DataAnalysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "source": [
    "# Data analysis\n",
    "\n",
    "This notebook generates all the plots used in the paper. In order to use this notebook specify the *checkpoint_folder*, in which is located *fmnist.csv* and *CIFAR.csv*."
   ],
   "metadata": {
    "id": "9veb-4Oni6bb",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "40zPtJsx_njC",
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "-IyZV96e_k_u",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "#libraries import\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from collections import namedtuple\n",
    "from datetime import datetime\n",
    "from itertools import product\n",
    "import numpy as np\n",
    "import scipy.stats\n",
    "import pandas as pd\n",
    "import os\n",
    "\n",
    "from helpers import * \n",
    "\n",
    "sns.set_style(\"dark\")"
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "# specify the folder in which are csv files located\n",
    "results_folder = 'results/'\n",
    "fmnist_df_path = f'{results_folder}fmnist.csv'\n",
    "cifar_df_path = f'{results_folder}CIFAR10.csv'\n",
    "\n",
    "# specify the folder in which plots are going to be saved. If None, plots are not going to be saved\n",
    "plot_folder = ''\n",
    "checkpoints = \"checkpoints/\""
   ],
   "metadata": {
    "id": "xe0qkPrtXnUh",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Creating and loading dataframes"
   ],
   "metadata": {
    "id": "c1ah_tCHru6n",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "### FashionMNIST"
   ],
   "metadata": {
    "id": "YOnVY0lSsOgN",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "if os.path.exists(fmnist_df_path):\n",
    "  fmnist_df = pd.read_csv(fmnist_df_path, index_col=[0])\n",
    "else:\n",
    "  print(\"Data needs to be downloaded form GDrive to extract the results for plotting. Please refer to the README.md file.\")\n",
    "  fmnist_df = generate_dataframe_per_dataset(\"FashionMNIST\", checkpoints)\n",
    "  fmnist_df.to_csv(fmnist_df_path)\n",
    "print('Done')"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "0N9tAxm2eBCi",
    "outputId": "accf4eea-4e29-46c6-dd48-45d8970ee943",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Done\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "fmnist_df.head()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 206
    },
    "id": "XZnbi_unoa19",
    "outputId": "9d567860-e5ba-4d82-c1bd-777b9349bef1",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "  Minimization     optim Optimizer  training_accuracy  Validation accuracy  \\\n",
       "0     standard      Adam      Adam          99.513336            91.979996   \n",
       "1     standard   SGD 0.1       SGD         100.000000            92.769997   \n",
       "2     standard  AdaBound  AdaBound         100.000000            92.720001   \n",
       "3     standard  AdaShift  AdaShift          99.781670            92.389999   \n",
       "4     standard   Adagrad   Adagrad         100.000000            92.659996   \n",
       "\n",
       "   training_loss  val_loss  Generalization gap epoch     Sharpness  \\\n",
       "0       0.014862  0.729570            7.533340  50.0  40364.594347   \n",
       "1       0.000050  0.516581            7.230003  50.0  37819.346180   \n",
       "2       0.000055  0.537387            7.279999  50.0  35747.061795   \n",
       "3       0.005874  0.600980            7.391670  50.0  47427.310191   \n",
       "4       0.000585  0.440312            7.340004  50.0  42589.065929   \n",
       "\n",
       "  Architecture  \n",
       "0  SimpleBatch  \n",
       "1  SimpleBatch  \n",
       "2  SimpleBatch  \n",
       "3  SimpleBatch  \n",
       "4  SimpleBatch  "
      ],
      "text/html": [
       "\n",
       "  <div id=\"df-3243719f-9077-4a12-9b12-090090b43628\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>Minimization</th>\n",
       "      <th>optim</th>\n",
       "      <th>Optimizer</th>\n",
       "      <th>training_accuracy</th>\n",
       "      <th>Validation accuracy</th>\n",
       "      <th>training_loss</th>\n",
       "      <th>val_loss</th>\n",
       "      <th>Generalization gap</th>\n",
       "      <th>epoch</th>\n",
       "      <th>Sharpness</th>\n",
       "      <th>Architecture</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>standard</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.513336</td>\n",
       "      <td>91.979996</td>\n",
       "      <td>0.014862</td>\n",
       "      <td>0.729570</td>\n",
       "      <td>7.533340</td>\n",
       "      <td>50.0</td>\n",
       "      <td>40364.594347</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>standard</td>\n",
       "      <td>SGD 0.1</td>\n",
       "      <td>SGD</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>92.769997</td>\n",
       "      <td>0.000050</td>\n",
       "      <td>0.516581</td>\n",
       "      <td>7.230003</td>\n",
       "      <td>50.0</td>\n",
       "      <td>37819.346180</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>standard</td>\n",
       "      <td>AdaBound</td>\n",
       "      <td>AdaBound</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>92.720001</td>\n",
       "      <td>0.000055</td>\n",
       "      <td>0.537387</td>\n",
       "      <td>7.279999</td>\n",
       "      <td>50.0</td>\n",
       "      <td>35747.061795</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>standard</td>\n",
       "      <td>AdaShift</td>\n",
       "      <td>AdaShift</td>\n",
       "      <td>99.781670</td>\n",
       "      <td>92.389999</td>\n",
       "      <td>0.005874</td>\n",
       "      <td>0.600980</td>\n",
       "      <td>7.391670</td>\n",
       "      <td>50.0</td>\n",
       "      <td>47427.310191</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>standard</td>\n",
       "      <td>Adagrad</td>\n",
       "      <td>Adagrad</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>92.659996</td>\n",
       "      <td>0.000585</td>\n",
       "      <td>0.440312</td>\n",
       "      <td>7.340004</td>\n",
       "      <td>50.0</td>\n",
       "      <td>42589.065929</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-3243719f-9077-4a12-9b12-090090b43628')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-3243719f-9077-4a12-9b12-090090b43628 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-3243719f-9077-4a12-9b12-090090b43628');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ]
     },
     "metadata": {},
     "execution_count": 43
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "### CIFAR10"
   ],
   "metadata": {
    "id": "BDHYTrkWsQrK",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "if os.path.exists(cifar_df_path):\n",
    "  cifar_df = pd.read_csv(cifar_df_path, index_col=[0])\n",
    "else:\n",
    "  print(\"Data needs to be downloaded form GDrive to extract the results for plotting. Please refer to the README.md file.\")\n",
    "  cifar_df = generate_dataframe_per_dataset('CIFAR10', checkpoints)\n",
    "  cifar_df.to_csv(cifar_df_path)\n",
    "print('Done')"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "zu1aOfjDsSS4",
    "outputId": "5abbe1d3-afaa-485a-e29f-4a9ac5652721",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Done\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "cifar_df.head()\n"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 395
    },
    "id": "kfUEbnddsUPL",
    "outputId": "062157c7-0269-4f74-a815-a802ac4dbdf9",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "   Minimization           optim Optimizer  training_accuracy  \\\n",
       "3      standard  AdaShift 0.001  AdaShift          99.459999   \n",
       "10     standard   AdaShift 0.01  AdaShift          99.593994   \n",
       "13     standard            Adam      Adam          99.691994   \n",
       "21          SAM   AdaShift 0.01  AdaShift          99.610001   \n",
       "27          SAM            Adam      Adam          99.627998   \n",
       "10          SAM            Adam      Adam          99.559998   \n",
       "29     standard            Adam      Adam          99.787994   \n",
       "34          SAM            Adam      Adam          99.711998   \n",
       "3      standard   AdaShift 0.01  AdaShift          99.631996   \n",
       "6      standard            Adam      Adam          99.591995   \n",
       "10          SAM            Adam      Adam          99.624001   \n",
       "\n",
       "    Validation accuracy  training_loss  val_loss  Generalization gap  epoch  \\\n",
       "3             71.589996       0.019315  1.619780           27.870003   50.0   \n",
       "10            76.610001       0.012215  2.136920           22.983994  100.0   \n",
       "13            76.199997       0.010103  2.927628           23.491997  150.0   \n",
       "21            78.209999       0.010428  1.436348           21.400002  150.0   \n",
       "27            77.549995       0.010072  1.882256           22.078003  200.0   \n",
       "10            84.089996       0.012892  1.125786           15.470001  100.0   \n",
       "29            83.779999       0.010034  2.593193           16.007996  200.0   \n",
       "34            84.279999       0.010790  2.117534           15.431999  200.0   \n",
       "3             84.320000       0.010287  1.151364           15.311996   50.0   \n",
       "6             85.689995       0.014837  1.434866           13.902000  100.0   \n",
       "10            86.779999       0.011002  0.970687           12.844002  100.0   \n",
       "\n",
       "        Sharpness  Architecture  \n",
       "3   467108.858864   SimpleBatch  \n",
       "10   99015.547609   SimpleBatch  \n",
       "13   85274.343582   SimpleBatch  \n",
       "21   27490.104654   SimpleBatch  \n",
       "27   21864.985911   SimpleBatch  \n",
       "10   15772.315452   MiddleBatch  \n",
       "29  125334.552863   MiddleBatch  \n",
       "34   45614.067676   MiddleBatch  \n",
       "3    64448.557400  ComplexBatch  \n",
       "6    96346.232905  ComplexBatch  \n",
       "10   14567.035782  ComplexBatch  "
      ],
      "text/html": [
       "\n",
       "  <div id=\"df-e0487a0c-f812-48bb-985d-fe009ae9e76b\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>Minimization</th>\n",
       "      <th>optim</th>\n",
       "      <th>Optimizer</th>\n",
       "      <th>training_accuracy</th>\n",
       "      <th>Validation accuracy</th>\n",
       "      <th>training_loss</th>\n",
       "      <th>val_loss</th>\n",
       "      <th>Generalization gap</th>\n",
       "      <th>epoch</th>\n",
       "      <th>Sharpness</th>\n",
       "      <th>Architecture</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>standard</td>\n",
       "      <td>AdaShift 0.001</td>\n",
       "      <td>AdaShift</td>\n",
       "      <td>99.459999</td>\n",
       "      <td>71.589996</td>\n",
       "      <td>0.019315</td>\n",
       "      <td>1.619780</td>\n",
       "      <td>27.870003</td>\n",
       "      <td>50.0</td>\n",
       "      <td>467108.858864</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>standard</td>\n",
       "      <td>AdaShift 0.01</td>\n",
       "      <td>AdaShift</td>\n",
       "      <td>99.593994</td>\n",
       "      <td>76.610001</td>\n",
       "      <td>0.012215</td>\n",
       "      <td>2.136920</td>\n",
       "      <td>22.983994</td>\n",
       "      <td>100.0</td>\n",
       "      <td>99015.547609</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>standard</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.691994</td>\n",
       "      <td>76.199997</td>\n",
       "      <td>0.010103</td>\n",
       "      <td>2.927628</td>\n",
       "      <td>23.491997</td>\n",
       "      <td>150.0</td>\n",
       "      <td>85274.343582</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>SAM</td>\n",
       "      <td>AdaShift 0.01</td>\n",
       "      <td>AdaShift</td>\n",
       "      <td>99.610001</td>\n",
       "      <td>78.209999</td>\n",
       "      <td>0.010428</td>\n",
       "      <td>1.436348</td>\n",
       "      <td>21.400002</td>\n",
       "      <td>150.0</td>\n",
       "      <td>27490.104654</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>SAM</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.627998</td>\n",
       "      <td>77.549995</td>\n",
       "      <td>0.010072</td>\n",
       "      <td>1.882256</td>\n",
       "      <td>22.078003</td>\n",
       "      <td>200.0</td>\n",
       "      <td>21864.985911</td>\n",
       "      <td>SimpleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>SAM</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.559998</td>\n",
       "      <td>84.089996</td>\n",
       "      <td>0.012892</td>\n",
       "      <td>1.125786</td>\n",
       "      <td>15.470001</td>\n",
       "      <td>100.0</td>\n",
       "      <td>15772.315452</td>\n",
       "      <td>MiddleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>standard</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.787994</td>\n",
       "      <td>83.779999</td>\n",
       "      <td>0.010034</td>\n",
       "      <td>2.593193</td>\n",
       "      <td>16.007996</td>\n",
       "      <td>200.0</td>\n",
       "      <td>125334.552863</td>\n",
       "      <td>MiddleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>SAM</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.711998</td>\n",
       "      <td>84.279999</td>\n",
       "      <td>0.010790</td>\n",
       "      <td>2.117534</td>\n",
       "      <td>15.431999</td>\n",
       "      <td>200.0</td>\n",
       "      <td>45614.067676</td>\n",
       "      <td>MiddleBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>standard</td>\n",
       "      <td>AdaShift 0.01</td>\n",
       "      <td>AdaShift</td>\n",
       "      <td>99.631996</td>\n",
       "      <td>84.320000</td>\n",
       "      <td>0.010287</td>\n",
       "      <td>1.151364</td>\n",
       "      <td>15.311996</td>\n",
       "      <td>50.0</td>\n",
       "      <td>64448.557400</td>\n",
       "      <td>ComplexBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>standard</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.591995</td>\n",
       "      <td>85.689995</td>\n",
       "      <td>0.014837</td>\n",
       "      <td>1.434866</td>\n",
       "      <td>13.902000</td>\n",
       "      <td>100.0</td>\n",
       "      <td>96346.232905</td>\n",
       "      <td>ComplexBatch</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>SAM</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.624001</td>\n",
       "      <td>86.779999</td>\n",
       "      <td>0.011002</td>\n",
       "      <td>0.970687</td>\n",
       "      <td>12.844002</td>\n",
       "      <td>100.0</td>\n",
       "      <td>14567.035782</td>\n",
       "      <td>ComplexBatch</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-e0487a0c-f812-48bb-985d-fe009ae9e76b')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-e0487a0c-f812-48bb-985d-fe009ae9e76b button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-e0487a0c-f812-48bb-985d-fe009ae9e76b');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ]
     },
     "metadata": {},
     "execution_count": 45
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "# Plotting\n"
   ],
   "metadata": {
    "id": "iB1myu-br2AE",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Validation accuracies and sharpnesses"
   ],
   "metadata": {
    "id": "LMMt2-a7s6lH",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "\n",
    "### FashionMNIST"
   ],
   "metadata": {
    "id": "EzPhWmoIs-JL",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "########## Plot that is included in the paper\n",
    "fig, ax = plt.subplots(1,3, figsize=(21, 7), sharex=True)\n",
    "fig.suptitle(f'FashionMNIST dataset', fontsize=20)\n",
    "\n",
    "for i,arc in enumerate(['SimpleBatch', 'MiddleBatch', 'ComplexBatch']):\n",
    "    df = fmnist_df[(fmnist_df['Architecture']==arc)&(fmnist_df['epoch']=='converged')]\n",
    "    orient = 'h'\n",
    "    \n",
    "    sns.barplot(data=df.sort_values('Validation accuracy', ascending=False), \n",
    "                y='Optimizer', hue='Minimization', x='Validation accuracy',\n",
    "                ax=ax[i], \n",
    "                palette=\"deep\", orient=orient)\n",
    "    \n",
    "    ax[i].set_title(arc, fontsize=15)\n",
    "    if i != 0:\n",
    "      ax[i].set(ylabel=None)\n",
    "      ax[i].get_legend().remove()\n",
    "    ax[i].set_xlabel('Test accuracy')\n",
    "\n",
    "ax[0].set_ylabel('Optimizer', fontsize=12)\n",
    "ax[0].set_xlim(91, 95)\n",
    "ax[0].legend(fontsize=14)\n",
    "\n",
    "\n",
    "if plot_folder != None:\n",
    "    plt.savefig(f'{plot_folder}FMNIST_val_acc_same_scale.png', dpi=300)\n",
    "plt.show()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 493
    },
    "id": "W2-nij56e9W1",
    "outputId": "4b65a17f-3c4f-4740-914a-6dceb3d8d57b",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 1512x504 with 3 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "########## Plot that is included in the paper\n",
    "fig, ax = plt.subplots(1,3, figsize=(21, 7), sharex=True)\n",
    "fig.suptitle(f'FashionMNIST dataset', fontsize=20)\n",
    "\n",
    "for i,arc in enumerate(['SimpleBatch', 'MiddleBatch', 'ComplexBatch']):\n",
    "    df = fmnist_df[(fmnist_df['Architecture']==arc)&(fmnist_df['epoch']=='converged')]\n",
    "    orient = 'h'\n",
    "    \n",
    "    sns.barplot(data=df.sort_values('Sharpness'), \n",
    "                y='Optimizer', hue='Minimization', x='Sharpness',\n",
    "                ax=ax[i], \n",
    "                palette=\"deep\", orient=orient)\n",
    "    \n",
    "    ax[i].set_title(arc, fontsize=15)\n",
    "    if i != 0:\n",
    "      ax[i].set(ylabel=None)\n",
    "      ax[i].get_legend().remove()\n",
    "\n",
    "ax[0].legend(fontsize=14)\n",
    "ax[0].set_ylabel('Optimizer', fontsize=12)\n",
    "\n",
    "\n",
    "if plot_folder != None:\n",
    "    plt.savefig(f'{plot_folder}FMNIST_sharpness_same_scale.png', dpi=300)\n",
    "plt.show()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 493
    },
    "id": "_zV9QXl-fZ4L",
    "outputId": "6b2beca0-48c4-498d-a3a1-f3a48a802c47",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 1512x504 with 3 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "########## Plot that is included in the paper\n",
    "fig, ax = plt.subplots(1,3, figsize=(21, 7), sharex=True)\n",
    "fig.suptitle(f'CIFAR10 dataset', fontsize=20)\n",
    "\n",
    "for i,arc in enumerate(['SimpleBatch', 'MiddleBatch', 'ComplexBatch']):\n",
    "    df = cifar_df[(cifar_df['Architecture']==arc)&(cifar_df['epoch']=='converged')]\n",
    "    orient = 'h'\n",
    "    \n",
    "    sns.barplot(data=df.sort_values('Validation accuracy', ascending=False), \n",
    "                y='Optimizer', hue='Minimization', x='Validation accuracy',\n",
    "                ax=ax[i], \n",
    "                palette=\"deep\", orient=orient)\n",
    "    \n",
    "    ax[i].set_title(arc, fontsize=15)\n",
    "    if i != 0:\n",
    "      ax[i].set(ylabel=None)\n",
    "      ax[i].get_legend().remove()\n",
    "    ax[i].set_xlabel('Test accuracy')\n",
    "\n",
    "ax[0].set_ylabel('Optimizer', fontsize=12)\n",
    "ax[0].set_xlim(74, 88)\n",
    "ax[0].legend(fontsize=14)\n",
    "\n",
    "if plot_folder != None:\n",
    "    plt.savefig(f'{plot_folder}CIFAR10_val_acc.png', dpi=300)\n",
    "plt.show()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 493
    },
    "id": "n7e5zP_lWHOp",
    "outputId": "a882ea47-a65b-4cd9-e7b0-676975bd2d50",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 1512x504 with 3 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "########## Plot that is included in the paper\n",
    "fig, ax = plt.subplots(1,3, figsize=(21, 7), sharex=True)\n",
    "fig.suptitle(f'CIFAR10 dataset', fontsize=20)\n",
    "\n",
    "for i,arc in enumerate(['SimpleBatch', 'MiddleBatch', 'ComplexBatch']):\n",
    "    df = cifar_df[(cifar_df['Architecture']==arc)&(cifar_df['epoch']=='converged')]\n",
    "    orient = 'h'\n",
    "    \n",
    "    sns.barplot(data=df.sort_values('Sharpness'), \n",
    "                y='Optimizer', hue='Minimization', x='Sharpness',\n",
    "                ax=ax[i], \n",
    "                palette=\"deep\", orient=orient)\n",
    "    \n",
    "    ax[i].set_title(arc, fontsize=15)\n",
    "    if i != 0:\n",
    "      ax[i].set(ylabel=None)\n",
    "      ax[i].get_legend().remove()\n",
    "\n",
    "ax[0].legend(fontsize=14)\n",
    "ax[0].set_ylabel('Optimizer', fontsize=12)\n",
    "\n",
    "if plot_folder != None:\n",
    "    plt.savefig(f'{plot_folder}CIFAR10_sharpness_same_scale.png', dpi=300)\n",
    "plt.show()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 493
    },
    "id": "M197bPadcKGx",
    "outputId": "a7be8c96-e8a0-46c3-dd94-431b2391b28e",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 1512x504 with 3 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Corelation between minimum sharpness and generalization gap"
   ],
   "metadata": {
    "id": "Ngl7RtdHsCd7",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "########## Plot not included in the paper - measures correlation for each architecture for FashionMNIST\n",
    "fig, ax = plt.subplots(1,3, figsize=(21, 7))\n",
    "fig.suptitle(f'FashionMNIST dataset', fontsize=20)\n",
    "\n",
    "for i,arc in enumerate(['SimpleBatch', 'MiddleBatch', 'ComplexBatch']): \n",
    "    sns.scatterplot(data=fmnist_df[fmnist_df.Architecture == arc], x='Generalization gap', y='Sharpness', hue='Minimization', ax=ax[i])\n",
    "    if i != 0:\n",
    "        ax[i].set_ylabel('')\n",
    "        ax[i].get_legend().remove()\n",
    "\n",
    "    x = fmnist_df[fmnist_df.Architecture == arc].dropna()['Generalization gap']\n",
    "    y = fmnist_df[fmnist_df.Architecture == arc].dropna()['Sharpness']\n",
    "    k = scipy.stats.kendalltau(x, y)\n",
    "    ax[i].set_title(f'{arc} ({k[0]:.2f})', fontsize=15)\n",
    "\n",
    "ax[0].set_ylabel('Sharpness', fontsize=12)\n",
    "ax[0].legend(fontsize=14)\n",
    "\n",
    "plt.savefig(f'{plot_folder}correlation_fmnist.png', dpi=300)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 493
    },
    "id": "5UcCVo0J4fYP",
    "outputId": "98be4acd-1640-43c3-9f5e-f4aeeb98fd3b",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 1512x504 with 3 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "########## Plot that is not included in the paper - measures correlation for each architecture for CIFAR10\n",
    "fig, ax = plt.subplots(1,3, figsize=(21, 7))\n",
    "fig.suptitle(f'CIFAR10 dataset', fontsize=20)\n",
    "\n",
    "for i,arc in enumerate(['SimpleBatch', 'MiddleBatch', 'ComplexBatch']): \n",
    "    sns.scatterplot(data=cifar_df[cifar_df.Architecture == arc], x='Generalization gap', y='Sharpness', hue='Minimization', ax=ax[i])\n",
    "    if i != 0:\n",
    "        ax[i].set_ylabel('')\n",
    "        ax[i].get_legend().remove()\n",
    "\n",
    "    x = cifar_df[cifar_df.Architecture == arc].dropna()['Generalization gap']\n",
    "    y = cifar_df[cifar_df.Architecture == arc].dropna()['Sharpness']\n",
    "    k = scipy.stats.kendalltau(x, y)\n",
    "    ax[i].set_title(f'{arc} ({k[0]:.2f})', fontsize=15)\n",
    "\n",
    "ax[0].set_ylabel('Sharpness', fontsize=12)\n",
    "ax[0].legend(fontsize=14)\n",
    "\n",
    "if plot_folder != None:\n",
    "    plt.savefig(f'{plot_folder}correlation_cifar.png', dpi=300)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 493
    },
    "id": "otigfVwdk5gr",
    "outputId": "9b8dca84-13f2-4956-f695-8f110f43cf9e",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 1512x504 with 3 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "### Plot in the paper \n",
    "fig, ax = plt.subplots(1,2, figsize=(12, 6))\n",
    "\n",
    "sns.scatterplot(data=fmnist_df[fmnist_df.epoch=='converged'], x='Generalization gap', y='Sharpness', hue='Minimization', ax=ax[0], palette='deep')\n",
    "x = fmnist_df[fmnist_df.epoch=='converged'].dropna()['Generalization gap']\n",
    "y = fmnist_df[fmnist_df.epoch=='converged'].dropna()['Sharpness']\n",
    "k = scipy.stats.kendalltau(x, y)\n",
    "ax[0].set_title(f'FashionMNIST ({k[0]:.2f})', fontsize=15)\n",
    "ax[0].set_ylabel('Sharpness', fontsize=12)\n",
    "print(f'p-value {k[1]}')\n",
    "\n",
    "sns.scatterplot(data=cifar_df[cifar_df.epoch =='converged'], x='Generalization gap', y='Sharpness', hue='Minimization', ax=ax[1], palette='deep')\n",
    "x = cifar_df[cifar_df.epoch =='converged'].dropna()['Generalization gap']\n",
    "y = cifar_df[cifar_df.epoch =='converged'].dropna()['Sharpness']\n",
    "k = scipy.stats.kendalltau(x, y)\n",
    "ax[1].set_title(f'CIFAR10 ({k[0]:.2f})', fontsize=15)\n",
    "ax[1].set(ylabel=None)\n",
    "print(f'p-value {k[1]}')\n",
    "\n",
    "if plot_folder != None:\n",
    "    plt.savefig(f'{plot_folder}correlation_final.png', dpi=300)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 441
    },
    "id": "HfPsNG6McESH",
    "outputId": "87c294ab-932d-4f00-dd04-bffc28705165",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "p-value 1.675882899619825e-05\n",
      "p-value 0.011293415200324022\n"
     ]
    },
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 864x432 with 2 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Choice of optimize - adaptive vs. non-adaptive"
   ],
   "metadata": {
    "id": "r5IHTRQKtVov",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "# Adding indicator if optimizor is adaptive or not\n",
    "is_adaptive = lambda x : 'adaptive' if x.lower().startswith('ada') else 'non-adaptive'\n",
    "\n",
    "fmnist_df['adaptive'] = fmnist_df['optim'].apply(is_adaptive)\n",
    "fmnist_df['adaptive'].describe()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "HRM6aULVv_oO",
    "outputId": "ee07a1f0-1a13-47de-b575-1bba2aee3d04",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "count          169\n",
       "unique           2\n",
       "top       adaptive\n",
       "freq           112\n",
       "Name: adaptive, dtype: object"
      ]
     },
     "metadata": {},
     "execution_count": 159
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "# Reporting average validation accuracy and sharpness\n",
    "\n",
    "fmnist_df_no_sam = fmnist_df[fmnist_df['Minimization']=='standard'].dropna()\n",
    "fmnist_df_no_sam['dataset'] = 'FashionMNIST'\n",
    "fmnist_df_no_sam.groupby('Optimizer').mean().sort_values('Validation accuracy', ascending = False)[['Validation accuracy', 'Sharpness']]"
   ],
   "metadata": {
    "id": "m-gNcFvsc1ic",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 269
    },
    "outputId": "5dcf3de9-6b6c-47ab-883c-3a1c5e85030e",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "           Validation accuracy      Sharpness\n",
       "Optimizer                                    \n",
       "SGD                  93.355997   90347.966340\n",
       "AdaBound             93.344665  111408.603229\n",
       "Adam                 93.040663   59814.534787\n",
       "PHB                  92.969331  185464.466488\n",
       "AdaShift             92.742855   95656.100080\n",
       "Adagrad              92.602665  561759.559346"
      ],
      "text/html": [
       "\n",
       "  <div id=\"df-37f890ce-38d9-4ecf-a871-63384126d01b\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>Validation accuracy</th>\n",
       "      <th>Sharpness</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Optimizer</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>SGD</th>\n",
       "      <td>93.355997</td>\n",
       "      <td>90347.966340</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AdaBound</th>\n",
       "      <td>93.344665</td>\n",
       "      <td>111408.603229</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Adam</th>\n",
       "      <td>93.040663</td>\n",
       "      <td>59814.534787</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PHB</th>\n",
       "      <td>92.969331</td>\n",
       "      <td>185464.466488</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AdaShift</th>\n",
       "      <td>92.742855</td>\n",
       "      <td>95656.100080</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Adagrad</th>\n",
       "      <td>92.602665</td>\n",
       "      <td>561759.559346</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-37f890ce-38d9-4ecf-a871-63384126d01b')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-37f890ce-38d9-4ecf-a871-63384126d01b button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-37f890ce-38d9-4ecf-a871-63384126d01b');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ]
     },
     "metadata": {},
     "execution_count": 161
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "cifar_df_no_sam = cifar_df[cifar_df['Minimization']=='standard']\n",
    "cifar_df_no_sam['dataset'] = 'CIFAR10' \n",
    "cifar_df_no_sam.groupby('Optimizer').mean().sort_values('Validation accuracy', ascending = False)[['Validation accuracy', 'Sharpness']]"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 373
    },
    "id": "b56I5opBk8eU",
    "outputId": "e51442b6-5c32-495a-cf7a-6b89607d2e4f",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stderr",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  \n"
     ]
    },
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "           Validation accuracy      Sharpness\n",
       "Optimizer                                    \n",
       "SGD                  83.200665   48122.125176\n",
       "AdaBound             83.031331   49213.654733\n",
       "Adam                 82.558998   99841.557479\n",
       "PHB                  81.937999   97502.878728\n",
       "AdaShift             81.262665  103066.032418\n",
       "Adagrad              78.486664  111665.522629"
      ],
      "text/html": [
       "\n",
       "  <div id=\"df-470b1517-4e71-4a55-907b-1bbbdb7e24e0\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>Validation accuracy</th>\n",
       "      <th>Sharpness</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Optimizer</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>SGD</th>\n",
       "      <td>83.200665</td>\n",
       "      <td>48122.125176</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AdaBound</th>\n",
       "      <td>83.031331</td>\n",
       "      <td>49213.654733</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Adam</th>\n",
       "      <td>82.558998</td>\n",
       "      <td>99841.557479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PHB</th>\n",
       "      <td>81.937999</td>\n",
       "      <td>97502.878728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AdaShift</th>\n",
       "      <td>81.262665</td>\n",
       "      <td>103066.032418</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Adagrad</th>\n",
       "      <td>78.486664</td>\n",
       "      <td>111665.522629</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-470b1517-4e71-4a55-907b-1bbbdb7e24e0')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-470b1517-4e71-4a55-907b-1bbbdb7e24e0 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-470b1517-4e71-4a55-907b-1bbbdb7e24e0');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ]
     },
     "metadata": {},
     "execution_count": 162
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "# Plotting distribution of validation accuracies and sharpnesses in the case of FashionMNIST\n",
    "fig, ax = plt.subplots(1,2,figsize=(12,6))\n",
    "fig.suptitle('FashionMNIST', fontsize=20)\n",
    "\n",
    "order = ['SGD', 'AdaBound', 'Adam', 'PHB', 'AdaShift', 'Adagrad']\n",
    "sns.boxplot(data=fmnist_df_no_sam, x='Optimizer', y='Validation accuracy', ax=ax[0], order=order, palette='deep')\n",
    "ax[0].set_ylabel('Test accuracy', fontsize=14)\n",
    "sns.boxplot(data=fmnist_df_no_sam, x='Optimizer', y='Sharpness', ax=ax[1], order=order, palette='deep')\n",
    "ax[1].set_ylabel('Sharpnesses', fontsize=14)\n",
    "\n",
    "if plot_folder != None:\n",
    "    plt.savefig(f'{plot_folder}non_sam_fmnist.png')"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 431
    },
    "id": "p1HNCEYNlbZW",
    "outputId": "f2c4608b-57c5-45e7-8358-0f2adc0f5bab",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 864x432 with 2 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "# Plotting distribution of validation accuracies and sharpnesses in the case of CIFAR10\n",
    "fig, ax = plt.subplots(1,2,figsize=(12,6))\n",
    "fig.suptitle('CIFAR10', fontsize=20)\n",
    "\n",
    "sns.boxplot(data=cifar_df_no_sam, x='Optimizer', y='Validation accuracy', ax=ax[0], order=order, palette='deep')\n",
    "ax[0].set_ylabel('Test accuracy', fontsize=14)\n",
    "sns.boxplot(data=cifar_df_no_sam, x='Optimizer', y='Sharpness', ax=ax[1], order=order, palette='deep')\n",
    "ax[1].set_ylabel('Sharpnesses', fontsize=14)\n",
    "\n",
    "if plot_folder != None:\n",
    "    plt.savefig(f'{plot_folder}non_sam_cifar.png')"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 431
    },
    "id": "P8uGMGR0my7i",
    "outputId": "6c661d0a-06fc-44d4-d13f-1a0c38cf75fc",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 864x432 with 2 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "# Concatenation of dataframes corresponding to different datasets\n",
    "big_df_no_sam = pd.concat((fmnist_df_no_sam,cifar_df_no_sam), axis=0)\n",
    "big_df_no_sam"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 424
    },
    "id": "aQDk-OCgn9X7",
    "outputId": "7c35d3cf-c80d-46c7-ea72-e390eec77efa",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "   Minimization          optim Optimizer  training_accuracy  \\\n",
       "0      standard           Adam      Adam          99.513336   \n",
       "1      standard        SGD 0.1       SGD         100.000000   \n",
       "2      standard       AdaBound  AdaBound         100.000000   \n",
       "3      standard       AdaShift  AdaShift          99.781670   \n",
       "4      standard        Adagrad   Adagrad         100.000000   \n",
       "..          ...            ...       ...                ...   \n",
       "42     standard           Adam      Adam          99.767998   \n",
       "43     standard        Adagrad   Adagrad          99.888000   \n",
       "47     standard       AdaBound  AdaBound          99.709999   \n",
       "48     standard  AdaShift 0.01  AdaShift          99.792000   \n",
       "50     standard   PHB 0.01 0.8       PHB          99.706001   \n",
       "\n",
       "    Validation accuracy  training_loss  val_loss  Generalization gap  \\\n",
       "0             91.979996       0.014862  0.729570            7.533340   \n",
       "1             92.769997       0.000050  0.516581            7.230003   \n",
       "2             92.720001       0.000055  0.537387            7.279999   \n",
       "3             92.389999       0.005874  0.600980            7.391670   \n",
       "4             92.659996       0.000585  0.440312            7.340004   \n",
       "..                  ...            ...       ...                 ...   \n",
       "42            86.290001       0.008549  1.441957           13.477997   \n",
       "43            79.519997       0.009620  0.959188           20.368004   \n",
       "47            84.820000       0.009147  0.913536           14.889999   \n",
       "48            84.650002       0.005884  1.056762           15.141998   \n",
       "50            84.209999       0.009123  0.784922           15.496002   \n",
       "\n",
       "        epoch      Sharpness  Architecture      adaptive       dataset  \n",
       "0        50.0   40364.594347   SimpleBatch      adaptive  FashionMNIST  \n",
       "1        50.0   37819.346180   SimpleBatch  non-adaptive  FashionMNIST  \n",
       "2        50.0   35747.061795   SimpleBatch      adaptive  FashionMNIST  \n",
       "3        50.0   47427.310191   SimpleBatch      adaptive  FashionMNIST  \n",
       "4        50.0   42589.065929   SimpleBatch      adaptive  FashionMNIST  \n",
       "..        ...            ...           ...           ...           ...  \n",
       "42  converged   97314.887747  ComplexBatch      adaptive       CIFAR10  \n",
       "43  converged  130116.135802  ComplexBatch      adaptive       CIFAR10  \n",
       "47  converged   34710.009932  ComplexBatch      adaptive       CIFAR10  \n",
       "48  converged   70562.848900  ComplexBatch      adaptive       CIFAR10  \n",
       "50  converged   53154.373354  ComplexBatch  non-adaptive       CIFAR10  \n",
       "\n",
       "[174 rows x 13 columns]"
      ],
      "text/html": [
       "\n",
       "  <div id=\"df-00eb5efa-3183-4d73-9041-5b0e77b9866b\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>Minimization</th>\n",
       "      <th>optim</th>\n",
       "      <th>Optimizer</th>\n",
       "      <th>training_accuracy</th>\n",
       "      <th>Validation accuracy</th>\n",
       "      <th>training_loss</th>\n",
       "      <th>val_loss</th>\n",
       "      <th>Generalization gap</th>\n",
       "      <th>epoch</th>\n",
       "      <th>Sharpness</th>\n",
       "      <th>Architecture</th>\n",
       "      <th>adaptive</th>\n",
       "      <th>dataset</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>standard</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.513336</td>\n",
       "      <td>91.979996</td>\n",
       "      <td>0.014862</td>\n",
       "      <td>0.729570</td>\n",
       "      <td>7.533340</td>\n",
       "      <td>50.0</td>\n",
       "      <td>40364.594347</td>\n",
       "      <td>SimpleBatch</td>\n",
       "      <td>adaptive</td>\n",
       "      <td>FashionMNIST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>standard</td>\n",
       "      <td>SGD 0.1</td>\n",
       "      <td>SGD</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>92.769997</td>\n",
       "      <td>0.000050</td>\n",
       "      <td>0.516581</td>\n",
       "      <td>7.230003</td>\n",
       "      <td>50.0</td>\n",
       "      <td>37819.346180</td>\n",
       "      <td>SimpleBatch</td>\n",
       "      <td>non-adaptive</td>\n",
       "      <td>FashionMNIST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>standard</td>\n",
       "      <td>AdaBound</td>\n",
       "      <td>AdaBound</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>92.720001</td>\n",
       "      <td>0.000055</td>\n",
       "      <td>0.537387</td>\n",
       "      <td>7.279999</td>\n",
       "      <td>50.0</td>\n",
       "      <td>35747.061795</td>\n",
       "      <td>SimpleBatch</td>\n",
       "      <td>adaptive</td>\n",
       "      <td>FashionMNIST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>standard</td>\n",
       "      <td>AdaShift</td>\n",
       "      <td>AdaShift</td>\n",
       "      <td>99.781670</td>\n",
       "      <td>92.389999</td>\n",
       "      <td>0.005874</td>\n",
       "      <td>0.600980</td>\n",
       "      <td>7.391670</td>\n",
       "      <td>50.0</td>\n",
       "      <td>47427.310191</td>\n",
       "      <td>SimpleBatch</td>\n",
       "      <td>adaptive</td>\n",
       "      <td>FashionMNIST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>standard</td>\n",
       "      <td>Adagrad</td>\n",
       "      <td>Adagrad</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>92.659996</td>\n",
       "      <td>0.000585</td>\n",
       "      <td>0.440312</td>\n",
       "      <td>7.340004</td>\n",
       "      <td>50.0</td>\n",
       "      <td>42589.065929</td>\n",
       "      <td>SimpleBatch</td>\n",
       "      <td>adaptive</td>\n",
       "      <td>FashionMNIST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>standard</td>\n",
       "      <td>Adam</td>\n",
       "      <td>Adam</td>\n",
       "      <td>99.767998</td>\n",
       "      <td>86.290001</td>\n",
       "      <td>0.008549</td>\n",
       "      <td>1.441957</td>\n",
       "      <td>13.477997</td>\n",
       "      <td>converged</td>\n",
       "      <td>97314.887747</td>\n",
       "      <td>ComplexBatch</td>\n",
       "      <td>adaptive</td>\n",
       "      <td>CIFAR10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>standard</td>\n",
       "      <td>Adagrad</td>\n",
       "      <td>Adagrad</td>\n",
       "      <td>99.888000</td>\n",
       "      <td>79.519997</td>\n",
       "      <td>0.009620</td>\n",
       "      <td>0.959188</td>\n",
       "      <td>20.368004</td>\n",
       "      <td>converged</td>\n",
       "      <td>130116.135802</td>\n",
       "      <td>ComplexBatch</td>\n",
       "      <td>adaptive</td>\n",
       "      <td>CIFAR10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>standard</td>\n",
       "      <td>AdaBound</td>\n",
       "      <td>AdaBound</td>\n",
       "      <td>99.709999</td>\n",
       "      <td>84.820000</td>\n",
       "      <td>0.009147</td>\n",
       "      <td>0.913536</td>\n",
       "      <td>14.889999</td>\n",
       "      <td>converged</td>\n",
       "      <td>34710.009932</td>\n",
       "      <td>ComplexBatch</td>\n",
       "      <td>adaptive</td>\n",
       "      <td>CIFAR10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>standard</td>\n",
       "      <td>AdaShift 0.01</td>\n",
       "      <td>AdaShift</td>\n",
       "      <td>99.792000</td>\n",
       "      <td>84.650002</td>\n",
       "      <td>0.005884</td>\n",
       "      <td>1.056762</td>\n",
       "      <td>15.141998</td>\n",
       "      <td>converged</td>\n",
       "      <td>70562.848900</td>\n",
       "      <td>ComplexBatch</td>\n",
       "      <td>adaptive</td>\n",
       "      <td>CIFAR10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>standard</td>\n",
       "      <td>PHB 0.01 0.8</td>\n",
       "      <td>PHB</td>\n",
       "      <td>99.706001</td>\n",
       "      <td>84.209999</td>\n",
       "      <td>0.009123</td>\n",
       "      <td>0.784922</td>\n",
       "      <td>15.496002</td>\n",
       "      <td>converged</td>\n",
       "      <td>53154.373354</td>\n",
       "      <td>ComplexBatch</td>\n",
       "      <td>non-adaptive</td>\n",
       "      <td>CIFAR10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>174 rows × 13 columns</p>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-00eb5efa-3183-4d73-9041-5b0e77b9866b')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-00eb5efa-3183-4d73-9041-5b0e77b9866b button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-00eb5efa-3183-4d73-9041-5b0e77b9866b');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ]
     },
     "metadata": {},
     "execution_count": 164
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "plt.hist(big_df_no_sam[(big_df_no_sam['adaptive']=='adaptive') ]['Validation accuracy'], alpha=0.5, density=True, label='adaptive', bins=50)\n",
    "plt.hist(big_df_no_sam[(big_df_no_sam['adaptive']=='non-adaptive') ]['Validation accuracy'], alpha=0.5, density=True, label='non-adaptive', bins=50)\n",
    "plt.legend()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 285
    },
    "id": "c9_87S-LZxPR",
    "outputId": "75c033d1-1df6-47ba-ddad-f65b88b47b66",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f20911d8fd0>"
      ]
     },
     "metadata": {},
     "execution_count": 178
    },
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "# Testing if validation accuracies in the case of adaptive and non-adaptive algorithms are drawn from the same distribution\n",
    "# Since p-value is high - this hypothesis can not be rejected\n",
    "x = big_df_no_sam[(big_df_no_sam['adaptive']=='adaptive') ]['Validation accuracy'].values\n",
    "y =  big_df_no_sam[(big_df_no_sam['adaptive']=='non-adaptive') ]['Validation accuracy'].values\n",
    "scipy.stats.ks_2samp(x,y)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "6S2ilM1BcEfQ",
    "outputId": "833f44f5-8d01-4c81-ed39-37b890b3e941",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "Ks_2sampResult(statistic=0.12894736842105264, pvalue=0.48342957087214533)"
      ]
     },
     "metadata": {},
     "execution_count": 145
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "# Testing if sharpnesses in the case of adaptive and non-adaptive algorithms are drawn from the same distribution\n",
    "# Since p-value is high - this hypothesis can not be rejected\n",
    "x = big_df_no_sam[(big_df_no_sam['adaptive']=='adaptive') ]['Sharpness'].values\n",
    "y =  big_df_no_sam[(big_df_no_sam['adaptive']=='non-adaptive') ]['Sharpness'].values\n",
    "scipy.stats.ks_2samp(x,y)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "arB7WFZwcrDg",
    "outputId": "52a3e885-6eec-4905-8568-d4a41db44a32",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "Ks_2sampResult(statistic=0.11403508771929824, pvalue=0.636766610230814)"
      ]
     },
     "metadata": {},
     "execution_count": 167
    }
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Data analysis.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}