Figure_2_deep_noisy_signal_prop.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Deep signal propagation with and without noise.\n",
    "\n",
    "In this notebook, we provide code to reproduce Figure 2 in the paper. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# imports\n",
    "import os, sys, pickle\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "from matplotlib.lines import Line2D\n",
    "from matplotlib.collections import LineCollection\n",
    "from matplotlib.colors import ListedColormap\n",
    "import matplotlib.lines as mlines\n",
    "\n",
    "# custom import\n",
    "from src.numpy_simulation import *\n",
    "from src.viz import get_colours\n",
    "from src.utils import load_experiment\n",
    "from src.theory import depth\n",
    "from src.theory import critical_point\n",
    "\n",
    "# plot settings\n",
    "import matplotlib as mpl\n",
    "mpl.rcParams['font.size'] = 100\n",
    "sns.set_context(\"paper\", font_scale=2)\n",
    "plt.rcParams['pdf.fonttype'] = 42\n",
    "plt.rcParams['text.usetex'] = True\n",
    "plt.rcParams['text.latex.unicode'] = True\n",
    "plt.rcParams['image.cmap'] = 'viridis'\n",
    "\n",
    "# results directory\n",
    "relative_results_dir = \"results\"\n",
    "results_dir = os.path.join(relative_results_dir)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## --- Noisy variance map: Theory vs simulation ---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "####### EXPERIMENT: dist:  none ;  None :  None ; activation:  tanh ; initialisation:  xavier ##############\n",
      "qmap calculations...\n",
      "Calculating Theory:\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "11bbc215c2ed480c9a269a841b308a8b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='input lengths', max=25), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "852fdd1ba31c4b0ebccd94b461647c41",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='layers', max=14), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Simulating network\n",
      "Single layer sims...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c15a68f500bd4f5582bb055facbd8696",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='generate inputs', max=30), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "26b43b32f8054f6c87e943b5e447ed81",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='network inits', max=50), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "####### EXPERIMENT: dist:  add gauss ;  std :  1 ; activation:  tanh ; initialisation:  xavier ##############\n",
      "qmap calculations...\n",
      "Calculating Theory:\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "71c6bd41b27146b18ea4e49d5fb6b487",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='input lengths', max=25), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b6efeafa882d404e8d2626556414b308",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='layers', max=14), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Simulating network\n",
      "Single layer sims...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9a9bf6c6faca4db69ea292d33b13362e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='generate inputs', max=30), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dd85477df5ff4c4480816fe8b2d921bb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='network inits', max=50), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "####### EXPERIMENT: dist:  none ;  None :  None ; activation:  relu ; initialisation:  he ##############\n",
      "qmap calculations...\n",
      "Calculating Theory:\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2b64fc2a20b6447fb5c066524353c07b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='input lengths', max=25), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "978dfda4a5e34a9c9311cee1e27a0ff7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='layers', max=14), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Simulating network\n",
      "Single layer sims...\n",
      "Setting activation to ReLU...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "66f87126bbd842b08af656ee5eb6bac7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='generate inputs', max=30), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e3bdc4ded76a4e0b8d523d2d85af44c2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='network inits', max=50), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "####### EXPERIMENT: dist:  add gauss ;  std :  1 ; activation:  relu ; initialisation:  he ##############\n",
      "qmap calculations...\n",
      "Calculating Theory:\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "34d17f8f834f4d44a34aae72db5eb0d8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='input lengths', max=25), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fcda86fdc8d44a87b490d364595a1978",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='layers', max=14), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Simulating network\n",
      "Single layer sims...\n",
      "Setting activation to ReLU...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "60eed132e8f84cc2ab4216b7c95424cc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='generate inputs', max=30), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a1a93edd441f49b990f36fc433acbb6d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='network inits', max=50), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "experiments = [\n",
    "    {\"dist\": \"none\", \"noise\": (None, None), \"act\":\"tanh\", \"init\":\"xavier\"}, \n",
    "    {\"dist\": \"add gauss\", \"noise\": ('std', 1), \"act\":\"tanh\", \"init\":\"xavier\"},\n",
    "    {\"dist\": \"none\", \"noise\": (None, None), \"act\":\"relu\", \"init\":\"he\"},\n",
    "    {\"dist\": \"add gauss\", \"noise\": ('std', 1), \"act\":\"relu\", \"init\":\"he\"}\n",
    "]\n",
    "\n",
    "for i, experiment in enumerate(experiments):\n",
    "    dist = experiment['dist']\n",
    "    noise = experiment['noise']\n",
    "    act = experiment['act']\n",
    "    init = experiment['init']\n",
    "    \n",
    "    # run simulations for scenario\n",
    "    noisy_signal_prop_simulations(dist, noise, act, init, seed=i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Dictionary for data that needs to be extracted\n",
    "\n",
    "tests = [{\n",
    "        \"distributions\": [{\"dist\": \"none\"}],\n",
    "        \"activations\": [\"tanh\"],\n",
    "        \"inits\": [\"xavier\"]\n",
    "    }, {\n",
    "        \"distributions\": [{\"dist\": \"add gauss\", \"std\": 1}],\n",
    "        \"activations\": [\"tanh\"],\n",
    "        \"inits\": [\"xavier\"]\n",
    "    }, {\n",
    "        \"distributions\": [{\"dist\": \"none\"}],\n",
    "        \"activations\": [\"relu\"],\n",
    "        \"inits\": [\"he\"]\n",
    "    }, {\n",
    "        \"distributions\": [{\"dist\": \"add gauss\", \"std\": 1}],\n",
    "        \"activations\": [\"relu\"],\n",
    "        \"inits\": [\"he\"]\n",
    "    }]\n",
    "\n",
    "\n",
    "\n",
    "############################################################################\n",
    "# q - length / variance plots\n",
    "############################################################################\n",
    "nq = 30\n",
    "qmax = 15\n",
    "qrange = np.linspace(0, qmax, nq)\n",
    "widxs = [0]\n",
    "bidxs = [0]\n",
    "n_hidden_layers = 16\n",
    "\n",
    "n_tests = len(tests)\n",
    "#pal = get_colours(10, 7)\n",
    "pal = get_colours(6, 7)[2:]\n",
    "test_data = []\n",
    "for i, test in enumerate(tests):\n",
    "    test_data.append(load_experiment(test, [\"q_maps\", \"single_layer_qmap_sim\", \"multi_layer_qmap_sim\"], \"results\"))\n",
    "\n",
    "fig = plt.figure(figsize=(16, 6))\n",
    "\n",
    "gs = plt.GridSpec(1, 2)\n",
    "ax1 = plt.subplot(gs[0, 0])\n",
    "ax2 = plt.subplot(gs[0, 1])\n",
    "\n",
    "# Add unity line\n",
    "ax1.plot((0, qmax), (0, qmax), '--', color='k', zorder=900, dashes=(4, 8), label=\"Identity line\")\n",
    "ax1.set_xlim(0, qmax)\n",
    "ax1.set_ylim(0, qmax)\n",
    "ax1.set_xlabel('Input variance ($q^{l-1})$')\n",
    "ax1.set_ylabel('Output variance ($q^l$)')\n",
    "ax1.set_title(\"Iterative variance map\")\n",
    "\n",
    "# axis 2\n",
    "ax2.set_xlim(0, qmax-1) #n_hidden_layers-1)\n",
    "ax2.set_ylim(0, qmax)\n",
    "ax2.set_xlabel('Layer ($l$)')\n",
    "ax2.set_ylabel('Variance ($q^{l})$')\n",
    "ax2.set_title(\"Dynamics of $q$\")\n",
    "\n",
    "nn = len(test_data)\n",
    "col_i = 0\n",
    "shade_i = 5\n",
    "for test, attr in zip(test_data, tests):\n",
    "    for dist in attr[\"distributions\"]:\n",
    "        for act in attr[\"activations\"]:\n",
    "            if dist['dist'] == \"none\":\n",
    "                if act == \"tanh\":\n",
    "                    label = \"tanh - None\"\n",
    "                    col_i = 0\n",
    "                else:\n",
    "                    label = \"ReLU - None\"\n",
    "                    col_i = 2\n",
    "            elif \"gauss\" in dist['dist']: \n",
    "                if act == \"tanh\":\n",
    "                    label = \"tanh - Add Gauss $(\\sigma^2_\\epsilon = $ \" + str(dist['std']) + \")\"\n",
    "                    col_i = 1\n",
    "                else:\n",
    "                    label = \"ReLU - Add Gauss $(\\sigma^2_\\epsilon = $ \" + str(dist['std']) + \")\"\n",
    "                    col_i = 3\n",
    "            for init in attr[\"inits\"]:\n",
    "                dashes = (None, None)\n",
    "                \n",
    "                \n",
    "                # extract test data\n",
    "                qmaps = test[dist['dist']][act][init]['q_maps']['qmaps']\n",
    "                single_layer_sims = test[dist['dist']][act][init]['single_layer_qmap_sim']\n",
    "                multi_layer_sims = test[dist['dist']][act][init]['multi_layer_qmap_sim']['qmaps_sim']\n",
    "                multi_layer_theory = test[dist['dist']][act][init]['multi_layer_qmap_sim']['theory']\n",
    "                \n",
    "\n",
    "                ############################################################################\n",
    "                # left\n",
    "                ############################################################################\n",
    "                for w, b in zip(widxs, bidxs):\n",
    "\n",
    "                    # plot means of simulation (as dots)\n",
    "                    mu = single_layer_sims[w, b].mean(-1).mean(-2)\n",
    "                    ax1.plot(qrange, mu, w, b, marker='o', ls='none', markersize=1, alpha=0.9, zorder=0, c=pal[col_i][shade_i])\n",
    "\n",
    "                    # add confidence interval around simulation\n",
    "                    std = single_layer_sims[w, b].mean(-1).std(-2)\n",
    "                    ax1.fill_between(qrange, mu-std, mu+std, alpha=0.4, label='_nolegend_', color=pal[col_i][3])\n",
    "\n",
    "                    # theory line\n",
    "                    ax1.plot(qrange, qmaps[0, 0, :, 1], c=pal[col_i][shade_i], label=label, dashes=dashes)\n",
    "                    \n",
    "\n",
    "                ############################################################################\n",
    "                # right\n",
    "                ############################################################################\n",
    "                q = 1\n",
    "                xx = np.arange(multi_layer_sims.shape[-2])\n",
    "                for w, b in zip(widxs, bidxs):\n",
    "                    # confidence intervals\n",
    "                    mu = multi_layer_sims[w, b].mean(axis=-1).mean(axis=0)\n",
    "                    std = multi_layer_sims[w, b].mean(axis=-1).std(axis=0)\n",
    "\n",
    "                    # plot theory\n",
    "                    ax2.plot(multi_layer_theory, c=pal[col_i][shade_i], label=\"Theory\")\n",
    "\n",
    "                    # plot the simulation\n",
    "                    ax2.fill_between(xx, mu-std, mu+std, alpha=0.2, label='_nolegend_', color=pal[col_i][3])\n",
    "\n",
    "                    # dots for mean\n",
    "                    ax2.plot(xx, mu, 'o', markersize=4, alpha=0.9, color=pal[col_i][shade_i], label=\"Simulation\")\n",
    "\n",
    "##############\n",
    "# add labels #\n",
    "##############\n",
    "fig.text(0.02, 0.95, \"(a)\", fontsize=20)\n",
    "fig.text(0.52, 0.95, \"(b)\", fontsize=20)\n",
    "\n",
    "leg = ax1.legend(prop={'size': 15})\n",
    "\n",
    "# set the linewidth of each legend object\n",
    "for legobj in leg.legendHandles:\n",
    "    legobj.set_linewidth(3.0)\n",
    "\n",
    "plt.gcf().tight_layout()\n",
    "plt.show()\n",
    "# plt.savefig(\"tanh.pdf\", dpi=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$\\textit{Deep signal propagation with and without noise}$. $\\textbf{(a)}$: Iterative variance map. $\\textbf{(b)}$: Variance dynamics during forward signal propagation. In (a) and (b), lines correspond to theoretical predictions\n",
    "\tand points to numerical simulations (means over $50$ runs with shaded one standard deviation bounds), for noiseless tanh (yellow) and noiseless ReLU (purple) networks,\n",
    "\tas well as for noisy tanh (red) and noisy ReLU (brown) networks regularised using additive noise from a standard Gaussian.\n",
    "\tBoth tanh networks use $(\\sigma_w , \\sigma_b ) = (1, 0)$, the Xavier initialisation (Glorot and Bengio, 2010), while the ReLU networks use $(\\sigma_w, \\sigma_b) = (\\sqrt{2}, 0)$ the He initialisation (He et al., 2015). In all experiments, network layers consisted of $1000$ hidden units."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}