double_opponency.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Double opponency\n",
    "\n",
    "This notebook plots the distribution of double opponent cells in different layers of our model as a function of bottleneck size. It corresponds to Figure 2(c) in the paper.\n",
    "\n",
    "**Note**: The easiest way to use this is as a colab notebook, which allows you to dive in with no setup.\n",
    "\n",
    "## Load Dependencies - Colab Only"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from os.path import exists\n",
    "if not exists('opponency.zip'):\n",
    "    !wget -O opponency.zip https://github.com/ecs-vlc/opponency/archive/master.zip\n",
    "    !unzip -qq opponency.zip\n",
    "    !mv opponency-master/* ./\n",
    "    !rm -r opponency-master"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generate Plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x180 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import rc\n",
    "import matplotlib.font_manager\n",
    "rc('font',**{'family':'serif','serif':['Computer Modern Roman'],'size':13})\n",
    "rc('text', usetex=True)\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "from statistics import load\n",
    "# import seaborn as sns\n",
    "# sns.set()\n",
    "\n",
    "spatial = pd.read_pickle('logs/spatial.pd')\n",
    "spectral = pd.read_pickle('logs/devalois.pd')\n",
    "\n",
    "plt.figure(figsize=(3,2.5))\n",
    "opps = load.double(spectral, spatial)\n",
    "\n",
    "retina2 = opps[opps['layer'] == 'retina_relu2']\n",
    "ventral0 = opps[opps['layer'] == 'ventral_relu0']\n",
    "ventral1 = opps[opps['layer'] == 'ventral_relu1']\n",
    "\n",
    "plt.plot(retina2['n_bn'], retina2['mean_rel_amount'], label='Retina 2', linestyle=':')\n",
    "plt.fill_between(\n",
    "    retina2['n_bn'],\n",
    "    retina2['mean_rel_amount'] + retina2['std_rel_amount'],\n",
    "    retina2['mean_rel_amount'] - retina2['std_rel_amount'],\n",
    "    alpha=0.1\n",
    ")\n",
    "\n",
    "plt.plot(ventral0['n_bn'], ventral0['mean_rel_amount'], label='Ventral 1', linestyle='--')\n",
    "plt.fill_between(\n",
    "    ventral0['n_bn'],\n",
    "    ventral0['mean_rel_amount'] + ventral0['std_rel_amount'],\n",
    "    ventral0['mean_rel_amount'] - ventral0['std_rel_amount'],\n",
    "    alpha=0.1\n",
    ")\n",
    "\n",
    "plt.plot(ventral1['n_bn'], ventral1['mean_rel_amount'], label='Ventral 2', linestyle='-.')\n",
    "plt.fill_between(\n",
    "    ventral1['n_bn'],\n",
    "    ventral1['mean_rel_amount'] + ventral1['std_rel_amount'],\n",
    "    ventral1['mean_rel_amount'] - ventral1['std_rel_amount'],\n",
    "    alpha=0.1\n",
    ")\n",
    "\n",
    "plt.xlim(1, 32)\n",
    "plt.ylim(0, 1)\n",
    "\n",
    "plt.xlabel('Bottleneck Size')\n",
    "plt.ylabel('Percentage')\n",
    "plt.savefig('figures/double_opponent.pdf', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x180 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "spatial = pd.read_pickle('logs/spatial.pd')\n",
    "spectral = pd.read_pickle('logs/devalois.pd')\n",
    "\n",
    "plt.figure(figsize=(3,2.5))\n",
    "opps = load.double(spectral, spatial, groupby='d_vvs')\n",
    "\n",
    "retina2 = opps[opps['layer'] == 'retina_relu2']\n",
    "ventral0 = opps[opps['layer'] == 'ventral_relu0']\n",
    "ventral1 = opps[opps['layer'] == 'ventral_relu1']\n",
    "\n",
    "plt.plot(retina2['d_vvs'], retina2['mean_rel_amount'], label='Retina 2', linestyle=':')\n",
    "plt.fill_between(\n",
    "    retina2['d_vvs'],\n",
    "    retina2['mean_rel_amount'] + retina2['std_rel_amount'],\n",
    "    retina2['mean_rel_amount'] - retina2['std_rel_amount'],\n",
    "    alpha=0.1\n",
    ")\n",
    "\n",
    "plt.plot(ventral0['d_vvs'], ventral0['mean_rel_amount'], label='Ventral 1', linestyle='--')\n",
    "plt.fill_between(\n",
    "    ventral0['d_vvs'],\n",
    "    ventral0['mean_rel_amount'] + ventral0['std_rel_amount'],\n",
    "    ventral0['mean_rel_amount'] - ventral0['std_rel_amount'],\n",
    "    alpha=0.1\n",
    ")\n",
    "\n",
    "plt.plot(ventral1['d_vvs'], ventral1['mean_rel_amount'], label='Ventral 2', linestyle='-.')\n",
    "plt.fill_between(\n",
    "    ventral1['d_vvs'],\n",
    "    ventral1['mean_rel_amount'] + ventral1['std_rel_amount'],\n",
    "    ventral1['mean_rel_amount'] - ventral1['std_rel_amount'],\n",
    "    alpha=0.1\n",
    ")\n",
    "\n",
    "plt.xlim(0, 4)\n",
    "plt.ylim(0, 1)\n",
    "plt.legend(frameon=False)\n",
    "plt.xlabel('Ventral Depth')\n",
    "plt.ylabel('Percentage')\n",
    "plt.savefig('figures/double_opponent_depth.pdf', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}