [Download embeddings](https://drive.google.com/file/d/1HLEI7OWDGfP-__w6vlnZjgOCo5fshP-7/view?usp=sharing) (~18 GB)\n",
    "2. Unpack embeddings: `tar -xvf data.tar.gz`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ce4bc30a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Assuming you unpacked the data in the same directory\n",
    "# as this notebook, you shouldn't need to change this.\n",
    "metadata_dir='data/saved_metadata/'\n",
    "train_dir='data/saved_embeddings/train/'\n",
    "test_dir='data/saved_embeddings/test/'"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17d5e305",
   "metadata": {},
   "source": [
    "## Loading Data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "80e847a2",
   "metadata": {},
   "source": [
    "Don't worry about understanding all of this. You'll only use a small part of this class in the exercise below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "f8ddbcb8",
   "metadata": {},
   "outputs": [],
   "source": [
    "class OpenImagesDataset(Dataset):\n",
    "    def __init__(self,\n",
    "                 metadata_dir: str,\n",
    "                 embeddings_dir: str,\n",
    "                 embeddings_include: str = '*.parquet.gzip',\n",
    "                 target_class: str = None,\n",
    "                 split: str ='train'):\n",
    "\n",
    "        # Save attributes\n",
    "        self._metadata_dir = metadata_dir\n",
    "        self._embeddings_dir = embeddings_dir\n",
    "        self._embeddings_include = embeddings_include\n",
    "        self._target_class = None\n",
    "\n",
    "        # Define paths of metadata files\n",
    "        assert split in ['train', 'test'], 'Invalid value for split'\n",
    "        label_path = os.path.join(metadata_dir, 'oidv6-class-descriptions.csv')\n",
    "        data_path = None\n",
    "        url_path = None\n",
    "\n",
    "        if split == 'train':\n",
    "            data_path = os.path.join(\n",
    "                metadata_dir,\n",
    "                'oidv6-train-annotations-human-imagelabels.csv')\n",
    "            url_path = os.path.join(\n",
    "                metadata_dir,\n",
    "                'oidv6-train-images-with-labels-with-rotation.csv')\n",
    "\n",
    "        elif split == 'test':\n",
    "            data_path = os.path.join(\n",
    "                metadata_dir,\n",
    "                'test-annotations-human-imagelabels.csv')\n",
    "            url_path = os.path.join(\n",
    "                metadata_dir,\n",
    "                'test-images-with-rotation.csv')\n",
    "\n",
    "        # Load metadata\n",
    "        print('Loading metadata...', end=' ')\n",
    "        self._label_df = pd.read_csv(label_path)\n",
    "\n",
    "        self._data_df = pd.read_csv(data_path)\n",
    "        self._data_df.drop('Source', axis=1, inplace=True)\n",
    "        self._data_df.set_index('ImageID', inplace=True)\n",
    "\n",
    "        self._url_df = pd.read_csv(url_path)\n",
    "        self._url_df = self._url_df[['ImageID', 'OriginalURL']]\n",
    "        self._url_df.set_index('ImageID', inplace=True)\n",
    "        print('Done.')\n",
    "\n",
    "        # Load embeddings\n",
    "        self._embeddings_df = self._load_embeddings_df()\n",
    "\n",
    "        # (optional) Load labels\n",
    "        if target_class is not None:\n",
    "            self.target_class = target_class\n",
    "\n",
    "    def to_numpy(self):\n",
    "        \"\"\"\n",
    "        Returns embeddings and labels\n",
    "        Important: does not make a copy of the data in memory\n",
    "        \"\"\"\n",
    "        return self.embeddings, self.labels\n",
    "\n",
    "    @property\n",
    "    def target_class(self):\n",
    "        return self._target_class\n",
    "\n",
    "    @property\n",
    "    def urls(self):\n",
    "        return self._embeddings_df.index.values\n",
    "\n",
    "    @property\n",
    "    def labels(self):\n",
    "        return self._labels_df.Confidence.values\n",
    "\n",
    "    @property\n",
    "    def index(self):\n",
    "        return self._embeddings_df.index.values\n",
    "\n",
    "    @property\n",
    "    def embeddings(self):\n",
    "        return self._embeddings_df.values\n",
    "\n",
    "    @property\n",
    "    def embeddings_dimension(self):\n",
    "        return self._embeddings_df.shape[1]\n",
    "\n",
    "    @target_class.setter\n",
    "    def target_class(self, target_class):\n",
    "        self._labels_df = self._load_labels_df(target_class)\n",
    "        self._target_class = target_class\n",
    "\n",
    "    def _load_labels_df(self, label: str):\n",
    "        \"\"\"\n",
    "        Loads metadata in metadata_dir for target_class\n",
    "        \"\"\"\n",
    "        # Get label name\n",
    "        label_name = self._label_df.query(f\"DisplayName == '{label}'\")[\n",
    "            'LabelName'].to_list()[0]\n",
    "\n",
    "        # Get labeled data given label\n",
    "        df = self._data_df.query(f\"LabelName == '{label_name}'\")\n",
    "\n",
    "        # Join with URL data\n",
    "        df = df.join(self._url_df, on='ImageID', how='left')\n",
    "\n",
    "        # Rename columns and reset index\n",
    "        df.reset_index(inplace=True)\n",
    "        df.rename(columns={'OriginalURL': 'img_path'}, inplace=True)\n",
    "        df.set_index('img_path', inplace=True)\n",
    "\n",
    "        # Align label index with embeddings index\n",
    "        df = df.reindex(self._embeddings_df.index, fill_value=0)\n",
    "\n",
    "        return df\n",
    "\n",
    "    def _load_embeddings_df(self):\n",
    "        \"\"\"\n",
    "        Loads embeddings from partitions in embeddings_dir with specified\n",
    "        pattern. Each embedding value is loaded into a separate column\n",
    "        such that embeddings.values returns the array necessary for\n",
    "        inference, which avoids creating a copy in memory of the data\n",
    "        \"\"\"\n",
    "        # Get paths of dfs in embeddings_dir with pattern\n",
    "        pattern = os.path.join(self._embeddings_dir,\n",
    "                               self._embeddings_include)\n",
    "        filepaths_ls = glob(pattern, recursive=True)\n",
    "        filepaths_ls.sort()\n",
    "\n",
    "        # Store all embedding values and indices\n",
    "        index_ls = []\n",
    "        values_ls = []\n",
    "        wrapped_loader = tqdm(filepaths_ls,\n",
    "                              desc=\"Loading embeddings\",\n",
    "                              unit=\"partition\")\n",
    "        for filepath in wrapped_loader:\n",
    "\n",
    "            df = pd.read_parquet(filepath)\n",
    "            index_ls.append(df.index)\n",
    "            values_ls.append(df['embedding'].tolist())\n",
    "\n",
    "        # Join into single df and return\n",
    "        return pd.DataFrame(\n",
    "            np.concatenate(values_ls, axis=0, dtype=np.float32),\n",
    "            index=np.concatenate(index_ls, axis=0))\n",
    "\n",
    "    \"\"\"\n",
    "    Necessary for pytorch\n",
    "    \"\"\"\n",
    "    def __len__(self):\n",
    "        return len(self._embeddings_df)\n",
    "\n",
    "    def __getitem__(self, idx: int):\n",
    "        return (self._embeddings_df.iloc[idx].values,\n",
    "                self._labels_df.Confidence.iloc[idx])\n",
    "\n",
    "    \n",
    "def create_dataset(metadata_dir: str, embeddings_dir: str, split: str):\n",
    "    return OpenImagesDataset(metadata_dir, embeddings_dir, split=split)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "ea311556",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading metadata... Done.\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a964b192f7cf46cb989066ea74a0212b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Loading embeddings:   0%|          | 0/16 [00:00<?, ?partition/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of embeddings for test images: (113508, 256)\n"
     ]
    }
   ],
   "source": [
    "# Create test dataset object\n",
    "if 'd_test' not in locals():\n",
    "    d_test = create_dataset(metadata_dir, test_dir, 'test')\n",
    "print(\"Shape of embeddings for test images: {}\".format(d_test.embeddings.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "ebe209cf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading metadata... Done.\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "067d8d8c660644c9b56b03e0c0f96853",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Loading embeddings:   0%|          | 0/94 [00:00<?, ?partition/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of embeddings for training images: (8121242, 256)\n"
     ]
    }
   ],
   "source": [
    "# Create training dataset object\n",
    "if 'd_train' not in locals():\n",
    "    d_train = create_dataset(metadata_dir, train_dir, 'train')\n",
    "print(\"Shape of embeddings for training images: {}\".format(d_train.embeddings.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c87dbcd7",
   "metadata": {},
   "source": [
    "## Helper functions for active learning\n",
    "Don't worry about understanding this to complete the exercise."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "a110320d",
   "metadata": {},
   "outputs": [],
   "source": [
    "def set_random_seed(seed: Optional[int] = None):\n",
    "    seed = np.random.randint(10000) if seed is None else seed\n",
    "    np.random.seed(seed)\n",
    "    return seed\n",
    "\n",
    "\n",
    "def create_seed(labels: np.array,\n",
    "                emb_dimension: int,\n",
    "                npositive: int = 5,\n",
    "                nnegative: int = 95,\n",
    "                random_seed: int = 400):\n",
    "    # Make sure there are enough negative examples\n",
    "    nnegative = max(nnegative, emb_dimension - npositive)\n",
    "\n",
    "    # Select indices for random samples\n",
    "    pos_samples = np.random.choice(\n",
    "        np.where(labels == 1)[0], npositive, replace=False)\n",
    "    neg_samples = np.random.choice(\n",
    "        np.where(labels == 0)[0], nnegative, replace=False)\n",
    "\n",
    "    # Combine and sort\n",
    "    seed = np.sort(np.concatenate((pos_samples, neg_samples)))\n",
    "\n",
    "    return seed\n",
    "\n",
    "\n",
    "def visualize_urls(urls: List[str], message: str = ''):\n",
    "    print(message)\n",
    "    [display(Image(url=url, width=400)) for url in urls]\n",
    "    input('Press any key to continue')\n",
    "    clear_output()\n",
    "\n",
    "    \n",
    "def train_model(embeddings: np.array, labels: np.array):\n",
    "    return LogisticRegression().fit(embeddings, labels)\n",
    "\n",
    "\n",
    "def eval_model(model, dataset):\n",
    "    proba = model.predict_proba(dataset.embeddings)[:, 1]\n",
    "    return average_precision_score(dataset.labels, proba)\n",
    "\n",
    "\n",
    "def collect_labels(dataset, indices: np.array, target_class: str):\n",
    "    labels = []\n",
    "    for url in dataset.urls[indices]:\n",
    "\n",
    "        # Show link and image\n",
    "        clear_output()\n",
    "        print(f'Link: {url}')\n",
    "        display(Image(url=url, width=400))\n",
    "        time.sleep(0.25)\n",
    "\n",
    "        # Get user label\n",
    "        need_input = True\n",
    "        while need_input:\n",
    "            label = input(f'Is this an example of {target_class}? [Y/n] ')\n",
    "            label = 'Y' if label is None or label == '' else label[0].upper()\n",
    "\n",
    "            if label not in ['N', 'Y']:\n",
    "                print('Invalid input')\n",
    "                continue\n",
    "\n",
    "            labels.append(1 if label == 'Y' else 0)\n",
    "            need_input = False\n",
    "\n",
    "    clear_output()\n",
    "    print(f'Finished labeling {len(labels)} images')\n",
    "    return labels\n",
    "\n",
    "\n",
    "def visualize_scores(scores: List[float], ylabel: str = 'Average precision'):\n",
    "    plt.title('Model score for test data')\n",
    "    plt.xlabel('Active learning round (#)')\n",
    "    plt.ylabel(ylabel)\n",
    "\n",
    "    plt.plot(range(len(scores)), scores, 'o-', color='b')\n",
    "    plt.xticks(range(len(scores)))\n",
    "    plt.legend(loc=\"best\")\n",
    "    plt.show()\n",
    "    \n",
    "def select_maxent(model, dataset, indices: List[int], budget: int):\n",
    "    # Get prediction probability for elegible indices\n",
    "    proba = model.predict_proba(dataset.embeddings[indices])\n",
    "\n",
    "    # Calculate information entropy from probabilities\n",
    "    entropy = -1.0 * (np.log(proba) * proba).sum(axis=1)\n",
    "\n",
    "    # Select indices with highest entropy (i.e. MaxEnt)\n",
    "    selected = entropy.argsort(axis=0)[::-1][:budget]\n",
    "\n",
    "    # Convert to original index\n",
    "    selected = np.array(indices)[selected]\n",
    "\n",
    "    return selected"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17c30ba8",
   "metadata": {},
   "source": [
    "## Baseline approach for reference\n",
    "You might want to skip this if you're in a rush ;)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "0a067eb6",
   "metadata": {},
   "outputs": [],
   "source": [
    "def baseline_selection(\n",
    "    train, test, concept: str, rounds: Tuple[int],\n",
    "    npos: int = 5, nneg: int = 95, k: int = 100,\n",
    "    select: Callable = select_maxent):\n",
    "\n",
    "    # Set initial values\n",
    "    labeled = create_seed(train.labels, train.embeddings_dimension, npos, nneg)\n",
    "    new_labeled = labeled\n",
    "    labels = train.labels[labeled]\n",
    "    scores = []\n",
    "    model = None\n",
    "    candidates = set(np.arange(len(train))) - set(labeled)\n",
    "\n",
    "    # Visualize seed\n",
    "    visualize_urls(train.urls[labeled[np.where(labels == 1)]],\n",
    "                   message=f'Initial positive samples for {concept}')\n",
    "\n",
    "    # Main active learning loop\n",
    "    for budget in rounds:\n",
    "        # Train model using labeled samples and score\n",
    "        model = train_model(train.embeddings[labeled], labels)\n",
    "        scores.append(eval_model(model, test))\n",
    "\n",
    "        # Select points to label (MaxEnt for now)\n",
    "        new_labeled = select(model, train, list(candidates), budget)\n",
    "\n",
    "        # User labels selected points\n",
    "        new_labels = collect_labels(train, new_labeled, concept)\n",
    "\n",
    "        # Update arrays\n",
    "        labeled = np.concatenate((labeled, new_labeled))\n",
    "        labels = np.concatenate((labels, new_labels))\n",
    "        candidates -= set(new_labeled)\n",
    "\n",
    "    # Train and score final iteration of model\n",
    "    model = train_model(train.embeddings[labeled], labels)\n",
    "    scores.append(eval_model(model, test))\n",
    "    return model, scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "78bde7f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set rare concept (e.g. Bowling, Sushi, Rat, Beach, Hawk, Cupcake, Crowd, etc)\n",
    "concept = 'Bowling'\n",
    "\n",
    "# Set concept\n",
    "d_train.target_class = concept\n",
    "d_test.target_class = concept"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "id": "676a8935",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished labeling 5 images\n",
      "Total time: 0:09:12.237339\n"
     ]
    }
   ],
   "source": [
    "np.random.seed(400)  # for reproducibility\n",
    "start = datetime.now()\n",
    "model, scores = baseline_selection(\n",
    "    train=d_train, test=d_test, concept=concept, npos=5, nneg=300,\n",
    "    rounds=[5, 5, 5], select=select_maxent)\n",
    "print(\"Total time: {}\".format(datetime.now() - start))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "id": "b9beab7f",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "No artists with labels found to put in legend.  Note that artists whose label start with an underscore are ignored when legend() is called with no argument.\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_scores(scores)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d75a1506",
   "metadata": {},
   "source": [
    "## Similarity Search for Efficient Active Learning and Search (SEALS)\n",
    "That was really slow. Let's speed up the process by implementing the [SEALS](https://aaai-2022.virtualchair.net/poster_aaai12967) approach from lecture.\n",
    "\n",
    "First, we need to create a similarity search index:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "4b7bb423",
   "metadata": {},
   "outputs": [],
   "source": [
    "class IndexWrapper():\n",
    "\n",
    "    def __init__(self, dataset, method='LSH'):\n",
    "        assert method in ['LSH', 'flatL2'], 'Invalid index method'\n",
    "        self._dataset = dataset\n",
    "        self._method = method\n",
    "        self._index = self._create_index()\n",
    "\n",
    "    def __len__(self):\n",
    "        return self._index.ntotal\n",
    "\n",
    "    def __getitem__(self, idx: int):\n",
    "        # Return url and corresponding embedding\n",
    "        return self._dataset.index[idx], self._dataset.embeddings[idx]\n",
    "\n",
    "    @property\n",
    "    def d(self):\n",
    "        return self._index.d\n",
    "\n",
    "    @property\n",
    "    def ntotal(self):\n",
    "        return self._index.ntotal\n",
    "\n",
    "    def _create_index(self):\n",
    "        \"\"\"\n",
    "        Returns FAISS index created from dataset.embeddings using the\n",
    "        specified index method\n",
    "        \"\"\"\n",
    "        # Initialize index\n",
    "        index = None\n",
    "        d = self._dataset.embeddings_dimension\n",
    "\n",
    "        if self._method == 'flatL2':\n",
    "            index = faiss.IndexFlatL2(d)\n",
    "\n",
    "        elif self._method == 'LSH':\n",
    "            n_bits = 2 * d\n",
    "            index = faiss.IndexLSH(d, n_bits)\n",
    "            index.train(self._dataset.embeddings)\n",
    "\n",
    "        # Add embeddings\n",
    "        index.add(self._dataset.embeddings)\n",
    "\n",
    "        return index\n",
    "\n",
    "    def search(self, q: np.array, k: int, attrs=None):\n",
    "        \"\"\"\n",
    "        Returns tuple, where each element is an attribute array specified\n",
    "        in attrs. Valid attributes include 'distances', 'urls', 'indices', and\n",
    "        'embeddings'\n",
    "        \"\"\"\n",
    "        # If no attributes selected, use default\n",
    "        default_attrs = ('indices', 'distances')\n",
    "        attrs = default_attrs if attrs is None else attrs\n",
    "\n",
    "        # Initialize results dictionary\n",
    "        results = {}\n",
    "\n",
    "        # Convert query to 2D np array for FAISS index\n",
    "        q = q if len(q.shape) == 2 else np.array([q])\n",
    "\n",
    "        # Search k-NN of q using the index\n",
    "        nn_distances, nn_indices = self._index.search(q, k)\n",
    "        results['distances'], results['indices'] = nn_distances, nn_indices\n",
    "\n",
    "        # Get other attributes\n",
    "        other_attrs = [attr for attr in attrs if attr not in default_attrs]\n",
    "        for attr in other_attrs:\n",
    "            assert hasattr(self._dataset, attr), 'Invalid attribute'\n",
    "            results[attr] = [\n",
    "                getattr(self._dataset, attr)[idx] for idx in results['indices']]\n",
    "\n",
    "        # Return results in order given by attrs\n",
    "        return tuple([results[attr] for attr in attrs])\n",
    "    \n",
    "    \n",
    "def create_index(dataset, method: str = 'LSH'):\n",
    "    return IndexWrapper(dataset, method=method)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "6534d729",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create IndexWrapper for training data\n",
    "if 'index_train' not in locals():\n",
    "    index_train = create_index(d_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d9634a77",
   "metadata": {},
   "source": [
    "## Exercise 2: Manage Candidate Pool\n",
    "Now that we have a similarity search index let's use it to limit the candidate pool we select unlabeled examples from. We don't need to look at all the unlabeled examples from the beginning. Instead, we could start with the nearest neighbors to our currently labeled examples and expand the pool. Using the baseline approach as a starting point, fill in the sections of the code labeled `YOUR CODE HERE`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "0c47d19c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def SEALS(\n",
    "    train, test, concept: str, rounds: Tuple[int],\n",
    "    npos: int = 5, nneg: int = 95, k: int = 100,\n",
    "    select: Callable = select_maxent, knn = None):\n",
    "\n",
    "    # Set initial values\n",
    "    labeled = create_seed(train.labels, train.embeddings_dimension, npos, nneg)\n",
    "    new_labeled = labeled\n",
    "    labels = train.labels[labeled]\n",
    "    scores = []\n",
    "    model = None\n",
    "    candidates = set(np.arange(len(train))) - set(labeled)\n",
    "    \n",
    "    if knn is not None:\n",
    "        # YOUR CODE HERE\n",
    "        pass\n",
    "\n",
    "    # Visualize seed\n",
    "    visualize_urls(train.urls[labeled[np.where(labels == 1)]],\n",
    "                   message=f'Initial positive samples for {concept}')\n",
    "\n",
    "    # Main active learning loop\n",
    "    for budget in rounds:\n",
    "        if knn is not None:\n",
    "            # YOUR CODE HERE\n",
    "            pass\n",
    "            \n",
    "        # Train model using labeled samples and score\n",
    "        model = train_model(train.embeddings[labeled], labels)\n",
    "        scores.append(eval_model(model, test))\n",
    "\n",
    "        # Select points to label (MaxEnt for now)\n",
    "        new_labeled = select(model, train, list(candidates), budget)\n",
    "\n",
    "        # User labels selected points\n",
    "        new_labels = collect_labels(train, new_labeled, concept)\n",
    "\n",
    "        # Update arrays\n",
    "        labeled = np.concatenate((labeled, new_labeled))\n",
    "        labels = np.concatenate((labels, new_labels))\n",
    "        candidates -= set(new_labeled)\n",
    "\n",
    "    # Train and score final iteration of model\n",
    "    model = train_model(train.embeddings[labeled], labels)\n",
    "    scores.append(eval_model(model, test))\n",
    "    return model, scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "b4ce9817",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished labeling 5 images\n",
      "Total time: 0:00:46.942903\n"
     ]
    }
   ],
   "source": [
    "np.random.seed(400)  # for reproducibility\n",
    "start = datetime.now()\n",
    "model, scores = SEALS(\n",
    "    train=d_train, test=d_test, concept=concept, npos=5, nneg=300,\n",
    "    rounds=[5, 5, 5], select=select_maxent, knn=index_train)\n",
    "print(\"Total time: {}\".format(datetime.now() - start))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "1dcc53df",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "No artists with labels found to put in legend.  Note that artists whose label start with an underscore are ignored when legend() is called with no argument.\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_scores(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "283a65a0",
   "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": "3.11.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}