scenarios/similarity/11_exploring_hyperparameters.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<i>Copyright (c) Microsoft Corporation. All rights reserved.</i>\n",
    "\n",
    "<i>Licensed under the MIT License.</i>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Testing different Hyperparameters and Benchmarking"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this notebook, we'll cover how to test different hyperparameters for a particular dataset and how to benchmark different parameters across a group of datasets. Note that this re-uses functionality which was already introduced and described in the [classification/notebooks/11_exploring_hyperparameters.ipynb](../../classification/notebooks/11_exploring_hyperparameters.ipynb) notebook. **Please refer to that notebook for all explanations, which this notebook will not repeat.**\n",
    "\n",
    "For an example of how to scale up with remote GPU clusters on Azure Machine Learning, please view [24_exploring_hyperparameters_on_azureml.ipynb](../../classification/notebooks/24_exploring_hyperparameters_on_azureml.ipynb)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Testing hyperparameters"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ensure edits to libraries are loaded and plotting is shown in the notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We start by importing the utilities we need."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.0.48'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import sys\n",
    "import numpy as np\n",
    "import scrapbook as sb\n",
    "import torch\n",
    "import fastai\n",
    "from fastai.vision import DatasetType\n",
    "\n",
    "sys.path.append(\"../../\")\n",
    "from utils_cv.classification.data import Urls\n",
    "from utils_cv.common.data import unzip_url\n",
    "from utils_cv.classification.parameter_sweeper import ParameterSweeper, clean_sweeper_df, plot_sweeper_df\n",
    "from utils_cv.similarity.data import comparative_set_builder\n",
    "from utils_cv.similarity.metrics import positive_image_ranks\n",
    "from utils_cv.similarity.model import compute_features_learner\n",
    "\n",
    "fastai.__version__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define the datasets and parameters we will use in this notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "tags": [
     "parameters"
    ]
   },
   "outputs": [],
   "source": [
    "DATA_PATHS = [unzip_url(Urls.fridge_objects_path, exist_ok=True), unzip_url(Urls.fridge_objects_watermark_path, exist_ok=True)]\n",
    "REPS = 3\n",
    "LEARNING_RATES = [1e-3, 1e-4, 1e-5]\n",
    "IM_SIZES = [300, 500]\n",
    "EPOCHS = [16]\n",
    "DROPOUTS = [0]  #Leave dropout at zero. Higher values tend to perform significantly worse"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Similiarity accuracy metric\n",
    "\n",
    "For image classification, we used the percentage of correctly labeled images to measure accuracy. For image retrieval, our measure is the rank of the positive example among a large number of negatives. This was described in the [01_training_and_evaluation_introduction.ipynb](01_training_and_evaluation_introduction.ipynb) notebook, and we will re-use some of the code from that notebook in the definition of the _retrieval_rank()_ function below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def retrieval_rank(learn):\n",
    "    data = learn.data\n",
    "\n",
    "    # Build multiple sets of comparative images from the validation images\n",
    "    comparative_sets = comparative_set_builder(\n",
    "        data.valid_ds, num_sets=1000, num_negatives=99\n",
    "    )\n",
    "\n",
    "    # Use penultimate layer as image representation\n",
    "    embedding_layer = learn.model[1][-2]\n",
    "        \n",
    "    # Compute DNN features for all validation images\n",
    "    valid_features = compute_features_learner(\n",
    "        data, DatasetType.Valid, learn, embedding_layer\n",
    "    )\n",
    "    assert len(list(valid_features.values())[0]) == 512\n",
    "\n",
    "    # For each comparative set compute the distances between the query image and all reference images\n",
    "    for cs in comparative_sets:\n",
    "        cs.compute_distances(valid_features)\n",
    "\n",
    "    # Compute the median rank of the positive example over all comparative sets\n",
    "    ranks = positive_image_ranks(comparative_sets)\n",
    "    median_rank = np.median(ranks)\n",
    "    return median_rank"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Using Python <a name=\"python\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We start by creating the Parameter Sweeper object. Before we start testing, it's a good idea to see what the default parameters are. We can use a the property `parameters` to easily see those default values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('learning_rate', [0.0001]),\n",
       "             ('epochs', [15]),\n",
       "             ('batch_size', [16]),\n",
       "             ('im_size', [299]),\n",
       "             ('architecture',\n",
       "              [<Architecture.resnet18: functools.partial(<function resnet18 at 0x7f443ed99598>)>]),\n",
       "             ('transform', [True]),\n",
       "             ('dropout', [0.5]),\n",
       "             ('weight_decay', [0.01]),\n",
       "             ('training_schedule',\n",
       "              [<TrainingSchedule.head_first_then_body: 'head_first_then_body'>]),\n",
       "             ('discriminative_lr', [False]),\n",
       "             ('one_cycle_policy', [True])])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sweeper = ParameterSweeper(metric_name=\"rank\")\n",
    "sweeper.parameters"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we know the defaults, we can pass it the parameters we want to test, and run the parameter sweep."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "this Learner object self-destroyed - it still exists, but no longer usable\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>duration</th>\n",
       "      <th>rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"12\" valign=\"top\">0</th>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.0001]|[epochs: 16]|[batch_size: 16]|[im_size: 300]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>27.527227</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>29.327206</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.0001]|[epochs: 16]|[batch_size: 16]|[im_size: 500]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>46.494100</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>40.436677</td>\n",
       "      <td>9.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.001]|[epochs: 16]|[batch_size: 16]|[im_size: 300]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>29.745073</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>28.454158</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.001]|[epochs: 16]|[batch_size: 16]|[im_size: 500]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>44.277393</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>40.866518</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|[batch_size: 16]|[im_size: 300]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>28.009722</td>\n",
       "      <td>20.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>29.721222</td>\n",
       "      <td>22.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|[batch_size: 16]|[im_size: 500]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>40.376158</td>\n",
       "      <td>25.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>42.627545</td>\n",
       "      <td>34.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"12\" valign=\"top\">1</th>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.0001]|[epochs: 16]|[batch_size: 16]|[im_size: 300]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>30.931857</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>26.125927</td>\n",
       "      <td>8.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.0001]|[epochs: 16]|[batch_size: 16]|[im_size: 500]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>46.117437</td>\n",
       "      <td>11.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>40.555442</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.001]|[epochs: 16]|[batch_size: 16]|[im_size: 300]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>29.870988</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>25.864497</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.001]|[epochs: 16]|[batch_size: 16]|[im_size: 500]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>46.807896</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>41.351353</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|[batch_size: 16]|[im_size: 300]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>25.873023</td>\n",
       "      <td>26.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>25.889981</td>\n",
       "      <td>26.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|[batch_size: 16]|[im_size: 500]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>41.558083</td>\n",
       "      <td>23.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>41.196609</td>\n",
       "      <td>30.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"12\" valign=\"top\">2</th>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.0001]|[epochs: 16]|[batch_size: 16]|[im_size: 300]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>25.954923</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>25.766089</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.0001]|[epochs: 16]|[batch_size: 16]|[im_size: 500]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>40.162561</td>\n",
       "      <td>9.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>41.274331</td>\n",
       "      <td>9.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.001]|[epochs: 16]|[batch_size: 16]|[im_size: 300]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>26.026493</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>25.616917</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 0.001]|[epochs: 16]|[batch_size: 16]|[im_size: 500]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>40.691592</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>40.468641</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|[batch_size: 16]|[im_size: 300]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>26.239308</td>\n",
       "      <td>19.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>26.347744</td>\n",
       "      <td>23.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|[batch_size: 16]|[im_size: 500]|[arch: resnet18]|[transforms: True]|[dropout: 0]|[weight_decay: 0.01]|[training_schedule: head_first_then_body]|[discriminative_lr: False]|[one_cycle_policy: True]</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <td>41.121185</td>\n",
       "      <td>33.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <td>40.931316</td>\n",
       "      <td>30.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                              duration  \\\n",
       "0 PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects           27.527227   \n",
       "                                                     fridgeObjectsWatermark  29.327206   \n",
       "  PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects           46.494100   \n",
       "                                                     fridgeObjectsWatermark  40.436677   \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects           29.745073   \n",
       "                                                     fridgeObjectsWatermark  28.454158   \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects           44.277393   \n",
       "                                                     fridgeObjectsWatermark  40.866518   \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           28.009722   \n",
       "                                                     fridgeObjectsWatermark  29.721222   \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           40.376158   \n",
       "                                                     fridgeObjectsWatermark  42.627545   \n",
       "1 PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects           30.931857   \n",
       "                                                     fridgeObjectsWatermark  26.125927   \n",
       "  PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects           46.117437   \n",
       "                                                     fridgeObjectsWatermark  40.555442   \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects           29.870988   \n",
       "                                                     fridgeObjectsWatermark  25.864497   \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects           46.807896   \n",
       "                                                     fridgeObjectsWatermark  41.351353   \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           25.873023   \n",
       "                                                     fridgeObjectsWatermark  25.889981   \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           41.558083   \n",
       "                                                     fridgeObjectsWatermark  41.196609   \n",
       "2 PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects           25.954923   \n",
       "                                                     fridgeObjectsWatermark  25.766089   \n",
       "  PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects           40.162561   \n",
       "                                                     fridgeObjectsWatermark  41.274331   \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects           26.026493   \n",
       "                                                     fridgeObjectsWatermark  25.616917   \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects           40.691592   \n",
       "                                                     fridgeObjectsWatermark  40.468641   \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           26.239308   \n",
       "                                                     fridgeObjectsWatermark  26.347744   \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           41.121185   \n",
       "                                                     fridgeObjectsWatermark  40.931316   \n",
       "\n",
       "                                                                             rank  \n",
       "0 PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects            3.0  \n",
       "                                                     fridgeObjectsWatermark   7.0  \n",
       "  PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects            7.0  \n",
       "                                                     fridgeObjectsWatermark   9.0  \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects            2.0  \n",
       "                                                     fridgeObjectsWatermark   1.0  \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects            1.0  \n",
       "                                                     fridgeObjectsWatermark   1.0  \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           20.0  \n",
       "                                                     fridgeObjectsWatermark  22.0  \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           25.0  \n",
       "                                                     fridgeObjectsWatermark  34.0  \n",
       "1 PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects            4.0  \n",
       "                                                     fridgeObjectsWatermark   8.5  \n",
       "  PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects           11.0  \n",
       "                                                     fridgeObjectsWatermark  10.0  \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects            1.0  \n",
       "                                                     fridgeObjectsWatermark   1.0  \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects            1.0  \n",
       "                                                     fridgeObjectsWatermark   1.0  \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           26.0  \n",
       "                                                     fridgeObjectsWatermark  26.0  \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           23.0  \n",
       "                                                     fridgeObjectsWatermark  30.0  \n",
       "2 PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects            3.0  \n",
       "                                                     fridgeObjectsWatermark   4.0  \n",
       "  PARAMETERS [learning_rate: 0.0001]|[epochs: 16]... fridgeObjects            9.0  \n",
       "                                                     fridgeObjectsWatermark   9.0  \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects            3.0  \n",
       "                                                     fridgeObjectsWatermark   1.0  \n",
       "  PARAMETERS [learning_rate: 0.001]|[epochs: 16]|... fridgeObjects            1.0  \n",
       "                                                     fridgeObjectsWatermark   1.0  \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           19.0  \n",
       "                                                     fridgeObjectsWatermark  23.0  \n",
       "  PARAMETERS [learning_rate: 1e-05]|[epochs: 16]|... fridgeObjects           33.0  \n",
       "                                                     fridgeObjectsWatermark  30.0  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sweeper.update_parameters(learning_rate=LEARNING_RATES, im_size=IM_SIZES, epochs=EPOCHS, dropout=DROPOUTS)\n",
    "df = sweeper.run(datasets=DATA_PATHS, reps=REPS, metric_fct=retrieval_rank); \n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize Results <a name=\"visualize\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When we read in multi-index dataframe, index 0 represents the run number, index 1 represents a single permutation of parameters, and index 2 represents the dataset. To see the results, show the df using the `clean_sweeper_df` helper function. This will display all the hyperparameters in a nice, readable way."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = clean_sweeper_df(df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since we've run our benchmarking over 3 repetitions, we may want to just look at the averages across the different __run numbers__."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"2\" halign=\"left\">P: [learning_rate: 0.0001] [im_size: 300]</th>\n",
       "      <th colspan=\"2\" halign=\"left\">P: [learning_rate: 0.0001] [im_size: 500]</th>\n",
       "      <th colspan=\"2\" halign=\"left\">P: [learning_rate: 0.001] [im_size: 300]</th>\n",
       "      <th colspan=\"2\" halign=\"left\">P: [learning_rate: 0.001] [im_size: 500]</th>\n",
       "      <th colspan=\"2\" halign=\"left\">P: [learning_rate: 1e-05] [im_size: 300]</th>\n",
       "      <th colspan=\"2\" halign=\"left\">P: [learning_rate: 1e-05] [im_size: 500]</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "      <th>fridgeObjects</th>\n",
       "      <th>fridgeObjectsWatermark</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>duration</th>\n",
       "      <td>28.138002</td>\n",
       "      <td>27.073074</td>\n",
       "      <td>44.258033</td>\n",
       "      <td>40.755483</td>\n",
       "      <td>28.547518</td>\n",
       "      <td>26.645191</td>\n",
       "      <td>43.925627</td>\n",
       "      <td>40.895504</td>\n",
       "      <td>26.707351</td>\n",
       "      <td>27.319649</td>\n",
       "      <td>41.018475</td>\n",
       "      <td>41.585157</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rank</th>\n",
       "      <td>3.333333</td>\n",
       "      <td>6.500000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>9.333333</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>21.666667</td>\n",
       "      <td>23.666667</td>\n",
       "      <td>27.000000</td>\n",
       "      <td>31.333333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         P: [learning_rate: 0.0001] [im_size: 300]                         \\\n",
       "                                     fridgeObjects fridgeObjectsWatermark   \n",
       "duration                                 28.138002              27.073074   \n",
       "rank                                      3.333333               6.500000   \n",
       "\n",
       "         P: [learning_rate: 0.0001] [im_size: 500]                         \\\n",
       "                                     fridgeObjects fridgeObjectsWatermark   \n",
       "duration                                 44.258033              40.755483   \n",
       "rank                                      9.000000               9.333333   \n",
       "\n",
       "         P: [learning_rate: 0.001] [im_size: 300]                         \\\n",
       "                                    fridgeObjects fridgeObjectsWatermark   \n",
       "duration                                28.547518              26.645191   \n",
       "rank                                     2.000000               1.000000   \n",
       "\n",
       "         P: [learning_rate: 0.001] [im_size: 500]                         \\\n",
       "                                    fridgeObjects fridgeObjectsWatermark   \n",
       "duration                                43.925627              40.895504   \n",
       "rank                                     1.000000               1.000000   \n",
       "\n",
       "         P: [learning_rate: 1e-05] [im_size: 300]                         \\\n",
       "                                    fridgeObjects fridgeObjectsWatermark   \n",
       "duration                                26.707351              27.319649   \n",
       "rank                                    21.666667              23.666667   \n",
       "\n",
       "         P: [learning_rate: 1e-05] [im_size: 500]                         \n",
       "                                    fridgeObjects fridgeObjectsWatermark  \n",
       "duration                                41.018475              41.585157  \n",
       "rank                                    27.000000              31.333333  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.mean(level=(1,2)).T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Print the average accuracy over the different runs for each dataset independently."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = df.mean(level=(1,2))[\"rank\"].unstack().plot(kind='bar', figsize=(12, 6))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Additionally, we may want simply to see which set of hyperparameters perform the best across the different __datasets__. We can do that by averaging the results of the different datasets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>P: [learning_rate: 0.0001] [im_size: 300]</th>\n",
       "      <th>P: [learning_rate: 0.0001] [im_size: 500]</th>\n",
       "      <th>P: [learning_rate: 0.001] [im_size: 300]</th>\n",
       "      <th>P: [learning_rate: 0.001] [im_size: 500]</th>\n",
       "      <th>P: [learning_rate: 1e-05] [im_size: 300]</th>\n",
       "      <th>P: [learning_rate: 1e-05] [im_size: 500]</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>duration</th>\n",
       "      <td>27.605538</td>\n",
       "      <td>42.506758</td>\n",
       "      <td>27.596354</td>\n",
       "      <td>42.410565</td>\n",
       "      <td>27.013500</td>\n",
       "      <td>41.301816</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rank</th>\n",
       "      <td>4.916667</td>\n",
       "      <td>9.166667</td>\n",
       "      <td>1.500000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>22.666667</td>\n",
       "      <td>29.166667</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          P: [learning_rate: 0.0001] [im_size: 300]  \\\n",
       "duration                                  27.605538   \n",
       "rank                                       4.916667   \n",
       "\n",
       "          P: [learning_rate: 0.0001] [im_size: 500]  \\\n",
       "duration                                  42.506758   \n",
       "rank                                       9.166667   \n",
       "\n",
       "          P: [learning_rate: 0.001] [im_size: 300]  \\\n",
       "duration                                 27.596354   \n",
       "rank                                      1.500000   \n",
       "\n",
       "          P: [learning_rate: 0.001] [im_size: 500]  \\\n",
       "duration                                 42.410565   \n",
       "rank                                      1.000000   \n",
       "\n",
       "          P: [learning_rate: 1e-05] [im_size: 300]  \\\n",
       "duration                                 27.013500   \n",
       "rank                                     22.666667   \n",
       "\n",
       "          P: [learning_rate: 1e-05] [im_size: 500]  \n",
       "duration                                 41.301816  \n",
       "rank                                     29.166667  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.mean(level=(1)).T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To make it easier to see which permutation did the best, we can plot the results using the `plot_sweeper_df` helper function. This plot will help us easily see which parameters offer the highest accuracies."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sweeper_df(df.mean(level=(1)), sort_by=\"rank\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/scrapbook.scrap.json+json": {
       "data": 36,
       "encoder": "json",
       "name": "nr_elements",
       "version": 1
      }
     },
     "metadata": {
      "scrapbook": {
       "data": true,
       "display": false,
       "name": "nr_elements"
      }
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "application/scrapbook.scrap.json+json": {
       "data": [
        4.916666666666667,
        9.166666666666666,
        1.5,
        1,
        22.666666666666668,
        29.166666666666668
       ],
       "encoder": "json",
       "name": "ranks",
       "version": 1
      }
     },
     "metadata": {
      "scrapbook": {
       "data": true,
       "display": false,
       "name": "ranks"
      }
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "application/scrapbook.scrap.json+json": {
       "data": 46.807896,
       "encoder": "json",
       "name": "max_duration",
       "version": 1
      }
     },
     "metadata": {
      "scrapbook": {
       "data": true,
       "display": false,
       "name": "max_duration"
      }
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "application/scrapbook.scrap.json+json": {
       "data": 25.616917,
       "encoder": "json",
       "name": "min_duration",
       "version": 1
      }
     },
     "metadata": {
      "scrapbook": {
       "data": true,
       "display": false,
       "name": "min_duration"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Preserve some of the notebook outputs\n",
    "sb.glue(\"nr_elements\", len(df))\n",
    "sb.glue(\"ranks\", list(df.mean(level=(1))[\"rank\"]))\n",
    "sb.glue(\"max_duration\", df.max().duration)\n",
    "sb.glue(\"min_duration\", df.min().duration)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (cv)",
   "language": "python",
   "name": "cv"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}