ch11_dnn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ch11 training deep neural network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "The tensorboard extension is already loaded. To reload it, use:\n  %reload_ext tensorboard\n"
    }
   ],
   "source": [
    "# Python ≥3.5 is required\n",
    "import sys\n",
    "assert sys.version_info >= (3, 5)\n",
    "\n",
    "# Scikit-Learn ≥0.20 is required\n",
    "import sklearn\n",
    "assert sklearn.__version__ >= \"0.20\"\n",
    "\n",
    "try:\n",
    "    # %tensorflow_version only exists in Colab.\n",
    "    %tensorflow_version 2.x\n",
    "except Exception:\n",
    "    pass\n",
    "\n",
    "# TensorFlow ≥2.0 is required\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "assert tf.__version__ >= \"2.0\"\n",
    "\n",
    "%load_ext tensorboard\n",
    "\n",
    "# Common imports\n",
    "import numpy as np\n",
    "import os\n",
    "\n",
    "# to make this notebook's output stable across runs\n",
    "np.random.seed(42)\n",
    "\n",
    "# To plot pretty figures\n",
    "%matplotlib inline\n",
    "import matplotlib as mpl\n",
    "import matplotlib.pyplot as plt\n",
    "mpl.rc('axes', labelsize=14)\n",
    "mpl.rc('xtick', labelsize=12)\n",
    "mpl.rc('ytick', labelsize=12)\n",
    "\n",
    "# Where to save the figures\n",
    "PROJECT_ROOT_DIR = \".\"\n",
    "CHAPTER_ID = \"deep\"\n",
    "IMAGES_PATH = os.path.join(PROJECT_ROOT_DIR, \"images\", CHAPTER_ID)\n",
    "os.makedirs(IMAGES_PATH, exist_ok=True)\n",
    "\n",
    "def save_fig(fig_id, tight_layout=True, fig_extension=\"png\", resolution=300):\n",
    "    path = os.path.join(IMAGES_PATH, fig_id + \".\" + fig_extension)\n",
    "    print(\"Saving figure\", fig_id)\n",
    "    if tight_layout:\n",
    "        plt.tight_layout()\n",
    "    plt.savefig(path, format=fig_extension, dpi=resolution)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## vanishing/exploding gradient problem"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "def logit(z):\n",
    "    return 1 / (1 + np.exp(-z))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Saving figure sigmoid_saturation_plot\n"
    },
    {
     "output_type": "display_data",
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/svg+xml": "<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n  \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n<!-- Created with matplotlib (https://matplotlib.org/) -->\n<svg height=\"279.051563pt\" version=\"1.1\" viewBox=\"0 0 425.019434 279.051563\" width=\"425.019434pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n <defs>\n  <style type=\"text/css\">\n*{stroke-linecap:butt;stroke-linejoin:round;}\n  </style>\n </defs>\n <g id=\"figure_1\">\n  <g id=\"patch_1\">\n   <path d=\"M 0 279.051563 \nL 425.019434 279.051563 \nL 425.019434 0 \nL 0 0 \nz\n\" style=\"fill:none;\"/>\n  </g>\n  <g id=\"axes_1\">\n   <g id=\"patch_2\">\n    <path d=\"M 43.339375 253.237813 \nL 417.819434 253.237813 \nL 417.819434 23.837812 \nL 43.339375 23.837812 \nz\n\" style=\"fill:#ffffff;\"/>\n   </g>\n   <g id=\"matplotlib.axis_1\">\n    <g id=\"xtick_1\">\n     <g id=\"line2d_1\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 80.787381 253.237813 \nL 80.787381 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_2\">\n      <defs>\n       <path d=\"M 0 0 \nL 0 3.5 \n\" id=\"m5373cede86\" style=\"stroke:#000000;stroke-width:0.8;\"/>\n      </defs>\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"80.787381\" xlink:href=\"#m5373cede86\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_1\">\n      <!-- −4 -->\n      <defs>\n       <path d=\"M 10.59375 35.5 \nL 73.1875 35.5 \nL 73.1875 27.203125 \nL 10.59375 27.203125 \nz\n\" id=\"DejaVuSans-8722\"/>\n       <path d=\"M 37.796875 64.3125 \nL 12.890625 25.390625 \nL 37.796875 25.390625 \nz\nM 35.203125 72.90625 \nL 47.609375 72.90625 \nL 47.609375 25.390625 \nL 58.015625 25.390625 \nL 58.015625 17.1875 \nL 47.609375 17.1875 \nL 47.609375 0 \nL 37.796875 0 \nL 37.796875 17.1875 \nL 4.890625 17.1875 \nL 4.890625 26.703125 \nz\n\" id=\"DejaVuSans-52\"/>\n      </defs>\n      <g transform=\"translate(71.942068 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_2\">\n     <g id=\"line2d_3\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 155.683393 253.237813 \nL 155.683393 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_4\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"155.683393\" xlink:href=\"#m5373cede86\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_2\">\n      <!-- −2 -->\n      <defs>\n       <path d=\"M 19.1875 8.296875 \nL 53.609375 8.296875 \nL 53.609375 0 \nL 7.328125 0 \nL 7.328125 8.296875 \nQ 12.9375 14.109375 22.625 23.890625 \nQ 32.328125 33.6875 34.8125 36.53125 \nQ 39.546875 41.84375 41.421875 45.53125 \nQ 43.3125 49.21875 43.3125 52.78125 \nQ 43.3125 58.59375 39.234375 62.25 \nQ 35.15625 65.921875 28.609375 65.921875 \nQ 23.96875 65.921875 18.8125 64.3125 \nQ 13.671875 62.703125 7.8125 59.421875 \nL 7.8125 69.390625 \nQ 13.765625 71.78125 18.9375 73 \nQ 24.125 74.21875 28.421875 74.21875 \nQ 39.75 74.21875 46.484375 68.546875 \nQ 53.21875 62.890625 53.21875 53.421875 \nQ 53.21875 48.921875 51.53125 44.890625 \nQ 49.859375 40.875 45.40625 35.40625 \nQ 44.1875 33.984375 37.640625 27.21875 \nQ 31.109375 20.453125 19.1875 8.296875 \nz\n\" id=\"DejaVuSans-50\"/>\n      </defs>\n      <g transform=\"translate(146.83808 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_3\">\n     <g id=\"line2d_5\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 230.579405 253.237813 \nL 230.579405 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_6\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"230.579405\" xlink:href=\"#m5373cede86\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_3\">\n      <!-- 0 -->\n      <defs>\n       <path d=\"M 31.78125 66.40625 \nQ 24.171875 66.40625 20.328125 58.90625 \nQ 16.5 51.421875 16.5 36.375 \nQ 16.5 21.390625 20.328125 13.890625 \nQ 24.171875 6.390625 31.78125 6.390625 \nQ 39.453125 6.390625 43.28125 13.890625 \nQ 47.125 21.390625 47.125 36.375 \nQ 47.125 51.421875 43.28125 58.90625 \nQ 39.453125 66.40625 31.78125 66.40625 \nz\nM 31.78125 74.21875 \nQ 44.046875 74.21875 50.515625 64.515625 \nQ 56.984375 54.828125 56.984375 36.375 \nQ 56.984375 17.96875 50.515625 8.265625 \nQ 44.046875 -1.421875 31.78125 -1.421875 \nQ 19.53125 -1.421875 13.0625 8.265625 \nQ 6.59375 17.96875 6.59375 36.375 \nQ 6.59375 54.828125 13.0625 64.515625 \nQ 19.53125 74.21875 31.78125 74.21875 \nz\n\" id=\"DejaVuSans-48\"/>\n      </defs>\n      <g transform=\"translate(226.761905 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_4\">\n     <g id=\"line2d_7\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 305.475416 253.237813 \nL 305.475416 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_8\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"305.475416\" xlink:href=\"#m5373cede86\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_4\">\n      <!-- 2 -->\n      <g transform=\"translate(301.657916 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_5\">\n     <g id=\"line2d_9\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 380.371428 253.237813 \nL 380.371428 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_10\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"380.371428\" xlink:href=\"#m5373cede86\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_5\">\n      <!-- 4 -->\n      <g transform=\"translate(376.553928 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n   </g>\n   <g id=\"matplotlib.axis_2\">\n    <g id=\"ytick_1\">\n     <g id=\"line2d_11\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 253.237813 \nL 417.819434 253.237813 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_12\">\n      <defs>\n       <path d=\"M 0 0 \nL -3.5 0 \n\" id=\"mb5fea8535b\" style=\"stroke:#000000;stroke-width:0.8;\"/>\n      </defs>\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"43.339375\" xlink:href=\"#mb5fea8535b\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_6\">\n      <!-- −0.2 -->\n      <defs>\n       <path d=\"M 10.6875 12.40625 \nL 21 12.40625 \nL 21 0 \nL 10.6875 0 \nz\n\" id=\"DejaVuSans-46\"/>\n      </defs>\n      <g transform=\"translate(7.2 257.796875)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-48\"/>\n       <use x=\"147.412109\" xlink:href=\"#DejaVuSans-46\"/>\n       <use x=\"179.199219\" xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_2\">\n     <g id=\"line2d_13\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 220.466384 \nL 417.819434 220.466384 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_14\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"43.339375\" xlink:href=\"#mb5fea8535b\" y=\"220.466384\"/>\n      </g>\n     </g>\n     <g id=\"text_7\">\n      <!-- 0.0 -->\n      <g transform=\"translate(17.255625 225.025446)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n       <use x=\"63.623047\" xlink:href=\"#DejaVuSans-46\"/>\n       <use x=\"95.410156\" xlink:href=\"#DejaVuSans-48\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_3\">\n     <g id=\"line2d_15\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 187.694955 \nL 417.819434 187.694955 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_16\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"43.339375\" xlink:href=\"#mb5fea8535b\" y=\"187.694955\"/>\n      </g>\n     </g>\n     <g id=\"text_8\">\n      <!-- 0.2 -->\n      <g transform=\"translate(17.255625 192.254018)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n       <use x=\"63.623047\" xlink:href=\"#DejaVuSans-46\"/>\n       <use x=\"95.410156\" xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_4\">\n     <g id=\"line2d_17\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 154.923527 \nL 417.819434 154.923527 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_18\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"43.339375\" xlink:href=\"#mb5fea8535b\" y=\"154.923527\"/>\n      </g>\n     </g>\n     <g id=\"text_9\">\n      <!-- 0.4 -->\n      <g transform=\"translate(17.255625 159.482589)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n       <use x=\"63.623047\" xlink:href=\"#DejaVuSans-46\"/>\n       <use x=\"95.410156\" xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_5\">\n     <g id=\"line2d_19\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 122.152098 \nL 417.819434 122.152098 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_20\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"43.339375\" xlink:href=\"#mb5fea8535b\" y=\"122.152098\"/>\n      </g>\n     </g>\n     <g id=\"text_10\">\n      <!-- 0.6 -->\n      <defs>\n       <path d=\"M 33.015625 40.375 \nQ 26.375 40.375 22.484375 35.828125 \nQ 18.609375 31.296875 18.609375 23.390625 \nQ 18.609375 15.53125 22.484375 10.953125 \nQ 26.375 6.390625 33.015625 6.390625 \nQ 39.65625 6.390625 43.53125 10.953125 \nQ 47.40625 15.53125 47.40625 23.390625 \nQ 47.40625 31.296875 43.53125 35.828125 \nQ 39.65625 40.375 33.015625 40.375 \nz\nM 52.59375 71.296875 \nL 52.59375 62.3125 \nQ 48.875 64.0625 45.09375 64.984375 \nQ 41.3125 65.921875 37.59375 65.921875 \nQ 27.828125 65.921875 22.671875 59.328125 \nQ 17.53125 52.734375 16.796875 39.40625 \nQ 19.671875 43.65625 24.015625 45.921875 \nQ 28.375 48.1875 33.59375 48.1875 \nQ 44.578125 48.1875 50.953125 41.515625 \nQ 57.328125 34.859375 57.328125 23.390625 \nQ 57.328125 12.15625 50.6875 5.359375 \nQ 44.046875 -1.421875 33.015625 -1.421875 \nQ 20.359375 -1.421875 13.671875 8.265625 \nQ 6.984375 17.96875 6.984375 36.375 \nQ 6.984375 53.65625 15.1875 63.9375 \nQ 23.390625 74.21875 37.203125 74.21875 \nQ 40.921875 74.21875 44.703125 73.484375 \nQ 48.484375 72.75 52.59375 71.296875 \nz\n\" id=\"DejaVuSans-54\"/>\n      </defs>\n      <g transform=\"translate(17.255625 126.711161)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n       <use x=\"63.623047\" xlink:href=\"#DejaVuSans-46\"/>\n       <use x=\"95.410156\" xlink:href=\"#DejaVuSans-54\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_6\">\n     <g id=\"line2d_21\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 89.38067 \nL 417.819434 89.38067 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_22\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"43.339375\" xlink:href=\"#mb5fea8535b\" y=\"89.38067\"/>\n      </g>\n     </g>\n     <g id=\"text_11\">\n      <!-- 0.8 -->\n      <defs>\n       <path d=\"M 31.78125 34.625 \nQ 24.75 34.625 20.71875 30.859375 \nQ 16.703125 27.09375 16.703125 20.515625 \nQ 16.703125 13.921875 20.71875 10.15625 \nQ 24.75 6.390625 31.78125 6.390625 \nQ 38.8125 6.390625 42.859375 10.171875 \nQ 46.921875 13.96875 46.921875 20.515625 \nQ 46.921875 27.09375 42.890625 30.859375 \nQ 38.875 34.625 31.78125 34.625 \nz\nM 21.921875 38.8125 \nQ 15.578125 40.375 12.03125 44.71875 \nQ 8.5 49.078125 8.5 55.328125 \nQ 8.5 64.0625 14.71875 69.140625 \nQ 20.953125 74.21875 31.78125 74.21875 \nQ 42.671875 74.21875 48.875 69.140625 \nQ 55.078125 64.0625 55.078125 55.328125 \nQ 55.078125 49.078125 51.53125 44.71875 \nQ 48 40.375 41.703125 38.8125 \nQ 48.828125 37.15625 52.796875 32.3125 \nQ 56.78125 27.484375 56.78125 20.515625 \nQ 56.78125 9.90625 50.3125 4.234375 \nQ 43.84375 -1.421875 31.78125 -1.421875 \nQ 19.734375 -1.421875 13.25 4.234375 \nQ 6.78125 9.90625 6.78125 20.515625 \nQ 6.78125 27.484375 10.78125 32.3125 \nQ 14.796875 37.15625 21.921875 38.8125 \nz\nM 18.3125 54.390625 \nQ 18.3125 48.734375 21.84375 45.5625 \nQ 25.390625 42.390625 31.78125 42.390625 \nQ 38.140625 42.390625 41.71875 45.5625 \nQ 45.3125 48.734375 45.3125 54.390625 \nQ 45.3125 60.0625 41.71875 63.234375 \nQ 38.140625 66.40625 31.78125 66.40625 \nQ 25.390625 66.40625 21.84375 63.234375 \nQ 18.3125 60.0625 18.3125 54.390625 \nz\n\" id=\"DejaVuSans-56\"/>\n      </defs>\n      <g transform=\"translate(17.255625 93.939732)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n       <use x=\"63.623047\" xlink:href=\"#DejaVuSans-46\"/>\n       <use x=\"95.410156\" xlink:href=\"#DejaVuSans-56\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_7\">\n     <g id=\"line2d_23\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 56.609241 \nL 417.819434 56.609241 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_24\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"43.339375\" xlink:href=\"#mb5fea8535b\" y=\"56.609241\"/>\n      </g>\n     </g>\n     <g id=\"text_12\">\n      <!-- 1.0 -->\n      <defs>\n       <path d=\"M 12.40625 8.296875 \nL 28.515625 8.296875 \nL 28.515625 63.921875 \nL 10.984375 60.40625 \nL 10.984375 69.390625 \nL 28.421875 72.90625 \nL 38.28125 72.90625 \nL 38.28125 8.296875 \nL 54.390625 8.296875 \nL 54.390625 0 \nL 12.40625 0 \nz\n\" id=\"DejaVuSans-49\"/>\n      </defs>\n      <g transform=\"translate(17.255625 61.168304)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-49\"/>\n       <use x=\"63.623047\" xlink:href=\"#DejaVuSans-46\"/>\n       <use x=\"95.410156\" xlink:href=\"#DejaVuSans-48\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_8\">\n     <g id=\"line2d_25\">\n      <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 23.837812 \nL 417.819434 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_26\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"43.339375\" xlink:href=\"#mb5fea8535b\" y=\"23.837812\"/>\n      </g>\n     </g>\n     <g id=\"text_13\">\n      <!-- 1.2 -->\n      <g transform=\"translate(17.255625 28.396875)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-49\"/>\n       <use x=\"63.623047\" xlink:href=\"#DejaVuSans-46\"/>\n       <use x=\"95.410156\" xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n   </g>\n   <g id=\"line2d_27\">\n    <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 220.466384 \nL 417.819434 220.466384 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"line2d_28\">\n    <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 56.609241 \nL 417.819434 56.609241 \n\" style=\"fill:none;stroke:#000000;stroke-dasharray:5.55,2.4;stroke-dashoffset:0;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"line2d_29\">\n    <path clip-path=\"url(#pe870947a28)\" d=\"M 230.579405 253.237813 \nL 230.579405 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"line2d_30\">\n    <path clip-path=\"url(#pe870947a28)\" d=\"M 101.212868 280.051563 \nL 358.139613 -1 \n\" style=\"fill:none;stroke:#008000;stroke-dasharray:5.55,2.4;stroke-dashoffset:0;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"line2d_31\">\n    <path clip-path=\"url(#pe870947a28)\" d=\"M 43.339375 219.369712 \nL 58.39385 218.832462 \nL 71.566515 218.15338 \nL 82.857371 217.3549 \nL 92.266418 216.487718 \nL 101.675465 215.386486 \nL 109.202702 214.298117 \nL 116.729939 212.987548 \nL 124.257177 211.414508 \nL 129.902605 210.035433 \nL 135.548033 208.462545 \nL 141.193461 206.673802 \nL 146.838889 204.646279 \nL 152.484317 202.356677 \nL 158.129745 199.782007 \nL 163.775173 196.90048 \nL 169.420601 193.692596 \nL 175.066029 190.142429 \nL 180.711457 186.239052 \nL 186.356885 181.978045 \nL 192.002313 177.362963 \nL 197.647741 172.406633 \nL 203.293169 167.132107 \nL 208.938597 161.573116 \nL 216.465835 153.796426 \nL 225.874881 143.677322 \nL 244.692975 123.279199 \nL 252.220212 115.502509 \nL 257.86564 109.943518 \nL 263.511068 104.668992 \nL 269.156496 99.712662 \nL 274.801924 95.09758 \nL 280.447352 90.836573 \nL 286.09278 86.933196 \nL 291.738208 83.383029 \nL 297.383636 80.175145 \nL 303.029064 77.293618 \nL 308.674492 74.718948 \nL 314.31992 72.429346 \nL 319.965348 70.401823 \nL 325.610776 68.61308 \nL 331.256204 67.040192 \nL 336.901632 65.661117 \nL 344.42887 64.088077 \nL 351.956107 62.777508 \nL 359.483345 61.689139 \nL 368.892391 60.587907 \nL 378.301438 59.720725 \nL 389.592294 58.922245 \nL 402.764959 58.243163 \nL 417.819434 57.705913 \nL 417.819434 57.705913 \n\" style=\"fill:none;stroke:#0000ff;stroke-linecap:square;stroke-width:2;\"/>\n   </g>\n   <g id=\"patch_3\">\n    <path d=\"M 43.339375 253.237813 \nL 43.339375 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_4\">\n    <path d=\"M 417.819434 253.237813 \nL 417.819434 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_5\">\n    <path d=\"M 43.339375 253.237813 \nL 417.819434 253.237813 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_6\">\n    <path d=\"M 43.339375 23.837812 \nL 417.819434 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_7\">\n    <path d=\"M 399.482114 88.367696 \nQ 404.070253 79.314514 408.658392 70.261333 \nL 405.090439 68.453099 \nQ 410.477942 64.458948 415.865445 60.464796 \nQ 415.829872 67.171298 415.794299 73.8778 \nL 412.226346 72.069566 \nQ 407.638207 81.122748 403.050067 90.17593 \nL 399.482114 88.367696 \nz\n\" style=\"stroke:#000000;stroke-linecap:round;\"/>\n   </g>\n   <g id=\"text_14\">\n    <!-- Saturating -->\n    <defs>\n     <path d=\"M 53.515625 70.515625 \nL 53.515625 60.890625 \nQ 47.90625 63.578125 42.921875 64.890625 \nQ 37.9375 66.21875 33.296875 66.21875 \nQ 25.25 66.21875 20.875 63.09375 \nQ 16.5 59.96875 16.5 54.203125 \nQ 16.5 49.359375 19.40625 46.890625 \nQ 22.3125 44.4375 30.421875 42.921875 \nL 36.375 41.703125 \nQ 47.40625 39.59375 52.65625 34.296875 \nQ 57.90625 29 57.90625 20.125 \nQ 57.90625 9.515625 50.796875 4.046875 \nQ 43.703125 -1.421875 29.984375 -1.421875 \nQ 24.8125 -1.421875 18.96875 -0.25 \nQ 13.140625 0.921875 6.890625 3.21875 \nL 6.890625 13.375 \nQ 12.890625 10.015625 18.65625 8.296875 \nQ 24.421875 6.59375 29.984375 6.59375 \nQ 38.421875 6.59375 43.015625 9.90625 \nQ 47.609375 13.234375 47.609375 19.390625 \nQ 47.609375 24.75 44.3125 27.78125 \nQ 41.015625 30.8125 33.5 32.328125 \nL 27.484375 33.5 \nQ 16.453125 35.6875 11.515625 40.375 \nQ 6.59375 45.0625 6.59375 53.421875 \nQ 6.59375 63.09375 13.40625 68.65625 \nQ 20.21875 74.21875 32.171875 74.21875 \nQ 37.3125 74.21875 42.625 73.28125 \nQ 47.953125 72.359375 53.515625 70.515625 \nz\n\" id=\"DejaVuSans-83\"/>\n     <path d=\"M 34.28125 27.484375 \nQ 23.390625 27.484375 19.1875 25 \nQ 14.984375 22.515625 14.984375 16.5 \nQ 14.984375 11.71875 18.140625 8.90625 \nQ 21.296875 6.109375 26.703125 6.109375 \nQ 34.1875 6.109375 38.703125 11.40625 \nQ 43.21875 16.703125 43.21875 25.484375 \nL 43.21875 27.484375 \nz\nM 52.203125 31.203125 \nL 52.203125 0 \nL 43.21875 0 \nL 43.21875 8.296875 \nQ 40.140625 3.328125 35.546875 0.953125 \nQ 30.953125 -1.421875 24.3125 -1.421875 \nQ 15.921875 -1.421875 10.953125 3.296875 \nQ 6 8.015625 6 15.921875 \nQ 6 25.140625 12.171875 29.828125 \nQ 18.359375 34.515625 30.609375 34.515625 \nL 43.21875 34.515625 \nL 43.21875 35.40625 \nQ 43.21875 41.609375 39.140625 45 \nQ 35.0625 48.390625 27.6875 48.390625 \nQ 23 48.390625 18.546875 47.265625 \nQ 14.109375 46.140625 10.015625 43.890625 \nL 10.015625 52.203125 \nQ 14.9375 54.109375 19.578125 55.046875 \nQ 24.21875 56 28.609375 56 \nQ 40.484375 56 46.34375 49.84375 \nQ 52.203125 43.703125 52.203125 31.203125 \nz\n\" id=\"DejaVuSans-97\"/>\n     <path d=\"M 18.3125 70.21875 \nL 18.3125 54.6875 \nL 36.8125 54.6875 \nL 36.8125 47.703125 \nL 18.3125 47.703125 \nL 18.3125 18.015625 \nQ 18.3125 11.328125 20.140625 9.421875 \nQ 21.96875 7.515625 27.59375 7.515625 \nL 36.8125 7.515625 \nL 36.8125 0 \nL 27.59375 0 \nQ 17.1875 0 13.234375 3.875 \nQ 9.28125 7.765625 9.28125 18.015625 \nL 9.28125 47.703125 \nL 2.6875 47.703125 \nL 2.6875 54.6875 \nL 9.28125 54.6875 \nL 9.28125 70.21875 \nz\n\" id=\"DejaVuSans-116\"/>\n     <path d=\"M 8.5 21.578125 \nL 8.5 54.6875 \nL 17.484375 54.6875 \nL 17.484375 21.921875 \nQ 17.484375 14.15625 20.5 10.265625 \nQ 23.53125 6.390625 29.59375 6.390625 \nQ 36.859375 6.390625 41.078125 11.03125 \nQ 45.3125 15.671875 45.3125 23.6875 \nL 45.3125 54.6875 \nL 54.296875 54.6875 \nL 54.296875 0 \nL 45.3125 0 \nL 45.3125 8.40625 \nQ 42.046875 3.421875 37.71875 1 \nQ 33.40625 -1.421875 27.6875 -1.421875 \nQ 18.265625 -1.421875 13.375 4.4375 \nQ 8.5 10.296875 8.5 21.578125 \nz\nM 31.109375 56 \nz\n\" id=\"DejaVuSans-117\"/>\n     <path d=\"M 41.109375 46.296875 \nQ 39.59375 47.171875 37.8125 47.578125 \nQ 36.03125 48 33.890625 48 \nQ 26.265625 48 22.1875 43.046875 \nQ 18.109375 38.09375 18.109375 28.8125 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 54.6875 \nL 18.109375 54.6875 \nL 18.109375 46.1875 \nQ 20.953125 51.171875 25.484375 53.578125 \nQ 30.03125 56 36.53125 56 \nQ 37.453125 56 38.578125 55.875 \nQ 39.703125 55.765625 41.0625 55.515625 \nz\n\" id=\"DejaVuSans-114\"/>\n     <path d=\"M 9.421875 54.6875 \nL 18.40625 54.6875 \nL 18.40625 0 \nL 9.421875 0 \nz\nM 9.421875 75.984375 \nL 18.40625 75.984375 \nL 18.40625 64.59375 \nL 9.421875 64.59375 \nz\n\" id=\"DejaVuSans-105\"/>\n     <path d=\"M 54.890625 33.015625 \nL 54.890625 0 \nL 45.90625 0 \nL 45.90625 32.71875 \nQ 45.90625 40.484375 42.875 44.328125 \nQ 39.84375 48.1875 33.796875 48.1875 \nQ 26.515625 48.1875 22.3125 43.546875 \nQ 18.109375 38.921875 18.109375 30.90625 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 54.6875 \nL 18.109375 54.6875 \nL 18.109375 46.1875 \nQ 21.34375 51.125 25.703125 53.5625 \nQ 30.078125 56 35.796875 56 \nQ 45.21875 56 50.046875 50.171875 \nQ 54.890625 44.34375 54.890625 33.015625 \nz\n\" id=\"DejaVuSans-110\"/>\n     <path d=\"M 45.40625 27.984375 \nQ 45.40625 37.75 41.375 43.109375 \nQ 37.359375 48.484375 30.078125 48.484375 \nQ 22.859375 48.484375 18.828125 43.109375 \nQ 14.796875 37.75 14.796875 27.984375 \nQ 14.796875 18.265625 18.828125 12.890625 \nQ 22.859375 7.515625 30.078125 7.515625 \nQ 37.359375 7.515625 41.375 12.890625 \nQ 45.40625 18.265625 45.40625 27.984375 \nz\nM 54.390625 6.78125 \nQ 54.390625 -7.171875 48.1875 -13.984375 \nQ 42 -20.796875 29.203125 -20.796875 \nQ 24.46875 -20.796875 20.265625 -20.09375 \nQ 16.0625 -19.390625 12.109375 -17.921875 \nL 12.109375 -9.1875 \nQ 16.0625 -11.328125 19.921875 -12.34375 \nQ 23.78125 -13.375 27.78125 -13.375 \nQ 36.625 -13.375 41.015625 -8.765625 \nQ 45.40625 -4.15625 45.40625 5.171875 \nL 45.40625 9.625 \nQ 42.625 4.78125 38.28125 2.390625 \nQ 33.9375 0 27.875 0 \nQ 17.828125 0 11.671875 7.65625 \nQ 5.515625 15.328125 5.515625 27.984375 \nQ 5.515625 40.671875 11.671875 48.328125 \nQ 17.828125 56 27.875 56 \nQ 33.9375 56 38.28125 53.609375 \nQ 42.625 51.21875 45.40625 46.390625 \nL 45.40625 54.6875 \nL 54.390625 54.6875 \nz\n\" id=\"DejaVuSans-103\"/>\n    </defs>\n    <g transform=\"translate(324.996957 105.766384)scale(0.14 -0.14)\">\n     <use xlink:href=\"#DejaVuSans-83\"/>\n     <use x=\"63.476562\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"124.755859\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"163.964844\" xlink:href=\"#DejaVuSans-117\"/>\n     <use x=\"227.34375\" xlink:href=\"#DejaVuSans-114\"/>\n     <use x=\"268.457031\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"329.736328\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"368.945312\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"396.728516\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"460.107422\" xlink:href=\"#DejaVuSans-103\"/>\n    </g>\n   </g>\n   <g id=\"patch_8\">\n    <path d=\"M 61.905987 181.625717 \nQ 56.854482 193.592486 51.802978 205.559255 \nL 55.488101 207.114848 \nQ 50.39009 211.477685 45.292079 215.840522 \nQ 44.862405 209.144296 44.432732 202.44807 \nL 48.117855 204.003663 \nQ 53.169359 192.036894 58.220864 180.070125 \nL 61.905987 181.625717 \nz\n\" style=\"stroke:#000000;stroke-linecap:round;\"/>\n   </g>\n   <g id=\"text_15\">\n    <!-- Saturating -->\n    <g transform=\"translate(62.860915 171.309241)scale(0.14 -0.14)\">\n     <use xlink:href=\"#DejaVuSans-83\"/>\n     <use x=\"63.476562\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"124.755859\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"163.964844\" xlink:href=\"#DejaVuSans-117\"/>\n     <use x=\"227.34375\" xlink:href=\"#DejaVuSans-114\"/>\n     <use x=\"268.457031\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"329.736328\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"368.945312\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"396.728516\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"460.107422\" xlink:href=\"#DejaVuSans-103\"/>\n    </g>\n   </g>\n   <g id=\"patch_9\">\n    <path d=\"M 275.229117 173.372851 \nQ 259.832665 162.212549 244.436213 151.052247 \nL 242.088635 154.290901 \nQ 238.992421 148.34133 235.896206 142.391759 \nQ 242.513787 143.483349 249.131367 144.574938 \nL 246.78379 147.813593 \nQ 262.180242 158.973895 277.576694 170.134197 \nL 275.229117 173.372851 \nz\n\" style=\"stroke:#000000;stroke-linecap:round;\"/>\n   </g>\n   <g id=\"text_16\">\n    <!-- Linear -->\n    <defs>\n     <path d=\"M 9.8125 72.90625 \nL 19.671875 72.90625 \nL 19.671875 8.296875 \nL 55.171875 8.296875 \nL 55.171875 0 \nL 9.8125 0 \nz\n\" id=\"DejaVuSans-76\"/>\n     <path d=\"M 56.203125 29.59375 \nL 56.203125 25.203125 \nL 14.890625 25.203125 \nQ 15.484375 15.921875 20.484375 11.0625 \nQ 25.484375 6.203125 34.421875 6.203125 \nQ 39.59375 6.203125 44.453125 7.46875 \nQ 49.3125 8.734375 54.109375 11.28125 \nL 54.109375 2.78125 \nQ 49.265625 0.734375 44.1875 -0.34375 \nQ 39.109375 -1.421875 33.890625 -1.421875 \nQ 20.796875 -1.421875 13.15625 6.1875 \nQ 5.515625 13.8125 5.515625 26.8125 \nQ 5.515625 40.234375 12.765625 48.109375 \nQ 20.015625 56 32.328125 56 \nQ 43.359375 56 49.78125 48.890625 \nQ 56.203125 41.796875 56.203125 29.59375 \nz\nM 47.21875 32.234375 \nQ 47.125 39.59375 43.09375 43.984375 \nQ 39.0625 48.390625 32.421875 48.390625 \nQ 24.90625 48.390625 20.390625 44.140625 \nQ 15.875 39.890625 15.1875 32.171875 \nz\n\" id=\"DejaVuSans-101\"/>\n    </defs>\n    <g transform=\"translate(283.719635 187.694955)scale(0.14 -0.14)\">\n     <use xlink:href=\"#DejaVuSans-76\"/>\n     <use x=\"55.712891\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"83.496094\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"146.875\" xlink:href=\"#DejaVuSans-101\"/>\n     <use x=\"208.398438\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"269.677734\" xlink:href=\"#DejaVuSans-114\"/>\n    </g>\n   </g>\n   <g id=\"text_17\">\n    <!-- Sigmoid activation function -->\n    <defs>\n     <path d=\"M 52 44.1875 \nQ 55.375 50.25 60.0625 53.125 \nQ 64.75 56 71.09375 56 \nQ 79.640625 56 84.28125 50.015625 \nQ 88.921875 44.046875 88.921875 33.015625 \nL 88.921875 0 \nL 79.890625 0 \nL 79.890625 32.71875 \nQ 79.890625 40.578125 77.09375 44.375 \nQ 74.3125 48.1875 68.609375 48.1875 \nQ 61.625 48.1875 57.5625 43.546875 \nQ 53.515625 38.921875 53.515625 30.90625 \nL 53.515625 0 \nL 44.484375 0 \nL 44.484375 32.71875 \nQ 44.484375 40.625 41.703125 44.40625 \nQ 38.921875 48.1875 33.109375 48.1875 \nQ 26.21875 48.1875 22.15625 43.53125 \nQ 18.109375 38.875 18.109375 30.90625 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 54.6875 \nL 18.109375 54.6875 \nL 18.109375 46.1875 \nQ 21.1875 51.21875 25.484375 53.609375 \nQ 29.78125 56 35.6875 56 \nQ 41.65625 56 45.828125 52.96875 \nQ 50 49.953125 52 44.1875 \nz\n\" id=\"DejaVuSans-109\"/>\n     <path d=\"M 30.609375 48.390625 \nQ 23.390625 48.390625 19.1875 42.75 \nQ 14.984375 37.109375 14.984375 27.296875 \nQ 14.984375 17.484375 19.15625 11.84375 \nQ 23.34375 6.203125 30.609375 6.203125 \nQ 37.796875 6.203125 41.984375 11.859375 \nQ 46.1875 17.53125 46.1875 27.296875 \nQ 46.1875 37.015625 41.984375 42.703125 \nQ 37.796875 48.390625 30.609375 48.390625 \nz\nM 30.609375 56 \nQ 42.328125 56 49.015625 48.375 \nQ 55.71875 40.765625 55.71875 27.296875 \nQ 55.71875 13.875 49.015625 6.21875 \nQ 42.328125 -1.421875 30.609375 -1.421875 \nQ 18.84375 -1.421875 12.171875 6.21875 \nQ 5.515625 13.875 5.515625 27.296875 \nQ 5.515625 40.765625 12.171875 48.375 \nQ 18.84375 56 30.609375 56 \nz\n\" id=\"DejaVuSans-111\"/>\n     <path d=\"M 45.40625 46.390625 \nL 45.40625 75.984375 \nL 54.390625 75.984375 \nL 54.390625 0 \nL 45.40625 0 \nL 45.40625 8.203125 \nQ 42.578125 3.328125 38.25 0.953125 \nQ 33.9375 -1.421875 27.875 -1.421875 \nQ 17.96875 -1.421875 11.734375 6.484375 \nQ 5.515625 14.40625 5.515625 27.296875 \nQ 5.515625 40.1875 11.734375 48.09375 \nQ 17.96875 56 27.875 56 \nQ 33.9375 56 38.25 53.625 \nQ 42.578125 51.265625 45.40625 46.390625 \nz\nM 14.796875 27.296875 \nQ 14.796875 17.390625 18.875 11.75 \nQ 22.953125 6.109375 30.078125 6.109375 \nQ 37.203125 6.109375 41.296875 11.75 \nQ 45.40625 17.390625 45.40625 27.296875 \nQ 45.40625 37.203125 41.296875 42.84375 \nQ 37.203125 48.484375 30.078125 48.484375 \nQ 22.953125 48.484375 18.875 42.84375 \nQ 14.796875 37.203125 14.796875 27.296875 \nz\n\" id=\"DejaVuSans-100\"/>\n     <path id=\"DejaVuSans-32\"/>\n     <path d=\"M 48.78125 52.59375 \nL 48.78125 44.1875 \nQ 44.96875 46.296875 41.140625 47.34375 \nQ 37.3125 48.390625 33.40625 48.390625 \nQ 24.65625 48.390625 19.8125 42.84375 \nQ 14.984375 37.3125 14.984375 27.296875 \nQ 14.984375 17.28125 19.8125 11.734375 \nQ 24.65625 6.203125 33.40625 6.203125 \nQ 37.3125 6.203125 41.140625 7.25 \nQ 44.96875 8.296875 48.78125 10.40625 \nL 48.78125 2.09375 \nQ 45.015625 0.34375 40.984375 -0.53125 \nQ 36.96875 -1.421875 32.421875 -1.421875 \nQ 20.0625 -1.421875 12.78125 6.34375 \nQ 5.515625 14.109375 5.515625 27.296875 \nQ 5.515625 40.671875 12.859375 48.328125 \nQ 20.21875 56 33.015625 56 \nQ 37.15625 56 41.109375 55.140625 \nQ 45.0625 54.296875 48.78125 52.59375 \nz\n\" id=\"DejaVuSans-99\"/>\n     <path d=\"M 2.984375 54.6875 \nL 12.5 54.6875 \nL 29.59375 8.796875 \nL 46.6875 54.6875 \nL 56.203125 54.6875 \nL 35.6875 0 \nL 23.484375 0 \nz\n\" id=\"DejaVuSans-118\"/>\n     <path d=\"M 37.109375 75.984375 \nL 37.109375 68.5 \nL 28.515625 68.5 \nQ 23.6875 68.5 21.796875 66.546875 \nQ 19.921875 64.59375 19.921875 59.515625 \nL 19.921875 54.6875 \nL 34.71875 54.6875 \nL 34.71875 47.703125 \nL 19.921875 47.703125 \nL 19.921875 0 \nL 10.890625 0 \nL 10.890625 47.703125 \nL 2.296875 47.703125 \nL 2.296875 54.6875 \nL 10.890625 54.6875 \nL 10.890625 58.5 \nQ 10.890625 67.625 15.140625 71.796875 \nQ 19.390625 75.984375 28.609375 75.984375 \nz\n\" id=\"DejaVuSans-102\"/>\n    </defs>\n    <g transform=\"translate(134.544873 17.837812)scale(0.14 -0.14)\">\n     <use xlink:href=\"#DejaVuSans-83\"/>\n     <use x=\"63.476562\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"91.259766\" xlink:href=\"#DejaVuSans-103\"/>\n     <use x=\"154.736328\" xlink:href=\"#DejaVuSans-109\"/>\n     <use x=\"252.148438\" xlink:href=\"#DejaVuSans-111\"/>\n     <use x=\"313.330078\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"341.113281\" xlink:href=\"#DejaVuSans-100\"/>\n     <use x=\"404.589844\" xlink:href=\"#DejaVuSans-32\"/>\n     <use x=\"436.376953\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"497.65625\" xlink:href=\"#DejaVuSans-99\"/>\n     <use x=\"552.636719\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"591.845703\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"619.628906\" xlink:href=\"#DejaVuSans-118\"/>\n     <use x=\"678.808594\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"740.087891\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"779.296875\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"807.080078\" xlink:href=\"#DejaVuSans-111\"/>\n     <use x=\"868.261719\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"931.640625\" xlink:href=\"#DejaVuSans-32\"/>\n     <use x=\"963.427734\" xlink:href=\"#DejaVuSans-102\"/>\n     <use x=\"998.632812\" xlink:href=\"#DejaVuSans-117\"/>\n     <use x=\"1062.011719\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"1125.390625\" xlink:href=\"#DejaVuSans-99\"/>\n     <use x=\"1180.371094\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"1219.580078\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"1247.363281\" xlink:href=\"#DejaVuSans-111\"/>\n     <use x=\"1308.544922\" xlink:href=\"#DejaVuSans-110\"/>\n    </g>\n   </g>\n  </g>\n </g>\n <defs>\n  <clipPath id=\"pe870947a28\">\n   <rect height=\"229.4\" width=\"374.480059\" x=\"43.339375\" y=\"23.837812\"/>\n  </clipPath>\n </defs>\n</svg>\n",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "source": [
    "z = np.linspace(-5, 5, 200)\n",
    "\n",
    "plt.plot([-5, 5], [0, 0], 'k-')\n",
    "plt.plot([-5, 5], [1, 1], 'k--')\n",
    "plt.plot([0, 0], [-0.2, 1.2], 'k-')\n",
    "plt.plot([-5, 5], [-3/4, 7/4], 'g--')\n",
    "plt.plot(z, logit(z), \"b-\", linewidth=2)\n",
    "props = dict(facecolor='black', shrink=0.1)\n",
    "plt.annotate('Saturating', xytext=(3.5, 0.7), xy=(5, 1), arrowprops=props, fontsize=14, ha=\"center\")\n",
    "plt.annotate('Saturating', xytext=(-3.5, 0.3), xy=(-5, 0), arrowprops=props, fontsize=14, ha=\"center\")\n",
    "plt.annotate('Linear', xytext=(2, 0.2), xy=(0, 0.5), arrowprops=props, fontsize=14, ha=\"center\")\n",
    "plt.grid(True)\n",
    "plt.title(\"Sigmoid activation function\", fontsize=14)\n",
    "plt.axis([-5, 5, -0.2, 1.2])\n",
    "\n",
    "save_fig(\"sigmoid_saturation_plot\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Xavier and He Initialization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "['Constant',\n 'GlorotNormal',\n 'GlorotUniform',\n 'Identity',\n 'Initializer',\n 'Ones',\n 'Orthogonal',\n 'RandomNormal',\n 'RandomUniform',\n 'TruncatedNormal',\n 'VarianceScaling',\n 'Zeros',\n 'constant',\n 'deserialize',\n 'get',\n 'glorot_normal',\n 'glorot_uniform',\n 'he_normal',\n 'he_uniform',\n 'identity',\n 'lecun_normal',\n 'lecun_uniform',\n 'ones',\n 'orthogonal',\n 'serialize',\n 'zeros']"
     },
     "metadata": {},
     "execution_count": 41
    }
   ],
   "source": [
    "[name for name in dir(keras.initializers) if not name.startswith(\"_\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "<tensorflow.python.keras.layers.core.Dense at 0x7f8cf222b950>"
     },
     "metadata": {},
     "execution_count": 43
    }
   ],
   "source": [
    "keras.layers.Dense(10, activation=\"relu\", kernel_initializer=\"he_normal\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "<tensorflow.python.keras.layers.core.Dense at 0x7f8cf45aacd0>"
     },
     "metadata": {},
     "execution_count": 45
    }
   ],
   "source": [
    "\n",
    "init = keras.initializers.VarianceScaling(scale=2., mode='fan_avg',\n",
    "                                          distribution='uniform')\n",
    "keras.layers.Dense(10, activation=\"relu\", kernel_initializer=init)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## nonsaturating activation functions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## leaky ReLU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "def leaky_relu(z, alpha=0.01):\n",
    "    return np.maximum(alpha*z, z)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Saving figure leaky_relu_plot\n"
    },
    {
     "output_type": "display_data",
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/svg+xml": "<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n  \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n<!-- Created with matplotlib (https://matplotlib.org/) -->\n<svg height=\"279.051563pt\" version=\"1.1\" viewBox=\"0 0 424.81 279.051563\" width=\"424.81pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n <defs>\n  <style type=\"text/css\">\n*{stroke-linecap:butt;stroke-linejoin:round;}\n  </style>\n </defs>\n <g id=\"figure_1\">\n  <g id=\"patch_1\">\n   <path d=\"M 0 279.051563 \nL 424.81 279.051563 \nL 424.81 0 \nL 0 0 \nz\n\" style=\"fill:none;\"/>\n  </g>\n  <g id=\"axes_1\">\n   <g id=\"patch_2\">\n    <path d=\"M 21.835 253.237813 \nL 417.61 253.237813 \nL 417.61 23.837812 \nL 21.835 23.837812 \nz\n\" style=\"fill:#ffffff;\"/>\n   </g>\n   <g id=\"matplotlib.axis_1\">\n    <g id=\"xtick_1\">\n     <g id=\"line2d_1\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 61.4125 253.237813 \nL 61.4125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_2\">\n      <defs>\n       <path d=\"M 0 0 \nL 0 3.5 \n\" id=\"m8c23400201\" style=\"stroke:#000000;stroke-width:0.8;\"/>\n      </defs>\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"61.4125\" xlink:href=\"#m8c23400201\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_1\">\n      <!-- −4 -->\n      <defs>\n       <path d=\"M 10.59375 35.5 \nL 73.1875 35.5 \nL 73.1875 27.203125 \nL 10.59375 27.203125 \nz\n\" id=\"DejaVuSans-8722\"/>\n       <path d=\"M 37.796875 64.3125 \nL 12.890625 25.390625 \nL 37.796875 25.390625 \nz\nM 35.203125 72.90625 \nL 47.609375 72.90625 \nL 47.609375 25.390625 \nL 58.015625 25.390625 \nL 58.015625 17.1875 \nL 47.609375 17.1875 \nL 47.609375 0 \nL 37.796875 0 \nL 37.796875 17.1875 \nL 4.890625 17.1875 \nL 4.890625 26.703125 \nz\n\" id=\"DejaVuSans-52\"/>\n      </defs>\n      <g transform=\"translate(52.567187 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_2\">\n     <g id=\"line2d_3\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 140.5675 253.237813 \nL 140.5675 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_4\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"140.5675\" xlink:href=\"#m8c23400201\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_2\">\n      <!-- −2 -->\n      <defs>\n       <path d=\"M 19.1875 8.296875 \nL 53.609375 8.296875 \nL 53.609375 0 \nL 7.328125 0 \nL 7.328125 8.296875 \nQ 12.9375 14.109375 22.625 23.890625 \nQ 32.328125 33.6875 34.8125 36.53125 \nQ 39.546875 41.84375 41.421875 45.53125 \nQ 43.3125 49.21875 43.3125 52.78125 \nQ 43.3125 58.59375 39.234375 62.25 \nQ 35.15625 65.921875 28.609375 65.921875 \nQ 23.96875 65.921875 18.8125 64.3125 \nQ 13.671875 62.703125 7.8125 59.421875 \nL 7.8125 69.390625 \nQ 13.765625 71.78125 18.9375 73 \nQ 24.125 74.21875 28.421875 74.21875 \nQ 39.75 74.21875 46.484375 68.546875 \nQ 53.21875 62.890625 53.21875 53.421875 \nQ 53.21875 48.921875 51.53125 44.890625 \nQ 49.859375 40.875 45.40625 35.40625 \nQ 44.1875 33.984375 37.640625 27.21875 \nQ 31.109375 20.453125 19.1875 8.296875 \nz\n\" id=\"DejaVuSans-50\"/>\n      </defs>\n      <g transform=\"translate(131.722187 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_3\">\n     <g id=\"line2d_5\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 219.7225 253.237813 \nL 219.7225 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_6\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"219.7225\" xlink:href=\"#m8c23400201\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_3\">\n      <!-- 0 -->\n      <defs>\n       <path d=\"M 31.78125 66.40625 \nQ 24.171875 66.40625 20.328125 58.90625 \nQ 16.5 51.421875 16.5 36.375 \nQ 16.5 21.390625 20.328125 13.890625 \nQ 24.171875 6.390625 31.78125 6.390625 \nQ 39.453125 6.390625 43.28125 13.890625 \nQ 47.125 21.390625 47.125 36.375 \nQ 47.125 51.421875 43.28125 58.90625 \nQ 39.453125 66.40625 31.78125 66.40625 \nz\nM 31.78125 74.21875 \nQ 44.046875 74.21875 50.515625 64.515625 \nQ 56.984375 54.828125 56.984375 36.375 \nQ 56.984375 17.96875 50.515625 8.265625 \nQ 44.046875 -1.421875 31.78125 -1.421875 \nQ 19.53125 -1.421875 13.0625 8.265625 \nQ 6.59375 17.96875 6.59375 36.375 \nQ 6.59375 54.828125 13.0625 64.515625 \nQ 19.53125 74.21875 31.78125 74.21875 \nz\n\" id=\"DejaVuSans-48\"/>\n      </defs>\n      <g transform=\"translate(215.905 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_4\">\n     <g id=\"line2d_7\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 298.8775 253.237813 \nL 298.8775 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_8\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"298.8775\" xlink:href=\"#m8c23400201\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_4\">\n      <!-- 2 -->\n      <g transform=\"translate(295.06 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_5\">\n     <g id=\"line2d_9\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 378.0325 253.237813 \nL 378.0325 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_10\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"378.0325\" xlink:href=\"#m8c23400201\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_5\">\n      <!-- 4 -->\n      <g transform=\"translate(374.215 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n   </g>\n   <g id=\"matplotlib.axis_2\">\n    <g id=\"ytick_1\">\n     <g id=\"line2d_11\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 21.835 228.833557 \nL 417.61 228.833557 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_12\">\n      <defs>\n       <path d=\"M 0 0 \nL -3.5 0 \n\" id=\"mb0381fd7b2\" style=\"stroke:#000000;stroke-width:0.8;\"/>\n      </defs>\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"21.835\" xlink:href=\"#mb0381fd7b2\" y=\"228.833557\"/>\n      </g>\n     </g>\n     <g id=\"text_6\">\n      <!-- 0 -->\n      <g transform=\"translate(7.2 233.39262)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_2\">\n     <g id=\"line2d_13\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 21.835 180.025047 \nL 417.61 180.025047 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_14\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"21.835\" xlink:href=\"#mb0381fd7b2\" y=\"180.025047\"/>\n      </g>\n     </g>\n     <g id=\"text_7\">\n      <!-- 1 -->\n      <defs>\n       <path d=\"M 12.40625 8.296875 \nL 28.515625 8.296875 \nL 28.515625 63.921875 \nL 10.984375 60.40625 \nL 10.984375 69.390625 \nL 28.421875 72.90625 \nL 38.28125 72.90625 \nL 38.28125 8.296875 \nL 54.390625 8.296875 \nL 54.390625 0 \nL 12.40625 0 \nz\n\" id=\"DejaVuSans-49\"/>\n      </defs>\n      <g transform=\"translate(7.2 184.584109)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-49\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_3\">\n     <g id=\"line2d_15\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 21.835 131.216536 \nL 417.61 131.216536 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_16\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"21.835\" xlink:href=\"#mb0381fd7b2\" y=\"131.216536\"/>\n      </g>\n     </g>\n     <g id=\"text_8\">\n      <!-- 2 -->\n      <g transform=\"translate(7.2 135.775598)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_4\">\n     <g id=\"line2d_17\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 21.835 82.408025 \nL 417.61 82.408025 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_18\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"21.835\" xlink:href=\"#mb0381fd7b2\" y=\"82.408025\"/>\n      </g>\n     </g>\n     <g id=\"text_9\">\n      <!-- 3 -->\n      <defs>\n       <path d=\"M 40.578125 39.3125 \nQ 47.65625 37.796875 51.625 33 \nQ 55.609375 28.21875 55.609375 21.1875 \nQ 55.609375 10.40625 48.1875 4.484375 \nQ 40.765625 -1.421875 27.09375 -1.421875 \nQ 22.515625 -1.421875 17.65625 -0.515625 \nQ 12.796875 0.390625 7.625 2.203125 \nL 7.625 11.71875 \nQ 11.71875 9.328125 16.59375 8.109375 \nQ 21.484375 6.890625 26.8125 6.890625 \nQ 36.078125 6.890625 40.9375 10.546875 \nQ 45.796875 14.203125 45.796875 21.1875 \nQ 45.796875 27.640625 41.28125 31.265625 \nQ 36.765625 34.90625 28.71875 34.90625 \nL 20.21875 34.90625 \nL 20.21875 43.015625 \nL 29.109375 43.015625 \nQ 36.375 43.015625 40.234375 45.921875 \nQ 44.09375 48.828125 44.09375 54.296875 \nQ 44.09375 59.90625 40.109375 62.90625 \nQ 36.140625 65.921875 28.71875 65.921875 \nQ 24.65625 65.921875 20.015625 65.03125 \nQ 15.375 64.15625 9.8125 62.3125 \nL 9.8125 71.09375 \nQ 15.4375 72.65625 20.34375 73.4375 \nQ 25.25 74.21875 29.59375 74.21875 \nQ 40.828125 74.21875 47.359375 69.109375 \nQ 53.90625 64.015625 53.90625 55.328125 \nQ 53.90625 49.265625 50.4375 45.09375 \nQ 46.96875 40.921875 40.578125 39.3125 \nz\n\" id=\"DejaVuSans-51\"/>\n      </defs>\n      <g transform=\"translate(7.2 86.967088)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-51\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_5\">\n     <g id=\"line2d_19\">\n      <path clip-path=\"url(#p499d76ad51)\" d=\"M 21.835 33.599515 \nL 417.61 33.599515 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_20\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"21.835\" xlink:href=\"#mb0381fd7b2\" y=\"33.599515\"/>\n      </g>\n     </g>\n     <g id=\"text_10\">\n      <!-- 4 -->\n      <g transform=\"translate(7.2 38.158577)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n   </g>\n   <g id=\"line2d_21\">\n    <path clip-path=\"url(#p499d76ad51)\" d=\"M 21.835 241.035685 \nL 218.72809 228.894874 \nL 220.71691 227.607213 \nL 406.08831 -1 \nL 406.08831 -1 \n\" style=\"fill:none;stroke:#0000ff;stroke-linecap:square;stroke-width:2;\"/>\n   </g>\n   <g id=\"line2d_22\">\n    <path clip-path=\"url(#p499d76ad51)\" d=\"M 21.835 228.833557 \nL 417.61 228.833557 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"line2d_23\">\n    <path clip-path=\"url(#p499d76ad51)\" d=\"M 219.7225 253.237813 \nL 219.7225 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"patch_3\">\n    <path d=\"M 21.835 253.237813 \nL 21.835 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_4\">\n    <path d=\"M 417.61 253.237813 \nL 417.61 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_5\">\n    <path d=\"M 21.835 253.237813 \nL 417.61 253.237813 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_6\">\n    <path d=\"M 21.835 23.837812 \nL 417.61 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_7\">\n    <path d=\"M 59.549152 213.541679 \nQ 48.267889 221.776524 36.986627 230.011369 \nL 39.344983 233.242182 \nQ 32.731828 234.355271 26.118672 235.46836 \nQ 29.194292 229.509052 32.269913 223.549743 \nL 34.62827 226.780556 \nQ 45.909532 218.545711 57.190795 210.310866 \nL 59.549152 213.541679 \nz\n\" style=\"stroke:#000000;stroke-linecap:round;\"/>\n   </g>\n   <g id=\"text_11\">\n    <!-- Leak -->\n    <defs>\n     <path d=\"M 9.8125 72.90625 \nL 19.671875 72.90625 \nL 19.671875 8.296875 \nL 55.171875 8.296875 \nL 55.171875 0 \nL 9.8125 0 \nz\n\" id=\"DejaVuSans-76\"/>\n     <path d=\"M 56.203125 29.59375 \nL 56.203125 25.203125 \nL 14.890625 25.203125 \nQ 15.484375 15.921875 20.484375 11.0625 \nQ 25.484375 6.203125 34.421875 6.203125 \nQ 39.59375 6.203125 44.453125 7.46875 \nQ 49.3125 8.734375 54.109375 11.28125 \nL 54.109375 2.78125 \nQ 49.265625 0.734375 44.1875 -0.34375 \nQ 39.109375 -1.421875 33.890625 -1.421875 \nQ 20.796875 -1.421875 13.15625 6.1875 \nQ 5.515625 13.8125 5.515625 26.8125 \nQ 5.515625 40.234375 12.765625 48.109375 \nQ 20.015625 56 32.328125 56 \nQ 43.359375 56 49.78125 48.890625 \nQ 56.203125 41.796875 56.203125 29.59375 \nz\nM 47.21875 32.234375 \nQ 47.125 39.59375 43.09375 43.984375 \nQ 39.0625 48.390625 32.421875 48.390625 \nQ 24.90625 48.390625 20.390625 44.140625 \nQ 15.875 39.890625 15.1875 32.171875 \nz\n\" id=\"DejaVuSans-101\"/>\n     <path d=\"M 34.28125 27.484375 \nQ 23.390625 27.484375 19.1875 25 \nQ 14.984375 22.515625 14.984375 16.5 \nQ 14.984375 11.71875 18.140625 8.90625 \nQ 21.296875 6.109375 26.703125 6.109375 \nQ 34.1875 6.109375 38.703125 11.40625 \nQ 43.21875 16.703125 43.21875 25.484375 \nL 43.21875 27.484375 \nz\nM 52.203125 31.203125 \nL 52.203125 0 \nL 43.21875 0 \nL 43.21875 8.296875 \nQ 40.140625 3.328125 35.546875 0.953125 \nQ 30.953125 -1.421875 24.3125 -1.421875 \nQ 15.921875 -1.421875 10.953125 3.296875 \nQ 6 8.015625 6 15.921875 \nQ 6 25.140625 12.171875 29.828125 \nQ 18.359375 34.515625 30.609375 34.515625 \nL 43.21875 34.515625 \nL 43.21875 35.40625 \nQ 43.21875 41.609375 39.140625 45 \nQ 35.0625 48.390625 27.6875 48.390625 \nQ 23 48.390625 18.546875 47.265625 \nQ 14.109375 46.140625 10.015625 43.890625 \nL 10.015625 52.203125 \nQ 14.9375 54.109375 19.578125 55.046875 \nQ 24.21875 56 28.609375 56 \nQ 40.484375 56 46.34375 49.84375 \nQ 52.203125 43.703125 52.203125 31.203125 \nz\n\" id=\"DejaVuSans-97\"/>\n     <path d=\"M 9.078125 75.984375 \nL 18.109375 75.984375 \nL 18.109375 31.109375 \nL 44.921875 54.6875 \nL 56.390625 54.6875 \nL 27.390625 29.109375 \nL 57.625 0 \nL 45.90625 0 \nL 18.109375 26.703125 \nL 18.109375 0 \nL 9.078125 0 \nz\n\" id=\"DejaVuSans-107\"/>\n    </defs>\n    <g transform=\"translate(64.651719 204.429302)scale(0.14 -0.14)\">\n     <use xlink:href=\"#DejaVuSans-76\"/>\n     <use x=\"55.697266\" xlink:href=\"#DejaVuSans-101\"/>\n     <use x=\"117.220703\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"178.5\" xlink:href=\"#DejaVuSans-107\"/>\n    </g>\n   </g>\n   <g id=\"text_12\">\n    <!-- Leaky ReLU activation function -->\n    <defs>\n     <path d=\"M 32.171875 -5.078125 \nQ 28.375 -14.84375 24.75 -17.8125 \nQ 21.140625 -20.796875 15.09375 -20.796875 \nL 7.90625 -20.796875 \nL 7.90625 -13.28125 \nL 13.1875 -13.28125 \nQ 16.890625 -13.28125 18.9375 -11.515625 \nQ 21 -9.765625 23.484375 -3.21875 \nL 25.09375 0.875 \nL 2.984375 54.6875 \nL 12.5 54.6875 \nL 29.59375 11.921875 \nL 46.6875 54.6875 \nL 56.203125 54.6875 \nz\n\" id=\"DejaVuSans-121\"/>\n     <path id=\"DejaVuSans-32\"/>\n     <path d=\"M 44.390625 34.1875 \nQ 47.5625 33.109375 50.5625 29.59375 \nQ 53.5625 26.078125 56.59375 19.921875 \nL 66.609375 0 \nL 56 0 \nL 46.6875 18.703125 \nQ 43.0625 26.03125 39.671875 28.421875 \nQ 36.28125 30.8125 30.421875 30.8125 \nL 19.671875 30.8125 \nL 19.671875 0 \nL 9.8125 0 \nL 9.8125 72.90625 \nL 32.078125 72.90625 \nQ 44.578125 72.90625 50.734375 67.671875 \nQ 56.890625 62.453125 56.890625 51.90625 \nQ 56.890625 45.015625 53.6875 40.46875 \nQ 50.484375 35.9375 44.390625 34.1875 \nz\nM 19.671875 64.796875 \nL 19.671875 38.921875 \nL 32.078125 38.921875 \nQ 39.203125 38.921875 42.84375 42.21875 \nQ 46.484375 45.515625 46.484375 51.90625 \nQ 46.484375 58.296875 42.84375 61.546875 \nQ 39.203125 64.796875 32.078125 64.796875 \nz\n\" id=\"DejaVuSans-82\"/>\n     <path d=\"M 8.6875 72.90625 \nL 18.609375 72.90625 \nL 18.609375 28.609375 \nQ 18.609375 16.890625 22.84375 11.734375 \nQ 27.09375 6.59375 36.625 6.59375 \nQ 46.09375 6.59375 50.34375 11.734375 \nQ 54.59375 16.890625 54.59375 28.609375 \nL 54.59375 72.90625 \nL 64.5 72.90625 \nL 64.5 27.390625 \nQ 64.5 13.140625 57.4375 5.859375 \nQ 50.390625 -1.421875 36.625 -1.421875 \nQ 22.796875 -1.421875 15.734375 5.859375 \nQ 8.6875 13.140625 8.6875 27.390625 \nz\n\" id=\"DejaVuSans-85\"/>\n     <path d=\"M 48.78125 52.59375 \nL 48.78125 44.1875 \nQ 44.96875 46.296875 41.140625 47.34375 \nQ 37.3125 48.390625 33.40625 48.390625 \nQ 24.65625 48.390625 19.8125 42.84375 \nQ 14.984375 37.3125 14.984375 27.296875 \nQ 14.984375 17.28125 19.8125 11.734375 \nQ 24.65625 6.203125 33.40625 6.203125 \nQ 37.3125 6.203125 41.140625 7.25 \nQ 44.96875 8.296875 48.78125 10.40625 \nL 48.78125 2.09375 \nQ 45.015625 0.34375 40.984375 -0.53125 \nQ 36.96875 -1.421875 32.421875 -1.421875 \nQ 20.0625 -1.421875 12.78125 6.34375 \nQ 5.515625 14.109375 5.515625 27.296875 \nQ 5.515625 40.671875 12.859375 48.328125 \nQ 20.21875 56 33.015625 56 \nQ 37.15625 56 41.109375 55.140625 \nQ 45.0625 54.296875 48.78125 52.59375 \nz\n\" id=\"DejaVuSans-99\"/>\n     <path d=\"M 18.3125 70.21875 \nL 18.3125 54.6875 \nL 36.8125 54.6875 \nL 36.8125 47.703125 \nL 18.3125 47.703125 \nL 18.3125 18.015625 \nQ 18.3125 11.328125 20.140625 9.421875 \nQ 21.96875 7.515625 27.59375 7.515625 \nL 36.8125 7.515625 \nL 36.8125 0 \nL 27.59375 0 \nQ 17.1875 0 13.234375 3.875 \nQ 9.28125 7.765625 9.28125 18.015625 \nL 9.28125 47.703125 \nL 2.6875 47.703125 \nL 2.6875 54.6875 \nL 9.28125 54.6875 \nL 9.28125 70.21875 \nz\n\" id=\"DejaVuSans-116\"/>\n     <path d=\"M 9.421875 54.6875 \nL 18.40625 54.6875 \nL 18.40625 0 \nL 9.421875 0 \nz\nM 9.421875 75.984375 \nL 18.40625 75.984375 \nL 18.40625 64.59375 \nL 9.421875 64.59375 \nz\n\" id=\"DejaVuSans-105\"/>\n     <path d=\"M 2.984375 54.6875 \nL 12.5 54.6875 \nL 29.59375 8.796875 \nL 46.6875 54.6875 \nL 56.203125 54.6875 \nL 35.6875 0 \nL 23.484375 0 \nz\n\" id=\"DejaVuSans-118\"/>\n     <path d=\"M 30.609375 48.390625 \nQ 23.390625 48.390625 19.1875 42.75 \nQ 14.984375 37.109375 14.984375 27.296875 \nQ 14.984375 17.484375 19.15625 11.84375 \nQ 23.34375 6.203125 30.609375 6.203125 \nQ 37.796875 6.203125 41.984375 11.859375 \nQ 46.1875 17.53125 46.1875 27.296875 \nQ 46.1875 37.015625 41.984375 42.703125 \nQ 37.796875 48.390625 30.609375 48.390625 \nz\nM 30.609375 56 \nQ 42.328125 56 49.015625 48.375 \nQ 55.71875 40.765625 55.71875 27.296875 \nQ 55.71875 13.875 49.015625 6.21875 \nQ 42.328125 -1.421875 30.609375 -1.421875 \nQ 18.84375 -1.421875 12.171875 6.21875 \nQ 5.515625 13.875 5.515625 27.296875 \nQ 5.515625 40.765625 12.171875 48.375 \nQ 18.84375 56 30.609375 56 \nz\n\" id=\"DejaVuSans-111\"/>\n     <path d=\"M 54.890625 33.015625 \nL 54.890625 0 \nL 45.90625 0 \nL 45.90625 32.71875 \nQ 45.90625 40.484375 42.875 44.328125 \nQ 39.84375 48.1875 33.796875 48.1875 \nQ 26.515625 48.1875 22.3125 43.546875 \nQ 18.109375 38.921875 18.109375 30.90625 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 54.6875 \nL 18.109375 54.6875 \nL 18.109375 46.1875 \nQ 21.34375 51.125 25.703125 53.5625 \nQ 30.078125 56 35.796875 56 \nQ 45.21875 56 50.046875 50.171875 \nQ 54.890625 44.34375 54.890625 33.015625 \nz\n\" id=\"DejaVuSans-110\"/>\n     <path d=\"M 37.109375 75.984375 \nL 37.109375 68.5 \nL 28.515625 68.5 \nQ 23.6875 68.5 21.796875 66.546875 \nQ 19.921875 64.59375 19.921875 59.515625 \nL 19.921875 54.6875 \nL 34.71875 54.6875 \nL 34.71875 47.703125 \nL 19.921875 47.703125 \nL 19.921875 0 \nL 10.890625 0 \nL 10.890625 47.703125 \nL 2.296875 47.703125 \nL 2.296875 54.6875 \nL 10.890625 54.6875 \nL 10.890625 58.5 \nQ 10.890625 67.625 15.140625 71.796875 \nQ 19.390625 75.984375 28.609375 75.984375 \nz\n\" id=\"DejaVuSans-102\"/>\n     <path d=\"M 8.5 21.578125 \nL 8.5 54.6875 \nL 17.484375 54.6875 \nL 17.484375 21.921875 \nQ 17.484375 14.15625 20.5 10.265625 \nQ 23.53125 6.390625 29.59375 6.390625 \nQ 36.859375 6.390625 41.078125 11.03125 \nQ 45.3125 15.671875 45.3125 23.6875 \nL 45.3125 54.6875 \nL 54.296875 54.6875 \nL 54.296875 0 \nL 45.3125 0 \nL 45.3125 8.40625 \nQ 42.046875 3.421875 37.71875 1 \nQ 33.40625 -1.421875 27.6875 -1.421875 \nQ 18.265625 -1.421875 13.375 4.4375 \nQ 8.5 10.296875 8.5 21.578125 \nz\nM 31.109375 56 \nz\n\" id=\"DejaVuSans-117\"/>\n    </defs>\n    <g transform=\"translate(110.911875 17.837812)scale(0.14 -0.14)\">\n     <use xlink:href=\"#DejaVuSans-76\"/>\n     <use x=\"55.697266\" xlink:href=\"#DejaVuSans-101\"/>\n     <use x=\"117.220703\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"178.5\" xlink:href=\"#DejaVuSans-107\"/>\n     <use x=\"236.363281\" xlink:href=\"#DejaVuSans-121\"/>\n     <use x=\"295.542969\" xlink:href=\"#DejaVuSans-32\"/>\n     <use x=\"327.330078\" xlink:href=\"#DejaVuSans-82\"/>\n     <use x=\"396.75\" xlink:href=\"#DejaVuSans-101\"/>\n     <use x=\"458.273438\" xlink:href=\"#DejaVuSans-76\"/>\n     <use x=\"513.908203\" xlink:href=\"#DejaVuSans-85\"/>\n     <use x=\"587.101562\" xlink:href=\"#DejaVuSans-32\"/>\n     <use x=\"618.888672\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"680.167969\" xlink:href=\"#DejaVuSans-99\"/>\n     <use x=\"735.148438\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"774.357422\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"802.140625\" xlink:href=\"#DejaVuSans-118\"/>\n     <use x=\"861.320312\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"922.599609\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"961.808594\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"989.591797\" xlink:href=\"#DejaVuSans-111\"/>\n     <use x=\"1050.773438\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"1114.152344\" xlink:href=\"#DejaVuSans-32\"/>\n     <use x=\"1145.939453\" xlink:href=\"#DejaVuSans-102\"/>\n     <use x=\"1181.144531\" xlink:href=\"#DejaVuSans-117\"/>\n     <use x=\"1244.523438\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"1307.902344\" xlink:href=\"#DejaVuSans-99\"/>\n     <use x=\"1362.882812\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"1402.091797\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"1429.875\" xlink:href=\"#DejaVuSans-111\"/>\n     <use x=\"1491.056641\" xlink:href=\"#DejaVuSans-110\"/>\n    </g>\n   </g>\n  </g>\n </g>\n <defs>\n  <clipPath id=\"p499d76ad51\">\n   <rect height=\"229.4\" width=\"395.775\" x=\"21.835\" y=\"23.837812\"/>\n  </clipPath>\n </defs>\n</svg>\n",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "source": [
    "plt.plot(z, leaky_relu(z, 0.05), \"b-\", linewidth=2)\n",
    "plt.plot([-5, 5], [0, 0], 'k-')\n",
    "plt.plot([0, 0], [-0.5, 4.2], 'k-')\n",
    "plt.grid(True)\n",
    "props = dict(facecolor='black', shrink=0.1)\n",
    "plt.annotate('Leak', xytext=(-3.5, 0.5), xy=(-5, -0.2), arrowprops=props, fontsize=14, ha=\"center\")\n",
    "plt.title(\"Leaky ReLU activation function\", fontsize=14)\n",
    "plt.axis([-5, 5, -0.5, 4.2])\n",
    "\n",
    "save_fig(\"leaky_relu_plot\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "['deserialize',\n 'elu',\n 'exponential',\n 'get',\n 'hard_sigmoid',\n 'linear',\n 'relu',\n 'selu',\n 'serialize',\n 'sigmoid',\n 'softmax',\n 'softplus',\n 'softsign',\n 'swish',\n 'tanh']"
     },
     "metadata": {},
     "execution_count": 51
    }
   ],
   "source": [
    "[m for m in dir(keras.activations) if not m.startswith(\"_\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "['LeakyReLU', 'PReLU', 'ReLU', 'ThresholdedReLU']"
     },
     "metadata": {},
     "execution_count": 53
    }
   ],
   "source": [
    "[m for m in dir(keras.layers) if \"relu\" in m.lower()]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## train a neural network on Fashion MNIST using the Leaky ReLU"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## leaky_ReLU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "(X_train_full, y_train_full), (X_test, y_test) \\\n",
    "= keras.datasets.fashion_mnist.load_data()\n",
    "\n",
    "X_train_full = X_train_full/255.0\n",
    "X_test = X_test/255.0\n",
    "X_valid, X_train = X_train_full[:5000], X_train_full[5000:]\n",
    "y_valid, y_train = y_train_full[:5000], y_train_full[5000:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "tf.random.set_seed(42)\n",
    "np.random.seed(42)\n",
    "\n",
    "model = keras.models.Sequential([\n",
    "    keras.layers.Flatten(input_shape=[28,28]), \n",
    "    keras.layers.Dense(300, kernel_initializer='he_normal'),\n",
    "    keras.layers.LeakyReLU(),\n",
    "    keras.layers.Dense(100, kernel_initializer='he_normal'), \n",
    "    keras.layers.LeakyReLU(),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(\n",
    "    loss='sparse_categorical_crossentropy', \n",
    "    optimizer=keras.optimizers.SGD(lr=1e-3), \n",
    "    metrics=['accuracy']\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 1.2819 - accuracy: 0.6229 - val_loss: 0.8886 - val_accuracy: 0.7160\nEpoch 2/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.7955 - accuracy: 0.7362 - val_loss: 0.7130 - val_accuracy: 0.7656\nEpoch 3/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.6816 - accuracy: 0.7721 - val_loss: 0.6427 - val_accuracy: 0.7898\nEpoch 4/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.6217 - accuracy: 0.7944 - val_loss: 0.5900 - val_accuracy: 0.8066\nEpoch 5/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5832 - accuracy: 0.8075 - val_loss: 0.5582 - val_accuracy: 0.8200\nEpoch 6/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5553 - accuracy: 0.8156 - val_loss: 0.5350 - val_accuracy: 0.8238\nEpoch 7/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5338 - accuracy: 0.8224 - val_loss: 0.5157 - val_accuracy: 0.8304\nEpoch 8/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5173 - accuracy: 0.8272 - val_loss: 0.5079 - val_accuracy: 0.8284\nEpoch 9/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5040 - accuracy: 0.8291 - val_loss: 0.4895 - val_accuracy: 0.8390\nEpoch 10/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.4924 - accuracy: 0.8321 - val_loss: 0.4817 - val_accuracy: 0.8396\n"
    }
   ],
   "source": [
    "histroy = model.fit(\n",
    "    X_train, \n",
    "    y_train, \n",
    "    epochs=10, \n",
    "    validation_data=(X_valid, y_valid)\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## PReLU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "tf.random.set_seed(42)\n",
    "np.random.seed(42)\n",
    "\n",
    "model = keras.models.Sequential([\n",
    "    keras.layers.Flatten(input_shape=[28,28]), \n",
    "    keras.layers.Dense(300, kernel_initializer='he_normal'),\n",
    "    keras.layers.PReLU(),\n",
    "    keras.layers.Dense(100, kernel_initializer='he_normal'), \n",
    "    keras.layers.PReLU(),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(\n",
    "    loss='sparse_categorical_crossentropy', \n",
    "    optimizer=keras.optimizers.SGD(lr=1e-3), \n",
    "    metrics=['accuracy']\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 1.3461 - accuracy: 0.6209 - val_loss: 0.9255 - val_accuracy: 0.7184\nEpoch 2/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.8197 - accuracy: 0.7355 - val_loss: 0.7305 - val_accuracy: 0.7632\nEpoch 3/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.6966 - accuracy: 0.7694 - val_loss: 0.6564 - val_accuracy: 0.7882\nEpoch 4/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.6331 - accuracy: 0.7910 - val_loss: 0.6003 - val_accuracy: 0.8046\nEpoch 5/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5917 - accuracy: 0.8057 - val_loss: 0.5656 - val_accuracy: 0.8184\nEpoch 6/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5618 - accuracy: 0.8135 - val_loss: 0.5406 - val_accuracy: 0.8238\nEpoch 7/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5390 - accuracy: 0.8205 - val_loss: 0.5196 - val_accuracy: 0.8314\nEpoch 8/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5213 - accuracy: 0.8258 - val_loss: 0.5113 - val_accuracy: 0.8314\nEpoch 9/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.5070 - accuracy: 0.8288 - val_loss: 0.4916 - val_accuracy: 0.8378\nEpoch 10/10\n1719/1719 [==============================] - 3s 2ms/step - loss: 0.4945 - accuracy: 0.8315 - val_loss: 0.4826 - val_accuracy: 0.8398\n"
    }
   ],
   "source": [
    "histroy = model.fit(\n",
    "    X_train, \n",
    "    y_train, \n",
    "    epochs=10, \n",
    "    validation_data=(X_valid, y_valid)\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ELU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "def elu(z, alpha=1):\n",
    "    return np.where(z < 0, alpha * (np.exp(z) - 1), z)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Saving figure elu_plot\n"
    },
    {
     "output_type": "display_data",
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/svg+xml": "<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n  \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n<!-- Created with matplotlib (https://matplotlib.org/) -->\n<svg height=\"277.551562pt\" version=\"1.1\" viewBox=\"0 0 424.865625 277.551562\" width=\"424.865625pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n <defs>\n  <style type=\"text/css\">\n*{stroke-linecap:butt;stroke-linejoin:round;}\n  </style>\n </defs>\n <g id=\"figure_1\">\n  <g id=\"patch_1\">\n   <path d=\"M -0 277.551562 \nL 424.865625 277.551562 \nL 424.865625 0 \nL -0 0 \nz\n\" style=\"fill:none;\"/>\n  </g>\n  <g id=\"axes_1\">\n   <g id=\"patch_2\">\n    <path d=\"M 31.890625 251.737812 \nL 417.665625 251.737812 \nL 417.665625 23.837812 \nL 31.890625 23.837812 \nz\n\" style=\"fill:#ffffff;\"/>\n   </g>\n   <g id=\"matplotlib.axis_1\">\n    <g id=\"xtick_1\">\n     <g id=\"line2d_1\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 70.468125 251.737812 \nL 70.468125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_2\">\n      <defs>\n       <path d=\"M 0 0 \nL 0 3.5 \n\" id=\"m62dde856a0\" style=\"stroke:#000000;stroke-width:0.8;\"/>\n      </defs>\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"70.468125\" xlink:href=\"#m62dde856a0\" y=\"251.737812\"/>\n      </g>\n     </g>\n     <g id=\"text_1\">\n      <!-- −4 -->\n      <defs>\n       <path d=\"M 10.59375 35.5 \nL 73.1875 35.5 \nL 73.1875 27.203125 \nL 10.59375 27.203125 \nz\n\" id=\"DejaVuSans-8722\"/>\n       <path d=\"M 37.796875 64.3125 \nL 12.890625 25.390625 \nL 37.796875 25.390625 \nz\nM 35.203125 72.90625 \nL 47.609375 72.90625 \nL 47.609375 25.390625 \nL 58.015625 25.390625 \nL 58.015625 17.1875 \nL 47.609375 17.1875 \nL 47.609375 0 \nL 37.796875 0 \nL 37.796875 17.1875 \nL 4.890625 17.1875 \nL 4.890625 26.703125 \nz\n\" id=\"DejaVuSans-52\"/>\n      </defs>\n      <g transform=\"translate(61.622812 267.855937)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_2\">\n     <g id=\"line2d_3\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 147.623125 251.737812 \nL 147.623125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_4\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"147.623125\" xlink:href=\"#m62dde856a0\" y=\"251.737812\"/>\n      </g>\n     </g>\n     <g id=\"text_2\">\n      <!-- −2 -->\n      <defs>\n       <path d=\"M 19.1875 8.296875 \nL 53.609375 8.296875 \nL 53.609375 0 \nL 7.328125 0 \nL 7.328125 8.296875 \nQ 12.9375 14.109375 22.625 23.890625 \nQ 32.328125 33.6875 34.8125 36.53125 \nQ 39.546875 41.84375 41.421875 45.53125 \nQ 43.3125 49.21875 43.3125 52.78125 \nQ 43.3125 58.59375 39.234375 62.25 \nQ 35.15625 65.921875 28.609375 65.921875 \nQ 23.96875 65.921875 18.8125 64.3125 \nQ 13.671875 62.703125 7.8125 59.421875 \nL 7.8125 69.390625 \nQ 13.765625 71.78125 18.9375 73 \nQ 24.125 74.21875 28.421875 74.21875 \nQ 39.75 74.21875 46.484375 68.546875 \nQ 53.21875 62.890625 53.21875 53.421875 \nQ 53.21875 48.921875 51.53125 44.890625 \nQ 49.859375 40.875 45.40625 35.40625 \nQ 44.1875 33.984375 37.640625 27.21875 \nQ 31.109375 20.453125 19.1875 8.296875 \nz\n\" id=\"DejaVuSans-50\"/>\n      </defs>\n      <g transform=\"translate(138.777812 267.855937)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_3\">\n     <g id=\"line2d_5\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 224.778125 251.737812 \nL 224.778125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_6\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"224.778125\" xlink:href=\"#m62dde856a0\" y=\"251.737812\"/>\n      </g>\n     </g>\n     <g id=\"text_3\">\n      <!-- 0 -->\n      <defs>\n       <path d=\"M 31.78125 66.40625 \nQ 24.171875 66.40625 20.328125 58.90625 \nQ 16.5 51.421875 16.5 36.375 \nQ 16.5 21.390625 20.328125 13.890625 \nQ 24.171875 6.390625 31.78125 6.390625 \nQ 39.453125 6.390625 43.28125 13.890625 \nQ 47.125 21.390625 47.125 36.375 \nQ 47.125 51.421875 43.28125 58.90625 \nQ 39.453125 66.40625 31.78125 66.40625 \nz\nM 31.78125 74.21875 \nQ 44.046875 74.21875 50.515625 64.515625 \nQ 56.984375 54.828125 56.984375 36.375 \nQ 56.984375 17.96875 50.515625 8.265625 \nQ 44.046875 -1.421875 31.78125 -1.421875 \nQ 19.53125 -1.421875 13.0625 8.265625 \nQ 6.59375 17.96875 6.59375 36.375 \nQ 6.59375 54.828125 13.0625 64.515625 \nQ 19.53125 74.21875 31.78125 74.21875 \nz\n\" id=\"DejaVuSans-48\"/>\n      </defs>\n      <g transform=\"translate(220.960625 267.855937)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_4\">\n     <g id=\"line2d_7\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 301.933125 251.737812 \nL 301.933125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_8\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"301.933125\" xlink:href=\"#m62dde856a0\" y=\"251.737812\"/>\n      </g>\n     </g>\n     <g id=\"text_4\">\n      <!-- 2 -->\n      <g transform=\"translate(298.115625 267.855937)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_5\">\n     <g id=\"line2d_9\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 379.088125 251.737812 \nL 379.088125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_10\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"379.088125\" xlink:href=\"#m62dde856a0\" y=\"251.737812\"/>\n      </g>\n     </g>\n     <g id=\"text_5\">\n      <!-- 4 -->\n      <g transform=\"translate(375.270625 267.855937)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n   </g>\n   <g id=\"matplotlib.axis_2\">\n    <g id=\"ytick_1\">\n     <g id=\"line2d_11\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 31.890625 243.297072 \nL 417.665625 243.297072 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_12\">\n      <defs>\n       <path d=\"M 0 0 \nL -3.5 0 \n\" id=\"m9ce129b377\" style=\"stroke:#000000;stroke-width:0.8;\"/>\n      </defs>\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m9ce129b377\" y=\"243.297072\"/>\n      </g>\n     </g>\n     <g id=\"text_6\">\n      <!-- −2 -->\n      <g transform=\"translate(7.2 247.856134)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_2\">\n     <g id=\"line2d_13\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 31.890625 201.093368 \nL 417.665625 201.093368 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_14\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m9ce129b377\" y=\"201.093368\"/>\n      </g>\n     </g>\n     <g id=\"text_7\">\n      <!-- −1 -->\n      <defs>\n       <path d=\"M 12.40625 8.296875 \nL 28.515625 8.296875 \nL 28.515625 63.921875 \nL 10.984375 60.40625 \nL 10.984375 69.390625 \nL 28.421875 72.90625 \nL 38.28125 72.90625 \nL 38.28125 8.296875 \nL 54.390625 8.296875 \nL 54.390625 0 \nL 12.40625 0 \nz\n\" id=\"DejaVuSans-49\"/>\n      </defs>\n      <g transform=\"translate(7.2 205.652431)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-49\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_3\">\n     <g id=\"line2d_15\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 31.890625 158.889664 \nL 417.665625 158.889664 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_16\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m9ce129b377\" y=\"158.889664\"/>\n      </g>\n     </g>\n     <g id=\"text_8\">\n      <!-- 0 -->\n      <g transform=\"translate(17.255625 163.448727)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_4\">\n     <g id=\"line2d_17\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 31.890625 116.685961 \nL 417.665625 116.685961 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_18\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m9ce129b377\" y=\"116.685961\"/>\n      </g>\n     </g>\n     <g id=\"text_9\">\n      <!-- 1 -->\n      <g transform=\"translate(17.255625 121.245023)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-49\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_5\">\n     <g id=\"line2d_19\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 31.890625 74.482257 \nL 417.665625 74.482257 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_20\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m9ce129b377\" y=\"74.482257\"/>\n      </g>\n     </g>\n     <g id=\"text_10\">\n      <!-- 2 -->\n      <g transform=\"translate(17.255625 79.041319)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_6\">\n     <g id=\"line2d_21\">\n      <path clip-path=\"url(#p821fbaac16)\" d=\"M 31.890625 32.278553 \nL 417.665625 32.278553 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_22\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m9ce129b377\" y=\"32.278553\"/>\n      </g>\n     </g>\n     <g id=\"text_11\">\n      <!-- 3 -->\n      <defs>\n       <path d=\"M 40.578125 39.3125 \nQ 47.65625 37.796875 51.625 33 \nQ 55.609375 28.21875 55.609375 21.1875 \nQ 55.609375 10.40625 48.1875 4.484375 \nQ 40.765625 -1.421875 27.09375 -1.421875 \nQ 22.515625 -1.421875 17.65625 -0.515625 \nQ 12.796875 0.390625 7.625 2.203125 \nL 7.625 11.71875 \nQ 11.71875 9.328125 16.59375 8.109375 \nQ 21.484375 6.890625 26.8125 6.890625 \nQ 36.078125 6.890625 40.9375 10.546875 \nQ 45.796875 14.203125 45.796875 21.1875 \nQ 45.796875 27.640625 41.28125 31.265625 \nQ 36.765625 34.90625 28.71875 34.90625 \nL 20.21875 34.90625 \nL 20.21875 43.015625 \nL 29.109375 43.015625 \nQ 36.375 43.015625 40.234375 45.921875 \nQ 44.09375 48.828125 44.09375 54.296875 \nQ 44.09375 59.90625 40.109375 62.90625 \nQ 36.140625 65.921875 28.71875 65.921875 \nQ 24.65625 65.921875 20.015625 65.03125 \nQ 15.375 64.15625 9.8125 62.3125 \nL 9.8125 71.09375 \nQ 15.4375 72.65625 20.34375 73.4375 \nQ 25.25 74.21875 29.59375 74.21875 \nQ 40.828125 74.21875 47.359375 69.109375 \nQ 53.90625 64.015625 53.90625 55.328125 \nQ 53.90625 49.265625 50.4375 45.09375 \nQ 46.96875 40.921875 40.578125 39.3125 \nz\n\" id=\"DejaVuSans-51\"/>\n      </defs>\n      <g transform=\"translate(17.255625 36.837616)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-51\"/>\n      </g>\n     </g>\n    </g>\n   </g>\n   <g id=\"line2d_23\">\n    <path clip-path=\"url(#p821fbaac16)\" d=\"M 31.890625 200.809002 \nL 60.969143 200.489091 \nL 82.293389 200.043105 \nL 97.801932 199.523407 \nL 111.371906 198.861566 \nL 123.003313 198.076205 \nL 132.696153 197.214384 \nL 142.388992 196.106393 \nL 150.143263 194.996138 \nL 157.897535 193.638706 \nL 163.713238 192.425755 \nL 169.528942 191.015445 \nL 175.344645 189.375663 \nL 181.160349 187.469071 \nL 186.976052 185.252258 \nL 190.853188 183.577434 \nL 194.730323 181.725537 \nL 198.607459 179.677845 \nL 202.484595 177.413659 \nL 206.361731 174.910088 \nL 210.238866 172.141823 \nL 214.116002 169.08088 \nL 217.993138 165.696314 \nL 221.870273 161.953911 \nL 225.747409 157.82927 \nL 370.929836 -1 \nL 370.929836 -1 \n\" style=\"fill:none;stroke:#0000ff;stroke-linecap:square;stroke-width:2;\"/>\n   </g>\n   <g id=\"line2d_24\">\n    <path clip-path=\"url(#p821fbaac16)\" d=\"M 31.890625 158.889664 \nL 417.665625 158.889664 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"line2d_25\">\n    <path clip-path=\"url(#p821fbaac16)\" d=\"M 31.890625 201.093368 \nL 417.665625 201.093368 \n\" style=\"fill:none;stroke:#000000;stroke-dasharray:5.55,2.4;stroke-dashoffset:0;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"line2d_26\">\n    <path clip-path=\"url(#p821fbaac16)\" d=\"M 224.778125 251.737812 \nL 224.778125 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"patch_3\">\n    <path d=\"M 31.890625 251.737812 \nL 31.890625 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_4\">\n    <path d=\"M 417.665625 251.737812 \nL 417.665625 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_5\">\n    <path d=\"M 31.890625 251.737812 \nL 417.665625 251.737812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_6\">\n    <path d=\"M 31.890625 23.837812 \nL 417.665625 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"text_12\">\n    <!-- ELU activation function ($\\alpha=1$) -->\n    <defs>\n     <path d=\"M 9.8125 72.90625 \nL 55.90625 72.90625 \nL 55.90625 64.59375 \nL 19.671875 64.59375 \nL 19.671875 43.015625 \nL 54.390625 43.015625 \nL 54.390625 34.71875 \nL 19.671875 34.71875 \nL 19.671875 8.296875 \nL 56.78125 8.296875 \nL 56.78125 0 \nL 9.8125 0 \nz\n\" id=\"DejaVuSans-69\"/>\n     <path d=\"M 9.8125 72.90625 \nL 19.671875 72.90625 \nL 19.671875 8.296875 \nL 55.171875 8.296875 \nL 55.171875 0 \nL 9.8125 0 \nz\n\" id=\"DejaVuSans-76\"/>\n     <path d=\"M 8.6875 72.90625 \nL 18.609375 72.90625 \nL 18.609375 28.609375 \nQ 18.609375 16.890625 22.84375 11.734375 \nQ 27.09375 6.59375 36.625 6.59375 \nQ 46.09375 6.59375 50.34375 11.734375 \nQ 54.59375 16.890625 54.59375 28.609375 \nL 54.59375 72.90625 \nL 64.5 72.90625 \nL 64.5 27.390625 \nQ 64.5 13.140625 57.4375 5.859375 \nQ 50.390625 -1.421875 36.625 -1.421875 \nQ 22.796875 -1.421875 15.734375 5.859375 \nQ 8.6875 13.140625 8.6875 27.390625 \nz\n\" id=\"DejaVuSans-85\"/>\n     <path id=\"DejaVuSans-32\"/>\n     <path d=\"M 34.28125 27.484375 \nQ 23.390625 27.484375 19.1875 25 \nQ 14.984375 22.515625 14.984375 16.5 \nQ 14.984375 11.71875 18.140625 8.90625 \nQ 21.296875 6.109375 26.703125 6.109375 \nQ 34.1875 6.109375 38.703125 11.40625 \nQ 43.21875 16.703125 43.21875 25.484375 \nL 43.21875 27.484375 \nz\nM 52.203125 31.203125 \nL 52.203125 0 \nL 43.21875 0 \nL 43.21875 8.296875 \nQ 40.140625 3.328125 35.546875 0.953125 \nQ 30.953125 -1.421875 24.3125 -1.421875 \nQ 15.921875 -1.421875 10.953125 3.296875 \nQ 6 8.015625 6 15.921875 \nQ 6 25.140625 12.171875 29.828125 \nQ 18.359375 34.515625 30.609375 34.515625 \nL 43.21875 34.515625 \nL 43.21875 35.40625 \nQ 43.21875 41.609375 39.140625 45 \nQ 35.0625 48.390625 27.6875 48.390625 \nQ 23 48.390625 18.546875 47.265625 \nQ 14.109375 46.140625 10.015625 43.890625 \nL 10.015625 52.203125 \nQ 14.9375 54.109375 19.578125 55.046875 \nQ 24.21875 56 28.609375 56 \nQ 40.484375 56 46.34375 49.84375 \nQ 52.203125 43.703125 52.203125 31.203125 \nz\n\" id=\"DejaVuSans-97\"/>\n     <path d=\"M 48.78125 52.59375 \nL 48.78125 44.1875 \nQ 44.96875 46.296875 41.140625 47.34375 \nQ 37.3125 48.390625 33.40625 48.390625 \nQ 24.65625 48.390625 19.8125 42.84375 \nQ 14.984375 37.3125 14.984375 27.296875 \nQ 14.984375 17.28125 19.8125 11.734375 \nQ 24.65625 6.203125 33.40625 6.203125 \nQ 37.3125 6.203125 41.140625 7.25 \nQ 44.96875 8.296875 48.78125 10.40625 \nL 48.78125 2.09375 \nQ 45.015625 0.34375 40.984375 -0.53125 \nQ 36.96875 -1.421875 32.421875 -1.421875 \nQ 20.0625 -1.421875 12.78125 6.34375 \nQ 5.515625 14.109375 5.515625 27.296875 \nQ 5.515625 40.671875 12.859375 48.328125 \nQ 20.21875 56 33.015625 56 \nQ 37.15625 56 41.109375 55.140625 \nQ 45.0625 54.296875 48.78125 52.59375 \nz\n\" id=\"DejaVuSans-99\"/>\n     <path d=\"M 18.3125 70.21875 \nL 18.3125 54.6875 \nL 36.8125 54.6875 \nL 36.8125 47.703125 \nL 18.3125 47.703125 \nL 18.3125 18.015625 \nQ 18.3125 11.328125 20.140625 9.421875 \nQ 21.96875 7.515625 27.59375 7.515625 \nL 36.8125 7.515625 \nL 36.8125 0 \nL 27.59375 0 \nQ 17.1875 0 13.234375 3.875 \nQ 9.28125 7.765625 9.28125 18.015625 \nL 9.28125 47.703125 \nL 2.6875 47.703125 \nL 2.6875 54.6875 \nL 9.28125 54.6875 \nL 9.28125 70.21875 \nz\n\" id=\"DejaVuSans-116\"/>\n     <path d=\"M 9.421875 54.6875 \nL 18.40625 54.6875 \nL 18.40625 0 \nL 9.421875 0 \nz\nM 9.421875 75.984375 \nL 18.40625 75.984375 \nL 18.40625 64.59375 \nL 9.421875 64.59375 \nz\n\" id=\"DejaVuSans-105\"/>\n     <path d=\"M 2.984375 54.6875 \nL 12.5 54.6875 \nL 29.59375 8.796875 \nL 46.6875 54.6875 \nL 56.203125 54.6875 \nL 35.6875 0 \nL 23.484375 0 \nz\n\" id=\"DejaVuSans-118\"/>\n     <path d=\"M 30.609375 48.390625 \nQ 23.390625 48.390625 19.1875 42.75 \nQ 14.984375 37.109375 14.984375 27.296875 \nQ 14.984375 17.484375 19.15625 11.84375 \nQ 23.34375 6.203125 30.609375 6.203125 \nQ 37.796875 6.203125 41.984375 11.859375 \nQ 46.1875 17.53125 46.1875 27.296875 \nQ 46.1875 37.015625 41.984375 42.703125 \nQ 37.796875 48.390625 30.609375 48.390625 \nz\nM 30.609375 56 \nQ 42.328125 56 49.015625 48.375 \nQ 55.71875 40.765625 55.71875 27.296875 \nQ 55.71875 13.875 49.015625 6.21875 \nQ 42.328125 -1.421875 30.609375 -1.421875 \nQ 18.84375 -1.421875 12.171875 6.21875 \nQ 5.515625 13.875 5.515625 27.296875 \nQ 5.515625 40.765625 12.171875 48.375 \nQ 18.84375 56 30.609375 56 \nz\n\" id=\"DejaVuSans-111\"/>\n     <path d=\"M 54.890625 33.015625 \nL 54.890625 0 \nL 45.90625 0 \nL 45.90625 32.71875 \nQ 45.90625 40.484375 42.875 44.328125 \nQ 39.84375 48.1875 33.796875 48.1875 \nQ 26.515625 48.1875 22.3125 43.546875 \nQ 18.109375 38.921875 18.109375 30.90625 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 54.6875 \nL 18.109375 54.6875 \nL 18.109375 46.1875 \nQ 21.34375 51.125 25.703125 53.5625 \nQ 30.078125 56 35.796875 56 \nQ 45.21875 56 50.046875 50.171875 \nQ 54.890625 44.34375 54.890625 33.015625 \nz\n\" id=\"DejaVuSans-110\"/>\n     <path d=\"M 37.109375 75.984375 \nL 37.109375 68.5 \nL 28.515625 68.5 \nQ 23.6875 68.5 21.796875 66.546875 \nQ 19.921875 64.59375 19.921875 59.515625 \nL 19.921875 54.6875 \nL 34.71875 54.6875 \nL 34.71875 47.703125 \nL 19.921875 47.703125 \nL 19.921875 0 \nL 10.890625 0 \nL 10.890625 47.703125 \nL 2.296875 47.703125 \nL 2.296875 54.6875 \nL 10.890625 54.6875 \nL 10.890625 58.5 \nQ 10.890625 67.625 15.140625 71.796875 \nQ 19.390625 75.984375 28.609375 75.984375 \nz\n\" id=\"DejaVuSans-102\"/>\n     <path d=\"M 8.5 21.578125 \nL 8.5 54.6875 \nL 17.484375 54.6875 \nL 17.484375 21.921875 \nQ 17.484375 14.15625 20.5 10.265625 \nQ 23.53125 6.390625 29.59375 6.390625 \nQ 36.859375 6.390625 41.078125 11.03125 \nQ 45.3125 15.671875 45.3125 23.6875 \nL 45.3125 54.6875 \nL 54.296875 54.6875 \nL 54.296875 0 \nL 45.3125 0 \nL 45.3125 8.40625 \nQ 42.046875 3.421875 37.71875 1 \nQ 33.40625 -1.421875 27.6875 -1.421875 \nQ 18.265625 -1.421875 13.375 4.4375 \nQ 8.5 10.296875 8.5 21.578125 \nz\nM 31.109375 56 \nz\n\" id=\"DejaVuSans-117\"/>\n     <path d=\"M 31 75.875 \nQ 24.46875 64.65625 21.28125 53.65625 \nQ 18.109375 42.671875 18.109375 31.390625 \nQ 18.109375 20.125 21.3125 9.0625 \nQ 24.515625 -2 31 -13.1875 \nL 23.1875 -13.1875 \nQ 15.875 -1.703125 12.234375 9.375 \nQ 8.59375 20.453125 8.59375 31.390625 \nQ 8.59375 42.28125 12.203125 53.3125 \nQ 15.828125 64.359375 23.1875 75.875 \nz\n\" id=\"DejaVuSans-40\"/>\n     <path d=\"M 40.921875 25.4375 \nL 40.96875 36.71875 \nQ 41.015625 48.25 32.328125 48.296875 \nQ 25.828125 48.34375 21.78125 42.921875 \nQ 16.703125 36.234375 14.984375 27.296875 \nQ 12.890625 16.546875 15.53125 11.421875 \nQ 18.265625 6.203125 24.171875 6.203125 \nQ 30.71875 6.203125 36.234375 16.609375 \nz\nM 33.84375 55.90625 \nQ 49.078125 56.15625 48.875 40.375 \nQ 48.875 40.375 56.5 54.6875 \nL 64.5 54.6875 \nL 48.734375 25.046875 \nL 48.578125 14.359375 \nQ 48.578125 11.96875 49.90625 9.96875 \nQ 51.421875 7.625 52.984375 7.625 \nL 57.328125 7.625 \nL 55.859375 0 \nL 50.4375 0 \nQ 45.84375 0 42.53125 4.109375 \nQ 40.96875 6.15625 40.921875 10.453125 \nQ 37.75 5.21875 32.28125 0.78125 \nQ 29.6875 -1.265625 22.703125 -1.21875 \nQ 11.28125 -1.125 7.125 6.203125 \nQ 2.875 13.8125 5.515625 27.296875 \nQ 8.34375 41.796875 15.765625 48.390625 \nQ 24.125 55.765625 33.84375 55.90625 \nz\n\" id=\"DejaVuSans-Oblique-945\"/>\n     <path d=\"M 10.59375 45.40625 \nL 73.1875 45.40625 \nL 73.1875 37.203125 \nL 10.59375 37.203125 \nz\nM 10.59375 25.484375 \nL 73.1875 25.484375 \nL 73.1875 17.1875 \nL 10.59375 17.1875 \nz\n\" id=\"DejaVuSans-61\"/>\n     <path d=\"M 8.015625 75.875 \nL 15.828125 75.875 \nQ 23.140625 64.359375 26.78125 53.3125 \nQ 30.421875 42.28125 30.421875 31.390625 \nQ 30.421875 20.453125 26.78125 9.375 \nQ 23.140625 -1.703125 15.828125 -13.1875 \nL 8.015625 -13.1875 \nQ 14.5 -2 17.703125 9.0625 \nQ 20.90625 20.125 20.90625 31.390625 \nQ 20.90625 42.671875 17.703125 53.65625 \nQ 14.5 64.65625 8.015625 75.875 \nz\n\" id=\"DejaVuSans-41\"/>\n    </defs>\n    <g transform=\"translate(118.238125 17.837812)scale(0.14 -0.14)\">\n     <use transform=\"translate(0 0.015625)\" xlink:href=\"#DejaVuSans-69\"/>\n     <use transform=\"translate(63.183594 0.015625)\" xlink:href=\"#DejaVuSans-76\"/>\n     <use transform=\"translate(118.896484 0.015625)\" xlink:href=\"#DejaVuSans-85\"/>\n     <use transform=\"translate(192.089844 0.015625)\" xlink:href=\"#DejaVuSans-32\"/>\n     <use transform=\"translate(223.876953 0.015625)\" xlink:href=\"#DejaVuSans-97\"/>\n     <use transform=\"translate(285.15625 0.015625)\" xlink:href=\"#DejaVuSans-99\"/>\n     <use transform=\"translate(340.136719 0.015625)\" xlink:href=\"#DejaVuSans-116\"/>\n     <use transform=\"translate(379.345703 0.015625)\" xlink:href=\"#DejaVuSans-105\"/>\n     <use transform=\"translate(407.128906 0.015625)\" xlink:href=\"#DejaVuSans-118\"/>\n     <use transform=\"translate(466.308594 0.015625)\" xlink:href=\"#DejaVuSans-97\"/>\n     <use transform=\"translate(527.587891 0.015625)\" xlink:href=\"#DejaVuSans-116\"/>\n     <use transform=\"translate(566.796875 0.015625)\" xlink:href=\"#DejaVuSans-105\"/>\n     <use transform=\"translate(594.580078 0.015625)\" xlink:href=\"#DejaVuSans-111\"/>\n     <use transform=\"translate(655.761719 0.015625)\" xlink:href=\"#DejaVuSans-110\"/>\n     <use transform=\"translate(719.140625 0.015625)\" xlink:href=\"#DejaVuSans-32\"/>\n     <use transform=\"translate(750.927734 0.015625)\" xlink:href=\"#DejaVuSans-102\"/>\n     <use transform=\"translate(786.132812 0.015625)\" xlink:href=\"#DejaVuSans-117\"/>\n     <use transform=\"translate(849.511719 0.015625)\" xlink:href=\"#DejaVuSans-110\"/>\n     <use transform=\"translate(912.890625 0.015625)\" xlink:href=\"#DejaVuSans-99\"/>\n     <use transform=\"translate(967.871094 0.015625)\" xlink:href=\"#DejaVuSans-116\"/>\n     <use transform=\"translate(1007.080078 0.015625)\" xlink:href=\"#DejaVuSans-105\"/>\n     <use transform=\"translate(1034.863281 0.015625)\" xlink:href=\"#DejaVuSans-111\"/>\n     <use transform=\"translate(1096.044922 0.015625)\" xlink:href=\"#DejaVuSans-110\"/>\n     <use transform=\"translate(1159.423828 0.015625)\" xlink:href=\"#DejaVuSans-32\"/>\n     <use transform=\"translate(1191.210938 0.015625)\" xlink:href=\"#DejaVuSans-40\"/>\n     <use transform=\"translate(1230.224609 0.015625)\" xlink:href=\"#DejaVuSans-Oblique-945\"/>\n     <use transform=\"translate(1315.625 0.015625)\" xlink:href=\"#DejaVuSans-61\"/>\n     <use transform=\"translate(1418.896484 0.015625)\" xlink:href=\"#DejaVuSans-49\"/>\n     <use transform=\"translate(1482.519531 0.015625)\" xlink:href=\"#DejaVuSans-41\"/>\n    </g>\n   </g>\n  </g>\n </g>\n <defs>\n  <clipPath id=\"p821fbaac16\">\n   <rect height=\"227.9\" width=\"385.775\" x=\"31.890625\" y=\"23.837812\"/>\n  </clipPath>\n </defs>\n</svg>\n",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "source": [
    "plt.plot(z, elu(z), \"b-\", linewidth=2)\n",
    "plt.plot([-5, 5], [0, 0], 'k-')\n",
    "plt.plot([-5, 5], [-1, -1], 'k--')\n",
    "plt.plot([0, 0], [-2.2, 3.2], 'k-')\n",
    "plt.grid(True)\n",
    "plt.title(r\"ELU activation function ($\\alpha=1$)\", fontsize=14)\n",
    "plt.axis([-5, 5, -2.2, 3.2])\n",
    "\n",
    "save_fig(\"elu_plot\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "<tensorflow.python.keras.layers.core.Dense at 0x7f8cf1b64d50>"
     },
     "metadata": {},
     "execution_count": 73
    }
   ],
   "source": [
    "keras.layers.Dense(10, activation=\"elu\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SELU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.special import erfc\n",
    "\n",
    "# alpha and scale to self normalize with mean 0 and standard deviation 1\n",
    "# (see equation 14 in the paper):\n",
    "alpha_0_1 = -np.sqrt(2 / np.pi) / (erfc(1/np.sqrt(2)) * np.exp(1/2) - 1)\n",
    "scale_0_1 = (1 - erfc(1 / np.sqrt(2)) * np.sqrt(np.e)) * np.sqrt(2 * np.pi) * (2 * erfc(np.sqrt(2))*np.e**2 + np.pi*erfc(1/np.sqrt(2))**2*np.e - 2*(2+np.pi)*erfc(1/np.sqrt(2))*np.sqrt(np.e)+np.pi+2)**(-1/2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "def selu(z, scale=scale_0_1, alpha=alpha_0_1):\n",
    "    return scale * elu(z, alpha)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Saving figure selu_plot\n"
    },
    {
     "output_type": "display_data",
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/svg+xml": "<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n  \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n<!-- Created with matplotlib (https://matplotlib.org/) -->\n<svg height=\"279.051563pt\" version=\"1.1\" viewBox=\"0 0 424.865625 279.051563\" width=\"424.865625pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n <defs>\n  <style type=\"text/css\">\n*{stroke-linecap:butt;stroke-linejoin:round;}\n  </style>\n </defs>\n <g id=\"figure_1\">\n  <g id=\"patch_1\">\n   <path d=\"M -0 279.051563 \nL 424.865625 279.051563 \nL 424.865625 0 \nL -0 0 \nz\n\" style=\"fill:none;\"/>\n  </g>\n  <g id=\"axes_1\">\n   <g id=\"patch_2\">\n    <path d=\"M 31.890625 253.237813 \nL 417.665625 253.237813 \nL 417.665625 23.837812 \nL 31.890625 23.837812 \nz\n\" style=\"fill:#ffffff;\"/>\n   </g>\n   <g id=\"matplotlib.axis_1\">\n    <g id=\"xtick_1\">\n     <g id=\"line2d_1\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 70.468125 253.237813 \nL 70.468125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_2\">\n      <defs>\n       <path d=\"M 0 0 \nL 0 3.5 \n\" id=\"mdef504ddcb\" style=\"stroke:#000000;stroke-width:0.8;\"/>\n      </defs>\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"70.468125\" xlink:href=\"#mdef504ddcb\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_1\">\n      <!-- −4 -->\n      <defs>\n       <path d=\"M 10.59375 35.5 \nL 73.1875 35.5 \nL 73.1875 27.203125 \nL 10.59375 27.203125 \nz\n\" id=\"DejaVuSans-8722\"/>\n       <path d=\"M 37.796875 64.3125 \nL 12.890625 25.390625 \nL 37.796875 25.390625 \nz\nM 35.203125 72.90625 \nL 47.609375 72.90625 \nL 47.609375 25.390625 \nL 58.015625 25.390625 \nL 58.015625 17.1875 \nL 47.609375 17.1875 \nL 47.609375 0 \nL 37.796875 0 \nL 37.796875 17.1875 \nL 4.890625 17.1875 \nL 4.890625 26.703125 \nz\n\" id=\"DejaVuSans-52\"/>\n      </defs>\n      <g transform=\"translate(61.622812 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_2\">\n     <g id=\"line2d_3\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 147.623125 253.237813 \nL 147.623125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_4\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"147.623125\" xlink:href=\"#mdef504ddcb\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_2\">\n      <!-- −2 -->\n      <defs>\n       <path d=\"M 19.1875 8.296875 \nL 53.609375 8.296875 \nL 53.609375 0 \nL 7.328125 0 \nL 7.328125 8.296875 \nQ 12.9375 14.109375 22.625 23.890625 \nQ 32.328125 33.6875 34.8125 36.53125 \nQ 39.546875 41.84375 41.421875 45.53125 \nQ 43.3125 49.21875 43.3125 52.78125 \nQ 43.3125 58.59375 39.234375 62.25 \nQ 35.15625 65.921875 28.609375 65.921875 \nQ 23.96875 65.921875 18.8125 64.3125 \nQ 13.671875 62.703125 7.8125 59.421875 \nL 7.8125 69.390625 \nQ 13.765625 71.78125 18.9375 73 \nQ 24.125 74.21875 28.421875 74.21875 \nQ 39.75 74.21875 46.484375 68.546875 \nQ 53.21875 62.890625 53.21875 53.421875 \nQ 53.21875 48.921875 51.53125 44.890625 \nQ 49.859375 40.875 45.40625 35.40625 \nQ 44.1875 33.984375 37.640625 27.21875 \nQ 31.109375 20.453125 19.1875 8.296875 \nz\n\" id=\"DejaVuSans-50\"/>\n      </defs>\n      <g transform=\"translate(138.777812 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_3\">\n     <g id=\"line2d_5\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 224.778125 253.237813 \nL 224.778125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_6\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"224.778125\" xlink:href=\"#mdef504ddcb\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_3\">\n      <!-- 0 -->\n      <defs>\n       <path d=\"M 31.78125 66.40625 \nQ 24.171875 66.40625 20.328125 58.90625 \nQ 16.5 51.421875 16.5 36.375 \nQ 16.5 21.390625 20.328125 13.890625 \nQ 24.171875 6.390625 31.78125 6.390625 \nQ 39.453125 6.390625 43.28125 13.890625 \nQ 47.125 21.390625 47.125 36.375 \nQ 47.125 51.421875 43.28125 58.90625 \nQ 39.453125 66.40625 31.78125 66.40625 \nz\nM 31.78125 74.21875 \nQ 44.046875 74.21875 50.515625 64.515625 \nQ 56.984375 54.828125 56.984375 36.375 \nQ 56.984375 17.96875 50.515625 8.265625 \nQ 44.046875 -1.421875 31.78125 -1.421875 \nQ 19.53125 -1.421875 13.0625 8.265625 \nQ 6.59375 17.96875 6.59375 36.375 \nQ 6.59375 54.828125 13.0625 64.515625 \nQ 19.53125 74.21875 31.78125 74.21875 \nz\n\" id=\"DejaVuSans-48\"/>\n      </defs>\n      <g transform=\"translate(220.960625 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_4\">\n     <g id=\"line2d_7\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 301.933125 253.237813 \nL 301.933125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_8\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"301.933125\" xlink:href=\"#mdef504ddcb\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_4\">\n      <!-- 2 -->\n      <g transform=\"translate(298.115625 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"xtick_5\">\n     <g id=\"line2d_9\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 379.088125 253.237813 \nL 379.088125 23.837812 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_10\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"379.088125\" xlink:href=\"#mdef504ddcb\" y=\"253.237813\"/>\n      </g>\n     </g>\n     <g id=\"text_5\">\n      <!-- 4 -->\n      <g transform=\"translate(375.270625 269.355938)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-52\"/>\n      </g>\n     </g>\n    </g>\n   </g>\n   <g id=\"matplotlib.axis_2\">\n    <g id=\"ytick_1\">\n     <g id=\"line2d_11\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 31.890625 244.741516 \nL 417.665625 244.741516 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_12\">\n      <defs>\n       <path d=\"M 0 0 \nL -3.5 0 \n\" id=\"m8f1fa62c4a\" style=\"stroke:#000000;stroke-width:0.8;\"/>\n      </defs>\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m8f1fa62c4a\" y=\"244.741516\"/>\n      </g>\n     </g>\n     <g id=\"text_6\">\n      <!-- −2 -->\n      <g transform=\"translate(7.2 249.300579)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_2\">\n     <g id=\"line2d_13\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 31.890625 202.260035 \nL 417.665625 202.260035 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_14\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m8f1fa62c4a\" y=\"202.260035\"/>\n      </g>\n     </g>\n     <g id=\"text_7\">\n      <!-- −1 -->\n      <defs>\n       <path d=\"M 12.40625 8.296875 \nL 28.515625 8.296875 \nL 28.515625 63.921875 \nL 10.984375 60.40625 \nL 10.984375 69.390625 \nL 28.421875 72.90625 \nL 38.28125 72.90625 \nL 38.28125 8.296875 \nL 54.390625 8.296875 \nL 54.390625 0 \nL 12.40625 0 \nz\n\" id=\"DejaVuSans-49\"/>\n      </defs>\n      <g transform=\"translate(7.2 206.819097)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-8722\"/>\n       <use x=\"83.789062\" xlink:href=\"#DejaVuSans-49\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_3\">\n     <g id=\"line2d_15\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 31.890625 159.778553 \nL 417.665625 159.778553 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_16\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m8f1fa62c4a\" y=\"159.778553\"/>\n      </g>\n     </g>\n     <g id=\"text_8\">\n      <!-- 0 -->\n      <g transform=\"translate(17.255625 164.337616)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-48\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_4\">\n     <g id=\"line2d_17\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 31.890625 117.297072 \nL 417.665625 117.297072 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_18\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m8f1fa62c4a\" y=\"117.297072\"/>\n      </g>\n     </g>\n     <g id=\"text_9\">\n      <!-- 1 -->\n      <g transform=\"translate(17.255625 121.856134)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-49\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_5\">\n     <g id=\"line2d_19\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 31.890625 74.81559 \nL 417.665625 74.81559 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_20\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m8f1fa62c4a\" y=\"74.81559\"/>\n      </g>\n     </g>\n     <g id=\"text_10\">\n      <!-- 2 -->\n      <g transform=\"translate(17.255625 79.374653)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-50\"/>\n      </g>\n     </g>\n    </g>\n    <g id=\"ytick_6\">\n     <g id=\"line2d_21\">\n      <path clip-path=\"url(#p3879959fc6)\" d=\"M 31.890625 32.334109 \nL 417.665625 32.334109 \n\" style=\"fill:none;stroke:#b0b0b0;stroke-linecap:square;stroke-width:0.8;\"/>\n     </g>\n     <g id=\"line2d_22\">\n      <g>\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"31.890625\" xlink:href=\"#m8f1fa62c4a\" y=\"32.334109\"/>\n      </g>\n     </g>\n     <g id=\"text_11\">\n      <!-- 3 -->\n      <defs>\n       <path d=\"M 40.578125 39.3125 \nQ 47.65625 37.796875 51.625 33 \nQ 55.609375 28.21875 55.609375 21.1875 \nQ 55.609375 10.40625 48.1875 4.484375 \nQ 40.765625 -1.421875 27.09375 -1.421875 \nQ 22.515625 -1.421875 17.65625 -0.515625 \nQ 12.796875 0.390625 7.625 2.203125 \nL 7.625 11.71875 \nQ 11.71875 9.328125 16.59375 8.109375 \nQ 21.484375 6.890625 26.8125 6.890625 \nQ 36.078125 6.890625 40.9375 10.546875 \nQ 45.796875 14.203125 45.796875 21.1875 \nQ 45.796875 27.640625 41.28125 31.265625 \nQ 36.765625 34.90625 28.71875 34.90625 \nL 20.21875 34.90625 \nL 20.21875 43.015625 \nL 29.109375 43.015625 \nQ 36.375 43.015625 40.234375 45.921875 \nQ 44.09375 48.828125 44.09375 54.296875 \nQ 44.09375 59.90625 40.109375 62.90625 \nQ 36.140625 65.921875 28.71875 65.921875 \nQ 24.65625 65.921875 20.015625 65.03125 \nQ 15.375 64.15625 9.8125 62.3125 \nL 9.8125 71.09375 \nQ 15.4375 72.65625 20.34375 73.4375 \nQ 25.25 74.21875 29.59375 74.21875 \nQ 40.828125 74.21875 47.359375 69.109375 \nQ 53.90625 64.015625 53.90625 55.328125 \nQ 53.90625 49.265625 50.4375 45.09375 \nQ 46.96875 40.921875 40.578125 39.3125 \nz\n\" id=\"DejaVuSans-51\"/>\n      </defs>\n      <g transform=\"translate(17.255625 36.893171)scale(0.12 -0.12)\">\n       <use xlink:href=\"#DejaVuSans-51\"/>\n      </g>\n     </g>\n    </g>\n   </g>\n   <g id=\"line2d_23\">\n    <path clip-path=\"url(#p3879959fc6)\" d=\"M 31.890625 233.961983 \nL 55.153439 233.545495 \nL 72.60055 233.019541 \nL 86.170524 232.410093 \nL 97.801932 231.686904 \nL 107.494771 230.893308 \nL 117.18761 229.873029 \nL 124.941881 228.850665 \nL 132.696153 227.600693 \nL 138.511856 226.483763 \nL 144.32756 225.185098 \nL 150.143263 223.675127 \nL 155.958967 221.919468 \nL 161.77467 219.878145 \nL 167.590374 217.504677 \nL 171.467509 215.7115 \nL 175.344645 213.728737 \nL 179.221781 211.536343 \nL 183.098916 209.112155 \nL 186.976052 206.431666 \nL 190.853188 203.467779 \nL 194.730323 200.190531 \nL 198.607459 196.566791 \nL 202.484595 192.559926 \nL 206.361731 188.12943 \nL 210.238866 183.230513 \nL 214.116002 177.813651 \nL 217.993138 171.824084 \nL 221.870273 165.201261 \nL 223.808841 161.631724 \nL 225.747409 158.657062 \nL 363.736076 -1 \nL 363.736076 -1 \n\" style=\"fill:none;stroke:#0000ff;stroke-linecap:square;stroke-width:2;\"/>\n   </g>\n   <g id=\"line2d_24\">\n    <path clip-path=\"url(#p3879959fc6)\" d=\"M 31.890625 159.778553 \nL 417.665625 159.778553 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"line2d_25\">\n    <path clip-path=\"url(#p3879959fc6)\" d=\"M 31.890625 234.460998 \nL 417.665625 234.460998 \n\" style=\"fill:none;stroke:#000000;stroke-dasharray:5.55,2.4;stroke-dashoffset:0;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"line2d_26\">\n    <path clip-path=\"url(#p3879959fc6)\" d=\"M 224.778125 253.237813 \nL 224.778125 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-width:1.5;\"/>\n   </g>\n   <g id=\"patch_3\">\n    <path d=\"M 31.890625 253.237813 \nL 31.890625 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_4\">\n    <path d=\"M 417.665625 253.237813 \nL 417.665625 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_5\">\n    <path d=\"M 31.890625 253.237813 \nL 417.665625 253.237813 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"patch_6\">\n    <path d=\"M 31.890625 23.837812 \nL 417.665625 23.837812 \n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\n   </g>\n   <g id=\"text_12\">\n    <!-- SELU activation function -->\n    <defs>\n     <path d=\"M 53.515625 70.515625 \nL 53.515625 60.890625 \nQ 47.90625 63.578125 42.921875 64.890625 \nQ 37.9375 66.21875 33.296875 66.21875 \nQ 25.25 66.21875 20.875 63.09375 \nQ 16.5 59.96875 16.5 54.203125 \nQ 16.5 49.359375 19.40625 46.890625 \nQ 22.3125 44.4375 30.421875 42.921875 \nL 36.375 41.703125 \nQ 47.40625 39.59375 52.65625 34.296875 \nQ 57.90625 29 57.90625 20.125 \nQ 57.90625 9.515625 50.796875 4.046875 \nQ 43.703125 -1.421875 29.984375 -1.421875 \nQ 24.8125 -1.421875 18.96875 -0.25 \nQ 13.140625 0.921875 6.890625 3.21875 \nL 6.890625 13.375 \nQ 12.890625 10.015625 18.65625 8.296875 \nQ 24.421875 6.59375 29.984375 6.59375 \nQ 38.421875 6.59375 43.015625 9.90625 \nQ 47.609375 13.234375 47.609375 19.390625 \nQ 47.609375 24.75 44.3125 27.78125 \nQ 41.015625 30.8125 33.5 32.328125 \nL 27.484375 33.5 \nQ 16.453125 35.6875 11.515625 40.375 \nQ 6.59375 45.0625 6.59375 53.421875 \nQ 6.59375 63.09375 13.40625 68.65625 \nQ 20.21875 74.21875 32.171875 74.21875 \nQ 37.3125 74.21875 42.625 73.28125 \nQ 47.953125 72.359375 53.515625 70.515625 \nz\n\" id=\"DejaVuSans-83\"/>\n     <path d=\"M 9.8125 72.90625 \nL 55.90625 72.90625 \nL 55.90625 64.59375 \nL 19.671875 64.59375 \nL 19.671875 43.015625 \nL 54.390625 43.015625 \nL 54.390625 34.71875 \nL 19.671875 34.71875 \nL 19.671875 8.296875 \nL 56.78125 8.296875 \nL 56.78125 0 \nL 9.8125 0 \nz\n\" id=\"DejaVuSans-69\"/>\n     <path d=\"M 9.8125 72.90625 \nL 19.671875 72.90625 \nL 19.671875 8.296875 \nL 55.171875 8.296875 \nL 55.171875 0 \nL 9.8125 0 \nz\n\" id=\"DejaVuSans-76\"/>\n     <path d=\"M 8.6875 72.90625 \nL 18.609375 72.90625 \nL 18.609375 28.609375 \nQ 18.609375 16.890625 22.84375 11.734375 \nQ 27.09375 6.59375 36.625 6.59375 \nQ 46.09375 6.59375 50.34375 11.734375 \nQ 54.59375 16.890625 54.59375 28.609375 \nL 54.59375 72.90625 \nL 64.5 72.90625 \nL 64.5 27.390625 \nQ 64.5 13.140625 57.4375 5.859375 \nQ 50.390625 -1.421875 36.625 -1.421875 \nQ 22.796875 -1.421875 15.734375 5.859375 \nQ 8.6875 13.140625 8.6875 27.390625 \nz\n\" id=\"DejaVuSans-85\"/>\n     <path id=\"DejaVuSans-32\"/>\n     <path d=\"M 34.28125 27.484375 \nQ 23.390625 27.484375 19.1875 25 \nQ 14.984375 22.515625 14.984375 16.5 \nQ 14.984375 11.71875 18.140625 8.90625 \nQ 21.296875 6.109375 26.703125 6.109375 \nQ 34.1875 6.109375 38.703125 11.40625 \nQ 43.21875 16.703125 43.21875 25.484375 \nL 43.21875 27.484375 \nz\nM 52.203125 31.203125 \nL 52.203125 0 \nL 43.21875 0 \nL 43.21875 8.296875 \nQ 40.140625 3.328125 35.546875 0.953125 \nQ 30.953125 -1.421875 24.3125 -1.421875 \nQ 15.921875 -1.421875 10.953125 3.296875 \nQ 6 8.015625 6 15.921875 \nQ 6 25.140625 12.171875 29.828125 \nQ 18.359375 34.515625 30.609375 34.515625 \nL 43.21875 34.515625 \nL 43.21875 35.40625 \nQ 43.21875 41.609375 39.140625 45 \nQ 35.0625 48.390625 27.6875 48.390625 \nQ 23 48.390625 18.546875 47.265625 \nQ 14.109375 46.140625 10.015625 43.890625 \nL 10.015625 52.203125 \nQ 14.9375 54.109375 19.578125 55.046875 \nQ 24.21875 56 28.609375 56 \nQ 40.484375 56 46.34375 49.84375 \nQ 52.203125 43.703125 52.203125 31.203125 \nz\n\" id=\"DejaVuSans-97\"/>\n     <path d=\"M 48.78125 52.59375 \nL 48.78125 44.1875 \nQ 44.96875 46.296875 41.140625 47.34375 \nQ 37.3125 48.390625 33.40625 48.390625 \nQ 24.65625 48.390625 19.8125 42.84375 \nQ 14.984375 37.3125 14.984375 27.296875 \nQ 14.984375 17.28125 19.8125 11.734375 \nQ 24.65625 6.203125 33.40625 6.203125 \nQ 37.3125 6.203125 41.140625 7.25 \nQ 44.96875 8.296875 48.78125 10.40625 \nL 48.78125 2.09375 \nQ 45.015625 0.34375 40.984375 -0.53125 \nQ 36.96875 -1.421875 32.421875 -1.421875 \nQ 20.0625 -1.421875 12.78125 6.34375 \nQ 5.515625 14.109375 5.515625 27.296875 \nQ 5.515625 40.671875 12.859375 48.328125 \nQ 20.21875 56 33.015625 56 \nQ 37.15625 56 41.109375 55.140625 \nQ 45.0625 54.296875 48.78125 52.59375 \nz\n\" id=\"DejaVuSans-99\"/>\n     <path d=\"M 18.3125 70.21875 \nL 18.3125 54.6875 \nL 36.8125 54.6875 \nL 36.8125 47.703125 \nL 18.3125 47.703125 \nL 18.3125 18.015625 \nQ 18.3125 11.328125 20.140625 9.421875 \nQ 21.96875 7.515625 27.59375 7.515625 \nL 36.8125 7.515625 \nL 36.8125 0 \nL 27.59375 0 \nQ 17.1875 0 13.234375 3.875 \nQ 9.28125 7.765625 9.28125 18.015625 \nL 9.28125 47.703125 \nL 2.6875 47.703125 \nL 2.6875 54.6875 \nL 9.28125 54.6875 \nL 9.28125 70.21875 \nz\n\" id=\"DejaVuSans-116\"/>\n     <path d=\"M 9.421875 54.6875 \nL 18.40625 54.6875 \nL 18.40625 0 \nL 9.421875 0 \nz\nM 9.421875 75.984375 \nL 18.40625 75.984375 \nL 18.40625 64.59375 \nL 9.421875 64.59375 \nz\n\" id=\"DejaVuSans-105\"/>\n     <path d=\"M 2.984375 54.6875 \nL 12.5 54.6875 \nL 29.59375 8.796875 \nL 46.6875 54.6875 \nL 56.203125 54.6875 \nL 35.6875 0 \nL 23.484375 0 \nz\n\" id=\"DejaVuSans-118\"/>\n     <path d=\"M 30.609375 48.390625 \nQ 23.390625 48.390625 19.1875 42.75 \nQ 14.984375 37.109375 14.984375 27.296875 \nQ 14.984375 17.484375 19.15625 11.84375 \nQ 23.34375 6.203125 30.609375 6.203125 \nQ 37.796875 6.203125 41.984375 11.859375 \nQ 46.1875 17.53125 46.1875 27.296875 \nQ 46.1875 37.015625 41.984375 42.703125 \nQ 37.796875 48.390625 30.609375 48.390625 \nz\nM 30.609375 56 \nQ 42.328125 56 49.015625 48.375 \nQ 55.71875 40.765625 55.71875 27.296875 \nQ 55.71875 13.875 49.015625 6.21875 \nQ 42.328125 -1.421875 30.609375 -1.421875 \nQ 18.84375 -1.421875 12.171875 6.21875 \nQ 5.515625 13.875 5.515625 27.296875 \nQ 5.515625 40.765625 12.171875 48.375 \nQ 18.84375 56 30.609375 56 \nz\n\" id=\"DejaVuSans-111\"/>\n     <path d=\"M 54.890625 33.015625 \nL 54.890625 0 \nL 45.90625 0 \nL 45.90625 32.71875 \nQ 45.90625 40.484375 42.875 44.328125 \nQ 39.84375 48.1875 33.796875 48.1875 \nQ 26.515625 48.1875 22.3125 43.546875 \nQ 18.109375 38.921875 18.109375 30.90625 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 54.6875 \nL 18.109375 54.6875 \nL 18.109375 46.1875 \nQ 21.34375 51.125 25.703125 53.5625 \nQ 30.078125 56 35.796875 56 \nQ 45.21875 56 50.046875 50.171875 \nQ 54.890625 44.34375 54.890625 33.015625 \nz\n\" id=\"DejaVuSans-110\"/>\n     <path d=\"M 37.109375 75.984375 \nL 37.109375 68.5 \nL 28.515625 68.5 \nQ 23.6875 68.5 21.796875 66.546875 \nQ 19.921875 64.59375 19.921875 59.515625 \nL 19.921875 54.6875 \nL 34.71875 54.6875 \nL 34.71875 47.703125 \nL 19.921875 47.703125 \nL 19.921875 0 \nL 10.890625 0 \nL 10.890625 47.703125 \nL 2.296875 47.703125 \nL 2.296875 54.6875 \nL 10.890625 54.6875 \nL 10.890625 58.5 \nQ 10.890625 67.625 15.140625 71.796875 \nQ 19.390625 75.984375 28.609375 75.984375 \nz\n\" id=\"DejaVuSans-102\"/>\n     <path d=\"M 8.5 21.578125 \nL 8.5 54.6875 \nL 17.484375 54.6875 \nL 17.484375 21.921875 \nQ 17.484375 14.15625 20.5 10.265625 \nQ 23.53125 6.390625 29.59375 6.390625 \nQ 36.859375 6.390625 41.078125 11.03125 \nQ 45.3125 15.671875 45.3125 23.6875 \nL 45.3125 54.6875 \nL 54.296875 54.6875 \nL 54.296875 0 \nL 45.3125 0 \nL 45.3125 8.40625 \nQ 42.046875 3.421875 37.71875 1 \nQ 33.40625 -1.421875 27.6875 -1.421875 \nQ 18.265625 -1.421875 13.375 4.4375 \nQ 8.5 10.296875 8.5 21.578125 \nz\nM 31.109375 56 \nz\n\" id=\"DejaVuSans-117\"/>\n    </defs>\n    <g transform=\"translate(139.18125 17.837812)scale(0.14 -0.14)\">\n     <use xlink:href=\"#DejaVuSans-83\"/>\n     <use x=\"63.476562\" xlink:href=\"#DejaVuSans-69\"/>\n     <use x=\"126.660156\" xlink:href=\"#DejaVuSans-76\"/>\n     <use x=\"182.294922\" xlink:href=\"#DejaVuSans-85\"/>\n     <use x=\"255.488281\" xlink:href=\"#DejaVuSans-32\"/>\n     <use x=\"287.275391\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"348.554688\" xlink:href=\"#DejaVuSans-99\"/>\n     <use x=\"403.535156\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"442.744141\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"470.527344\" xlink:href=\"#DejaVuSans-118\"/>\n     <use x=\"529.707031\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"590.986328\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"630.195312\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"657.978516\" xlink:href=\"#DejaVuSans-111\"/>\n     <use x=\"719.160156\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"782.539062\" xlink:href=\"#DejaVuSans-32\"/>\n     <use x=\"814.326172\" xlink:href=\"#DejaVuSans-102\"/>\n     <use x=\"849.53125\" xlink:href=\"#DejaVuSans-117\"/>\n     <use x=\"912.910156\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"976.289062\" xlink:href=\"#DejaVuSans-99\"/>\n     <use x=\"1031.269531\" xlink:href=\"#DejaVuSans-116\"/>\n     <use x=\"1070.478516\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"1098.261719\" xlink:href=\"#DejaVuSans-111\"/>\n     <use x=\"1159.443359\" xlink:href=\"#DejaVuSans-110\"/>\n    </g>\n   </g>\n  </g>\n </g>\n <defs>\n  <clipPath id=\"p3879959fc6\">\n   <rect height=\"229.4\" width=\"385.775\" x=\"31.890625\" y=\"23.837812\"/>\n  </clipPath>\n </defs>\n</svg>\n",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "source": [
    "plt.plot(z, selu(z), \"b-\", linewidth=2)\n",
    "plt.plot([-5, 5], [0, 0], 'k-')\n",
    "plt.plot([-5, 5], [-1.758, -1.758], 'k--')\n",
    "plt.plot([0, 0], [-2.2, 3.2], 'k-')\n",
    "plt.grid(True)\n",
    "plt.title(\"SELU activation function\", fontsize=14)\n",
    "plt.axis([-5, 5, -2.2, 3.2])\n",
    "\n",
    "save_fig(\"selu_plot\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Layer 0: mean -0.00, std deviation 1.00\nLayer 100: mean 0.02, std deviation 0.96\nLayer 200: mean 0.01, std deviation 0.90\nLayer 300: mean -0.02, std deviation 0.92\nLayer 400: mean 0.05, std deviation 0.89\nLayer 500: mean 0.01, std deviation 0.93\nLayer 600: mean 0.02, std deviation 0.92\nLayer 700: mean -0.02, std deviation 0.90\nLayer 800: mean 0.05, std deviation 0.83\nLayer 900: mean 0.02, std deviation 1.00\n"
    }
   ],
   "source": [
    "np.random.seed(42)\n",
    "Z = np.random.normal(size=(500, 100)) # standardized inputs\n",
    "for layer in range(1000):\n",
    "    W = np.random.normal(size=(100, 100), scale=np.sqrt(1 / 100)) # LeCun initialization\n",
    "    Z = selu(np.dot(Z, W))\n",
    "    means = np.mean(Z, axis=0).mean()\n",
    "    stds = np.std(Z, axis=0).mean()\n",
    "    if layer % 100 == 0:\n",
    "        print(\"Layer {}: mean {:.2f}, std deviation {:.2f}\".format(layer, means, stds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "<tensorflow.python.keras.layers.core.Dense at 0x7f8d004ef710>"
     },
     "metadata": {},
     "execution_count": 83
    }
   ],
   "source": [
    "keras.layers.Dense(\n",
    "    10, \n",
    "    activation=\"selu\",\n",
    "    kernel_initializer=\"lecun_normal\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## create a neural net for Fashion MNIST with 100 hidden layers, using the SELU activation function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.random.seed(42)\n",
    "tf.random.set_seed(42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.Sequential()\n",
    "model.add(keras.layers.Flatten(input_shape=[28, 28]))\n",
    "model.add(keras.layers.Dense(300, activation=\"selu\",\n",
    "                             kernel_initializer=\"lecun_normal\"))\n",
    "for layer in range(99):\n",
    "    model.add(keras.layers.Dense(100, activation=\"selu\",\n",
    "                                 kernel_initializer=\"lecun_normal\"))\n",
    "model.add(keras.layers.Dense(10, activation=\"softmax\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
    "              optimizer=keras.optimizers.SGD(lr=1e-3),\n",
    "              metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "pixel_means = X_train.mean(axis=0, keepdims=True)\n",
    "pixel_stds = X_train.std(axis=0, keepdims=True)\n",
    "X_train_scaled = (X_train - pixel_means) / pixel_stds\n",
    "X_valid_scaled = (X_valid - pixel_means) / pixel_stds\n",
    "X_test_scaled = (X_test - pixel_means) / pixel_stds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/5\n1719/1719 [==============================] - 14s 8ms/step - loss: 1.2829 - accuracy: 0.4983 - val_loss: 0.9338 - val_accuracy: 0.6178\nEpoch 2/5\n1719/1719 [==============================] - 13s 7ms/step - loss: 0.7986 - accuracy: 0.6979 - val_loss: 0.6783 - val_accuracy: 0.7522\nEpoch 3/5\n1719/1719 [==============================] - 12s 7ms/step - loss: 0.6677 - accuracy: 0.7569 - val_loss: 0.6068 - val_accuracy: 0.7758\nEpoch 4/5\n1719/1719 [==============================] - 13s 7ms/step - loss: 0.5748 - accuracy: 0.7895 - val_loss: 0.5438 - val_accuracy: 0.7980\nEpoch 5/5\n1719/1719 [==============================] - 12s 7ms/step - loss: 0.5289 - accuracy: 0.8073 - val_loss: 0.5301 - val_accuracy: 0.8112\n"
    }
   ],
   "source": [
    "history = model.fit(\n",
    "    X_train_scaled, \n",
    "    y_train, \n",
    "    epochs=5,\n",
    "    validation_data=(X_valid_scaled, y_valid))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Use ReLU, instead"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.random.seed(42)\n",
    "tf.random.set_seed(42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.add(keras.layers.Flatten(input_shape=[28, 28]))\n",
    "model.add(keras.layers.Dense(300, activation=\"relu\", kernel_initializer=\"he_normal\"))\n",
    "for layer in range(99):\n",
    "    model.add(keras.layers.Dense(100, activation=\"relu\", kernel_initializer=\"he_normal\"))\n",
    "model.add(keras.layers.Dense(10, activation=\"softmax\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
    "              optimizer=keras.optimizers.SGD(lr=1e-3),\n",
    "              metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/5\n1719/1719 [==============================] - 21s 12ms/step - loss: 1.9862 - accuracy: 0.2743 - val_loss: 1.8718 - val_accuracy: 0.2888\nEpoch 2/5\n1719/1719 [==============================] - 20s 12ms/step - loss: 1.7874 - accuracy: 0.2920 - val_loss: 1.3735 - val_accuracy: 0.3722\nEpoch 3/5\n1719/1719 [==============================] - 20s 12ms/step - loss: 1.7316 - accuracy: 0.2331 - val_loss: 1.7847 - val_accuracy: 0.1858\nEpoch 4/5\n1719/1719 [==============================] - 20s 12ms/step - loss: 1.8158 - accuracy: 0.1997 - val_loss: 1.8437 - val_accuracy: 0.2078\nEpoch 5/5\n1719/1719 [==============================] - 20s 12ms/step - loss: 1.7793 - accuracy: 0.2016 - val_loss: 1.7499 - val_accuracy: 0.2004\n"
    }
   ],
   "source": [
    "history = model.fit(X_train_scaled, y_train, epochs=5,\n",
    "                    validation_data=(X_valid_scaled, y_valid))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## batch normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.Sequential([\n",
    "    keras.layers.Flatten(input_shape=[28, 28]),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.Dense(300, activation=\"relu\"),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.Dense(100, activation=\"relu\"),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.Dense(10, activation=\"softmax\")\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Model: \"sequential_11\"\n_________________________________________________________________\nLayer (type)                 Output Shape              Param #   \n=================================================================\nflatten_12 (Flatten)         (None, 784)               0         \n_________________________________________________________________\nbatch_normalization_6 (Batch (None, 784)               3136      \n_________________________________________________________________\ndense_443 (Dense)            (None, 300)               235500    \n_________________________________________________________________\nbatch_normalization_7 (Batch (None, 300)               1200      \n_________________________________________________________________\ndense_444 (Dense)            (None, 100)               30100     \n_________________________________________________________________\nbatch_normalization_8 (Batch (None, 100)               400       \n_________________________________________________________________\ndense_445 (Dense)            (None, 10)                1010      \n=================================================================\nTotal params: 271,346\nTrainable params: 268,978\nNon-trainable params: 2,368\n_________________________________________________________________\n"
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "[('batch_normalization_6/gamma:0', True),\n ('batch_normalization_6/beta:0', True),\n ('batch_normalization_6/moving_mean:0', False),\n ('batch_normalization_6/moving_variance:0', False)]"
     },
     "metadata": {},
     "execution_count": 102
    }
   ],
   "source": [
    "bn1 = model.layers[1]\n",
    "[(var.name, var.trainable) for var in bn1.variables]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "[<tf.Operation 'cond_2/Identity' type=Identity>,\n <tf.Operation 'cond_3/Identity' type=Identity>]"
     },
     "metadata": {},
     "execution_count": 103
    }
   ],
   "source": [
    "bn1.updates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
    "              optimizer=keras.optimizers.SGD(lr=1e-3),\n",
    "              metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.8750 - accuracy: 0.7124 - val_loss: 0.5525 - val_accuracy: 0.8230\nEpoch 2/10\n1719/1719 [==============================] - 4s 2ms/step - loss: 0.5753 - accuracy: 0.8029 - val_loss: 0.4724 - val_accuracy: 0.8472\nEpoch 3/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.5189 - accuracy: 0.8206 - val_loss: 0.4375 - val_accuracy: 0.8554\nEpoch 4/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.4827 - accuracy: 0.8323 - val_loss: 0.4151 - val_accuracy: 0.8594\nEpoch 5/10\n1719/1719 [==============================] - 4s 2ms/step - loss: 0.4565 - accuracy: 0.8407 - val_loss: 0.3997 - val_accuracy: 0.8636\nEpoch 6/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.4397 - accuracy: 0.8474 - val_loss: 0.3867 - val_accuracy: 0.8694\nEpoch 7/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.4242 - accuracy: 0.8513 - val_loss: 0.3763 - val_accuracy: 0.8708\nEpoch 8/10\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.4143 - accuracy: 0.8541 - val_loss: 0.3712 - val_accuracy: 0.8738\nEpoch 9/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.4023 - accuracy: 0.8580 - val_loss: 0.3630 - val_accuracy: 0.8748\nEpoch 10/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3914 - accuracy: 0.8626 - val_loss: 0.3571 - val_accuracy: 0.8758\n"
    }
   ],
   "source": [
    "history = model.fit(X_train, y_train, epochs=10,\n",
    "                    validation_data=(X_valid, y_valid))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.Sequential([\n",
    "    keras.layers.Flatten(input_shape=[28, 28]),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.Dense(300, use_bias=False),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.Activation(\"relu\"),\n",
    "    keras.layers.Dense(100, use_bias=False),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.Activation(\"relu\"),\n",
    "    keras.layers.Dense(10, activation=\"softmax\")\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
    "              optimizer=keras.optimizers.SGD(lr=1e-3),\n",
    "              metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/10\n1719/1719 [==============================] - 5s 3ms/step - loss: 1.0317 - accuracy: 0.6757 - val_loss: 0.6767 - val_accuracy: 0.7812\nEpoch 2/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.6790 - accuracy: 0.7793 - val_loss: 0.5566 - val_accuracy: 0.8180\nEpoch 3/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.5960 - accuracy: 0.8037 - val_loss: 0.5007 - val_accuracy: 0.8362\nEpoch 4/10\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.5447 - accuracy: 0.8192 - val_loss: 0.4666 - val_accuracy: 0.8448\nEpoch 5/10\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.5109 - accuracy: 0.8277 - val_loss: 0.4434 - val_accuracy: 0.8536\nEpoch 6/10\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.4898 - accuracy: 0.8337 - val_loss: 0.4263 - val_accuracy: 0.8546\nEpoch 7/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.4712 - accuracy: 0.8395 - val_loss: 0.4130 - val_accuracy: 0.8572\nEpoch 8/10\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.4560 - accuracy: 0.8440 - val_loss: 0.4034 - val_accuracy: 0.8608\nEpoch 9/10\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.4441 - accuracy: 0.8475 - val_loss: 0.3943 - val_accuracy: 0.8642\nEpoch 10/10\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.4332 - accuracy: 0.8504 - val_loss: 0.3874 - val_accuracy: 0.8662\n"
    }
   ],
   "source": [
    "history = model.fit(X_train, y_train, epochs=10,\n",
    "                    validation_data=(X_valid, y_valid))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## gradient clipping"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## reusing pretrained layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "def split_dataset(X, y):\n",
    "    y_5_or_6 = (y == 5) | (y == 6) # sandals or shirts\n",
    "    y_A = y[~y_5_or_6]\n",
    "    y_A[y_A > 6] -= 2 # class indices 7, 8, 9 should be moved to 5, 6, 7\n",
    "    y_B = (y[y_5_or_6] == 6).astype(np.float32) # binary classification task: is it a shirt (class 6)?\n",
    "    return ((X[~y_5_or_6], y_A),\n",
    "            (X[y_5_or_6], y_B))\n",
    "\n",
    "(X_train_A, y_train_A), (X_train_B, y_train_B) = split_dataset(X_train, y_train)\n",
    "(X_valid_A, y_valid_A), (X_valid_B, y_valid_B) = split_dataset(X_valid, y_valid)\n",
    "(X_test_A, y_test_A), (X_test_B, y_test_B) = split_dataset(X_test, y_test)\n",
    "X_train_B = X_train_B[:200]\n",
    "y_train_B = y_train_B[:200]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "(43986, 28, 28)"
     },
     "metadata": {},
     "execution_count": 110
    }
   ],
   "source": [
    "X_train_A.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "(200, 28, 28)"
     },
     "metadata": {},
     "execution_count": 111
    }
   ],
   "source": [
    "X_train_B.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "array([4, 0, 5, 7, 7, 7, 4, 4, 3, 4, 0, 1, 6, 3, 4, 3, 2, 6, 5, 3, 4, 5,\n       1, 3, 4, 2, 0, 6, 7, 1], dtype=uint8)"
     },
     "metadata": {},
     "execution_count": 112
    }
   ],
   "source": [
    "y_train_A[:30]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "array([1., 1., 0., 0., 0., 0., 1., 1., 1., 0., 0., 1., 1., 0., 0., 0., 0.,\n       0., 0., 1., 1., 0., 0., 1., 1., 0., 1., 1., 1., 1.], dtype=float32)"
     },
     "metadata": {},
     "execution_count": 113
    }
   ],
   "source": [
    "y_train_B[:30]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [],
   "source": [
    "tf.random.set_seed(42)\n",
    "np.random.seed(42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_A = keras.models.Sequential()\n",
    "model_A.add(keras.layers.Flatten(input_shape=[28, 28]))\n",
    "for n_hidden in (300, 100, 50, 50, 50):\n",
    "    model_A.add(keras.layers.Dense(n_hidden, activation='selu'))\n",
    "model_A.add(keras.layers.Dense(8, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_A.compile(\n",
    "    loss='sparse_categorical_crossentropy', \n",
    "    optimizer=keras.optimizers.SGD(lr=1e-3), \n",
    "    metrics=['accuracy']\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.5926 - accuracy: 0.8103 - val_loss: 0.3892 - val_accuracy: 0.8675\nEpoch 2/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.3523 - accuracy: 0.8788 - val_loss: 0.3290 - val_accuracy: 0.8819\nEpoch 3/20\n1375/1375 [==============================] - 3s 2ms/step - loss: 0.3170 - accuracy: 0.8896 - val_loss: 0.3013 - val_accuracy: 0.8989\nEpoch 4/20\n1375/1375 [==============================] - 3s 2ms/step - loss: 0.2973 - accuracy: 0.8975 - val_loss: 0.2894 - val_accuracy: 0.9023\nEpoch 5/20\n1375/1375 [==============================] - 3s 2ms/step - loss: 0.2835 - accuracy: 0.9021 - val_loss: 0.2774 - val_accuracy: 0.9068\nEpoch 6/20\n1375/1375 [==============================] - 3s 2ms/step - loss: 0.2729 - accuracy: 0.9061 - val_loss: 0.2731 - val_accuracy: 0.9066\nEpoch 7/20\n1375/1375 [==============================] - 3s 2ms/step - loss: 0.2641 - accuracy: 0.9091 - val_loss: 0.2718 - val_accuracy: 0.9091\nEpoch 8/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2573 - accuracy: 0.9126 - val_loss: 0.2588 - val_accuracy: 0.9141\nEpoch 9/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2518 - accuracy: 0.9134 - val_loss: 0.2562 - val_accuracy: 0.9143\nEpoch 10/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2469 - accuracy: 0.9153 - val_loss: 0.2539 - val_accuracy: 0.9165\nEpoch 11/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2422 - accuracy: 0.9175 - val_loss: 0.2495 - val_accuracy: 0.9150\nEpoch 12/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2382 - accuracy: 0.9186 - val_loss: 0.2510 - val_accuracy: 0.9136\nEpoch 13/20\n1375/1375 [==============================] - 3s 2ms/step - loss: 0.2350 - accuracy: 0.9198 - val_loss: 0.2443 - val_accuracy: 0.9158\nEpoch 14/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2315 - accuracy: 0.9211 - val_loss: 0.2412 - val_accuracy: 0.9178\nEpoch 15/20\n1375/1375 [==============================] - 3s 2ms/step - loss: 0.2287 - accuracy: 0.9214 - val_loss: 0.2445 - val_accuracy: 0.9193\nEpoch 16/20\n1375/1375 [==============================] - 3s 2ms/step - loss: 0.2254 - accuracy: 0.9223 - val_loss: 0.2382 - val_accuracy: 0.9193\nEpoch 17/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2230 - accuracy: 0.9231 - val_loss: 0.2405 - val_accuracy: 0.9180\nEpoch 18/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2201 - accuracy: 0.9246 - val_loss: 0.2420 - val_accuracy: 0.9150\nEpoch 19/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2178 - accuracy: 0.9254 - val_loss: 0.2329 - val_accuracy: 0.9203\nEpoch 20/20\n1375/1375 [==============================] - 2s 2ms/step - loss: 0.2155 - accuracy: 0.9260 - val_loss: 0.2332 - val_accuracy: 0.9205\n"
    }
   ],
   "source": [
    "history = model_A.fit(\n",
    "    X_train_A, \n",
    "    y_train_A, \n",
    "    epochs=20,\n",
    "    validation_data=(X_valid_A, y_valid_A)\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_A.save('my_model_A.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_B = keras.models.Sequential()\n",
    "model_B.add(keras.layers.Flatten(input_shape=[28, 28]))\n",
    "for n_hidden in (300, 100, 50, 50, 50):\n",
    "    model_B.add(keras.layers.Dense(n_hidden, activation='selu'))\n",
    "model_B.add(keras.layers.Dense(1, activation='sigmoid'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_B.compile(\n",
    "    loss='binary_crossentropy', \n",
    "    optimizer=keras.optimizers.SGD(lr=1e-3), \n",
    "    metrics=['accuracy']\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/20\n7/7 [==============================] - 0s 22ms/step - loss: 0.9573 - accuracy: 0.4650 - val_loss: 0.6314 - val_accuracy: 0.6004\nEpoch 2/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.5692 - accuracy: 0.7450 - val_loss: 0.4784 - val_accuracy: 0.8529\nEpoch 3/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.4503 - accuracy: 0.8650 - val_loss: 0.4102 - val_accuracy: 0.8945\nEpoch 4/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.3879 - accuracy: 0.8950 - val_loss: 0.3647 - val_accuracy: 0.9178\nEpoch 5/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.3435 - accuracy: 0.9250 - val_loss: 0.3300 - val_accuracy: 0.9320\nEpoch 6/20\n7/7 [==============================] - 0s 11ms/step - loss: 0.3081 - accuracy: 0.9300 - val_loss: 0.3019 - val_accuracy: 0.9402\nEpoch 7/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.2800 - accuracy: 0.9350 - val_loss: 0.2804 - val_accuracy: 0.9422\nEpoch 8/20\n7/7 [==============================] - 0s 12ms/step - loss: 0.2564 - accuracy: 0.9450 - val_loss: 0.2606 - val_accuracy: 0.9473\nEpoch 9/20\n7/7 [==============================] - 0s 9ms/step - loss: 0.2362 - accuracy: 0.9550 - val_loss: 0.2428 - val_accuracy: 0.9523\nEpoch 10/20\n7/7 [==============================] - 0s 12ms/step - loss: 0.2188 - accuracy: 0.9600 - val_loss: 0.2281 - val_accuracy: 0.9544\nEpoch 11/20\n7/7 [==============================] - 0s 11ms/step - loss: 0.2036 - accuracy: 0.9700 - val_loss: 0.2150 - val_accuracy: 0.9584\nEpoch 12/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.1898 - accuracy: 0.9700 - val_loss: 0.2036 - val_accuracy: 0.9584\nEpoch 13/20\n7/7 [==============================] - 0s 9ms/step - loss: 0.1773 - accuracy: 0.9750 - val_loss: 0.1931 - val_accuracy: 0.9615\nEpoch 14/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.1668 - accuracy: 0.9800 - val_loss: 0.1838 - val_accuracy: 0.9635\nEpoch 15/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.1570 - accuracy: 0.9900 - val_loss: 0.1746 - val_accuracy: 0.9686\nEpoch 16/20\n7/7 [==============================] - 0s 11ms/step - loss: 0.1481 - accuracy: 0.9900 - val_loss: 0.1674 - val_accuracy: 0.9686\nEpoch 17/20\n7/7 [==============================] - 0s 9ms/step - loss: 0.1406 - accuracy: 0.9900 - val_loss: 0.1604 - val_accuracy: 0.9706\nEpoch 18/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.1334 - accuracy: 0.9900 - val_loss: 0.1539 - val_accuracy: 0.9706\nEpoch 19/20\n7/7 [==============================] - 0s 10ms/step - loss: 0.1268 - accuracy: 0.9900 - val_loss: 0.1482 - val_accuracy: 0.9716\nEpoch 20/20\n7/7 [==============================] - 0s 9ms/step - loss: 0.1208 - accuracy: 0.9900 - val_loss: 0.1431 - val_accuracy: 0.9716\n"
    }
   ],
   "source": [
    "history = model_B.fit(\n",
    "    X_train_B, \n",
    "    y_train_B, \n",
    "    epochs=20,\n",
    "    validation_data=(X_valid_B, y_valid_B)\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Model: \"sequential_12\"\n_________________________________________________________________\nLayer (type)                 Output Shape              Param #   \n=================================================================\nflatten_13 (Flatten)         (None, 784)               0         \n_________________________________________________________________\nbatch_normalization_9 (Batch (None, 784)               3136      \n_________________________________________________________________\ndense_446 (Dense)            (None, 300)               235200    \n_________________________________________________________________\nbatch_normalization_10 (Batc (None, 300)               1200      \n_________________________________________________________________\nactivation_2 (Activation)    (None, 300)               0         \n_________________________________________________________________\ndense_447 (Dense)            (None, 100)               30000     \n_________________________________________________________________\nbatch_normalization_11 (Batc (None, 100)               400       \n_________________________________________________________________\nactivation_3 (Activation)    (None, 100)               0         \n_________________________________________________________________\ndense_448 (Dense)            (None, 10)                1010      \n=================================================================\nTotal params: 270,946\nTrainable params: 268,578\nNon-trainable params: 2,368\n_________________________________________________________________\n"
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_A = keras.models.load_model(\"my_model_A.h5\")\n",
    "model_B_on_A = keras.models.Sequential(model_A.layers[:-1])\n",
    "model_B_on_A.add(keras.layers.Dense(1, activation=\"sigmoid\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_A_clone = keras.models.clone_model(model_A)\n",
    "model_A_clone.set_weights(model_A.get_weights())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [],
   "source": [
    "for layer in model_B_on_A.layers[:-1]:\n",
    "    layer.trainable = False\n",
    "\n",
    "model_B_on_A.compile(\n",
    "    loss=\"binary_crossentropy\",\n",
    "    optimizer=keras.optimizers.SGD(lr=1e-3),\n",
    "    metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/4\n7/7 [==============================] - 0s 31ms/step - loss: 0.5775 - accuracy: 0.6500 - val_loss: 0.5817 - val_accuracy: 0.6400\nEpoch 2/4\n7/7 [==============================] - 0s 12ms/step - loss: 0.5411 - accuracy: 0.6700 - val_loss: 0.5444 - val_accuracy: 0.6815\nEpoch 3/4\n7/7 [==============================] - 0s 11ms/step - loss: 0.5045 - accuracy: 0.7300 - val_loss: 0.5125 - val_accuracy: 0.7099\nEpoch 4/4\n7/7 [==============================] - 0s 23ms/step - loss: 0.4731 - accuracy: 0.7500 - val_loss: 0.4839 - val_accuracy: 0.7363\n"
    }
   ],
   "source": [
    "history = model_B_on_A.fit(\n",
    "    X_train_B, \n",
    "    y_train_B, \n",
    "    epochs=4,\n",
    "    validation_data=(X_valid_B, y_valid_B))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "Epoch 1/16\n7/7 [==============================] - 0s 26ms/step - loss: 0.3950 - accuracy: 0.8200 - val_loss: 0.3452 - val_accuracy: 0.8671\nEpoch 2/16\n7/7 [==============================] - 0s 15ms/step - loss: 0.2793 - accuracy: 0.9350 - val_loss: 0.2599 - val_accuracy: 0.9290\nEpoch 3/16\n7/7 [==============================] - 0s 17ms/step - loss: 0.2080 - accuracy: 0.9650 - val_loss: 0.2108 - val_accuracy: 0.9544\nEpoch 4/16\n7/7 [==============================] - 0s 17ms/step - loss: 0.1668 - accuracy: 0.9800 - val_loss: 0.1789 - val_accuracy: 0.9696\nEpoch 5/16\n7/7 [==============================] - 0s 23ms/step - loss: 0.1395 - accuracy: 0.9800 - val_loss: 0.1560 - val_accuracy: 0.9757\nEpoch 6/16\n7/7 [==============================] - 0s 14ms/step - loss: 0.1196 - accuracy: 0.9950 - val_loss: 0.1392 - val_accuracy: 0.9797\nEpoch 7/16\n7/7 [==============================] - 0s 13ms/step - loss: 0.1049 - accuracy: 0.9950 - val_loss: 0.1266 - val_accuracy: 0.9838\nEpoch 8/16\n7/7 [==============================] - 0s 17ms/step - loss: 0.0937 - accuracy: 0.9950 - val_loss: 0.1163 - val_accuracy: 0.9858\nEpoch 9/16\n7/7 [==============================] - 0s 14ms/step - loss: 0.0847 - accuracy: 1.0000 - val_loss: 0.1066 - val_accuracy: 0.9888\nEpoch 10/16\n7/7 [==============================] - 0s 24ms/step - loss: 0.0762 - accuracy: 1.0000 - val_loss: 0.1000 - val_accuracy: 0.9899\nEpoch 11/16\n7/7 [==============================] - 0s 16ms/step - loss: 0.0704 - accuracy: 1.0000 - val_loss: 0.0940 - val_accuracy: 0.9899\nEpoch 12/16\n7/7 [==============================] - 0s 13ms/step - loss: 0.0649 - accuracy: 1.0000 - val_loss: 0.0888 - val_accuracy: 0.9899\nEpoch 13/16\n7/7 [==============================] - 0s 16ms/step - loss: 0.0602 - accuracy: 1.0000 - val_loss: 0.0839 - val_accuracy: 0.9899\nEpoch 14/16\n7/7 [==============================] - 0s 12ms/step - loss: 0.0559 - accuracy: 1.0000 - val_loss: 0.0802 - val_accuracy: 0.9899\nEpoch 15/16\n7/7 [==============================] - 0s 12ms/step - loss: 0.0525 - accuracy: 1.0000 - val_loss: 0.0769 - val_accuracy: 0.9899\nEpoch 16/16\n7/7 [==============================] - 0s 13ms/step - loss: 0.0496 - accuracy: 1.0000 - val_loss: 0.0739 - val_accuracy: 0.9899\n"
    }
   ],
   "source": [
    "for layer in model_B_on_A.layers[:-1]:\n",
    "    layer.trainable = True\n",
    "\n",
    "model_B_on_A.compile(loss=\"binary_crossentropy\",\n",
    "                     optimizer=keras.optimizers.SGD(lr=1e-3),\n",
    "                     metrics=[\"accuracy\"])\n",
    "                     \n",
    "history = model_B_on_A.fit(X_train_B, y_train_B, epochs=16,\n",
    "                           validation_data=(X_valid_B, y_valid_B))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "63/63 [==============================] - 0s 2ms/step - loss: 0.1408 - accuracy: 0.9705\n"
    },
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "[0.1408407837152481, 0.9704999923706055]"
     },
     "metadata": {},
     "execution_count": 128
    }
   ],
   "source": [
    "model_B.evaluate(X_test_B, y_test_B)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "63/63 [==============================] - 0s 2ms/step - loss: 0.0683 - accuracy: 0.9930\n"
    },
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "[0.06825262308120728, 0.9929999709129333]"
     },
     "metadata": {},
     "execution_count": 129
    }
   ],
   "source": [
    "model_B_on_A.evaluate(X_test_B, y_test_B)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "4.066666666666663"
     },
     "metadata": {},
     "execution_count": 130
    }
   ],
   "source": [
    "(100 - 96.95) / (100 - 99.25)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## unsupervized pretraining"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## faster optimizers"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## learning rate scheduling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.5-final"
  },
  "orig_nbformat": 2,
  "kernelspec": {
   "name": "python37564bit42731fe29a1c4072b88558cb14c14b00",
   "display_name": "Python 3.7.5 64-bit"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}