Classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Lab 2C: Classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "import torchaudio\n",
    "import sys\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import IPython.display as ipd\n",
    "\n",
    "import os\n",
    "import numpy as np\n",
    "device = \"cpu\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "A classification is a partition of reality. Given realizations $x$ of some object we ascribe to each of them a unique label $y_c$ that marks them as members of class $c$. In a classification task we want to train an artificial intelligence that when presented with an input $x$ predicts a class label $\\hat{y}$ that matches the label $y_c$ ascribed by the real world.\n",
    "\n",
    "To model classification as empirical risk minimization (ERM) we just need to define a proper loss. The easiest is to define a hit loss that takes the value $\\ell(y, \\hat{y}) = 1$ when $\\hat{y} \\neq y$ and the value $\\ell(y, \\hat{y}) = 0$ when $\\hat{y} = y$. To write this loss formally we define the indicator function $I ( y = y_c )$ which takes the values \n",
    "\n",
    "$$\n",
    "   I ( y = y_c ) = 1 \\text{~when~} y = y_c, \\quad \\\\\n",
    "   I ( y = y_c ) = 0 \\text{~when~} y \\neq y_c.\n",
    "$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The previous loss is a reasonable choice but it is not differentiable with respect to the score variable $\\hat{s}$. This is a drawback for implementing gradient descent. To solve this problem we define the soft maximum function \n",
    "\n",
    "$$\n",
    "   \\text{softmax}(\\hat{s})\n",
    "      = \\log \\sum_{d=1}^{C} e^{\\hat{s}_d}.\n",
    "$$\n",
    "\n",
    "The soft maximum is a differentiable approximation of the maximum in the sense that $\\text{softmax}(\\hat{s}) \\approx \\max_d \\hat{s}_d$. We can therefore replace the maximum $\\max_d \\hat{s}_d$ in \\eqref{eqn_max_loss} by the soft maximum function to define the loss\n",
    "\n",
    "$$\n",
    "\n",
    "\\ell(y, \\hat{s}) \n",
    "      = \\sum_{c=1}^{C} \n",
    "         I [ y = y_c ] \n",
    "            \\times \n",
    "               \\Big[ \\log \\sum_{d=1}^{C} e^{\\hat{s}_d}  -  \\hat{s}_c \\Big].\n",
    "$$\n",
    "\n",
    "This is the cross entropy loss. Its use in classification tasks is standard."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loading and exploring the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_audios = torch.load('train_audios_3.pt')\n",
    "train_labels = torch.load('train_labels_3.pt')\n",
    "test_audios = torch.load('test_audios_3.pt')\n",
    "test_labels = torch.load('test_labels_3.pt')\n",
    "sample_rate = 8000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "sample, label = train_audios[0], train_labels[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of waveform: torch.Size([1, 8000])\n",
      "Sample rate of waveform: 8000\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f702bdc2740>]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAi8AAAGdCAYAAADaPpOnAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/P9b71AAAACXBIWXMAAA9hAAAPYQGoP6dpAABOlUlEQVR4nO3deXhU5aE/8O+ZmcxMQnaykZAQFiUgq0Ri3KupoNTWpb1oUREpvVap2Fgr1Aparw2t1p9LqVz33gqCbd0XFAOoIBJZgqxBEAhbEkJIJuskM/P+/ghM5pzJTGYmc2bmJN9PnzxPzpztPYnN+fKukhBCgIiIiEgjdOEuABEREZE/GF6IiIhIUxheiIiISFMYXoiIiEhTGF6IiIhIUxheiIiISFMYXoiIiEhTGF6IiIhIUwzhLkCwORwOHD9+HHFxcZAkKdzFISIiIh8IIdDY2IjMzEzodN7rVvpceDl+/Diys7PDXQwiIiIKwJEjRzB48GCvx/S58BIXFweg8+Hj4+PDXBoiIiLyhcViQXZ2tvM97k2fCy9nm4ri4+MZXoiIiDTGly4fIemwu2TJEuTm5sJsNqOgoABlZWU+nbdixQpIkoTrr79e3QISERGRZqgeXlauXIni4mIsWrQIW7duxfjx4zFlyhTU1NR4Pe/QoUP47W9/i0svvVTtIhIREZGGqB5ennrqKcyZMwezZs3C6NGjsXTpUsTExOCVV17xeI7dbseMGTPw6KOPYtiwYWoXkYiIiDRE1fDS3t6OLVu2oKioqOuGOh2KioqwceNGj+f98Y9/RFpaGmbPnq1m8YiIiEiDVO2wW1tbC7vdjvT0dNnn6enp2Lt3b7fnrF+/Hi+//DLKy8t9uofVaoXVanVuWyyWgMtLREREkS+iZthtbGzEbbfdhhdffBEpKSk+nVNSUoKEhATnF+d4ISIi6ttUrXlJSUmBXq9HdXW17PPq6mpkZGS4HX/gwAEcOnQI1113nfMzh8PRWVCDARUVFRg+fLjsnAULFqC4uNi5fXacOBEREfVNqoYXo9GISZMmobS01Dnc2eFwoLS0FHPnznU7Pi8vDzt27JB99oc//AGNjY145plnug0lJpMJJpNJlfITERFR5FF9krri4mLMnDkT+fn5mDx5Mp5++mk0Nzdj1qxZAIDbb78dWVlZKCkpgdlsxpgxY2TnJyYmAoDb50RERNQ/qR5epk+fjpMnT2LhwoWoqqrChAkTsGrVKmcn3srKyh4XYCIiIiI6SxJCiHAXIpgsFgsSEhLQ0NDA5QGIiIg0wp/3N6s8iIiISFMYXoiIfPTV/lq8uflIuItB1O/1uVWliYjU8vOXNgEAzsuMx3mZCWEuDVH/xZoXIiI/nahvC3cRiPo1hhciIiLSFIYXIiI/9akhmkQaxPBCREREmsLwQkTkJyncBSDq5xheiIj8xGYjovBieCEiIiJNYXghIiIiTWF4ISIiIk1heCEiIiJNYXghIiIiTWF4ISIiIk1heCEiIiJNYXghIiIiTWF4ISLykxCcpo4onBheiIiISFMYXoiI/CRJXN2IKJwYXoiI/MRmI6LwYnghIiIiTWF4ISIiIk1heCEiIiJNYXghIiIiTWF4ISIiIk1heCEi8hPHGhGFF8MLERERaQrDCxGRnzhFHVF4MbwQEfmJzUZE4cXwQkRERJrC8EJERESawvBCREREmsLwQkRERJrC8EJERESawvBCRBSAtg47fvX6Fry19Wi4i0LU7zC8EBEF4LWvDuHjnVUofnN7uItC1O8wvBARBeB0S3u4i0DUbzG8EBEFgjPVEYUNwwsRERFpCsMLERERaUpIwsuSJUuQm5sLs9mMgoIClJWVeTz2rbfeQn5+PhITEzFgwABMmDAB//znP0NRTCIiItIA1cPLypUrUVxcjEWLFmHr1q0YP348pkyZgpqamm6PT05OxkMPPYSNGzfi22+/xaxZszBr1ix88sknaheViMgngv1diMJK9fDy1FNPYc6cOZg1axZGjx6NpUuXIiYmBq+88kq3x19xxRW44YYbMGrUKAwfPhzz5s3DuHHjsH79erWLSkTkM+YXovBRNby0t7djy5YtKCoq6rqhToeioiJs3Lixx/OFECgtLUVFRQUuu+wyNYtKROQzSQp3CYj6N4OaF6+trYXdbkd6errs8/T0dOzdu9fjeQ0NDcjKyoLVaoVer8ff//53/PCHP+z2WKvVCqvV6ty2WCzBKTwRkQdCAMwvROGjangJVFxcHMrLy9HU1ITS0lIUFxdj2LBhuOKKK9yOLSkpwaOPPhr6QhJRv8ZmI6LwUTW8pKSkQK/Xo7q6WvZ5dXU1MjIyPJ6n0+kwYsQIAMCECROwZ88elJSUdBteFixYgOLiYue2xWJBdnZ2cB6AiIiIIo6qfV6MRiMmTZqE0tJS52cOhwOlpaUoLCz0+ToOh0PWNOTKZDIhPj5e9kVERER9l+rNRsXFxZg5cyby8/MxefJkPP3002hubsasWbMAALfffjuysrJQUlICoLMZKD8/H8OHD4fVasVHH32Ef/7zn3j++efVLioRkZvWdjuijXqfjhVCQGJvXiLVqR5epk+fjpMnT2LhwoWoqqrChAkTsGrVKmcn3srKSuh0XRVAzc3NuPvuu3H06FFER0cjLy8Pr7/+OqZPn652UYmIZD7ecQK/WrYVf5g2Cr+4dJjXY9/ffhyL3tuFpbdOwuShySEqIVH/JAnRt6ZbslgsSEhIQENDA5uQiKhXRv7hY1htDgDAocXTkDv/QwDA0lsnYcvhOrz45UG3fQOMeuz649TwFJhIw/x5f3NtIyKiILL3rX8PEkUkhhciogCwbwtR+DC8EBH5TcBTizsrXojUx/BCRBREzC5E6mN4ISLyG5uMiMKJ4YWIyG9e6ldY9UKkOoYXIqIgEkwvRKpjeCEiCgA75hKFD8MLEZEHgYyGZqghUh/DCxGRBwwiRJGJ4YWIKIiYd4jUx/BCRORBYM1GjC9EamN4ISLqozrsDpxqsoa7GERBx/BCRBREkVTv8qNn12PS/3yGQ7XNEELg3fJjqKhqDHexiHrNEO4CEBFpjRCeQ0oktRpVVHcGlY93ViEvIw7zVpQDAA4tnhbGUhH1HmteiIj6gV3HG8JdBKKgYXghIgqiQDr5qi0Sy0TUGwwvRET9FEdGkVYxvBAR9QPKnPLOtmMo+FMpth+pD0t5iHqD4YWISAUbD5xC6Z7qcBcDANBdq9F9K8tR02jF3cu2hrw8RL3F0UZERCq45cWvAQBlD12FtDhzmEvjmc3hCHcRiPzGmhcioiBS1nKcamoPSzlcscMu9TUML0REHkguUUTZudXXvq4OdoolCjqGFyIiP/kTRyIlu0RIMYiCguGFiKiPaGjtwJK1+3GkrkX2udRtl10i7WJ4ISLyQHior+gpCoRr/pTfv7UDT3xSgev+tj4s9ycKFYYXIiIfuOaRSG022vj9KQBAfUtH6G5KFAYML0REHnhqbqmsa/FYK6OkZoddIQTufWMbnvhkr3O7OxxtRH0NwwsRkZ8Wf7zX6/5Aa2n8VX6kHu9tP44law/0eKy3DOVwCNzywte4e9mWIJaOSD2cpI6IKACeamUkRTWHmv1frLbgTDD3fW2Ts8nJ4RDQ6VhVQ5GNNS9ERAHw1mwkPHwfbMGKGA6OoyaNYXghIvJA71IDEej7PZR1GMHIIMwxpAUML0REHkR6R1dlE1Wgx0XKRHpEvmJ4ISLyQB9g349QzfOiRrgK1xw1RP5geCEicvHwOzvxi39shhACukivelEIRu5gdCEt4GgjIiIX//z6MABg5zGLLLz4ujCj3eHrDDC952u0ktBTB2NGFtIW1rwQEXXD5gh8GHJTmy2IJfFMnWaj4F+TKNgYXoiIutGbd/jBU81BK0eosRaGtIDhhYioG72pgQhd7YWvo418vyJrXkgLGF6IiLolvGz5d24oCCE4Uoj6DYYXIqJuaC0HaK28RL3B8EJE1I3eZIFQBQl22KX+KiThZcmSJcjNzYXZbEZBQQHKyso8Hvviiy/i0ksvRVJSEpKSklBUVOT1eCIiNWjtJS7gOXD5k3HYYZe0QPXwsnLlShQXF2PRokXYunUrxo8fjylTpqCmpqbb49etW4dbbrkFa9euxcaNG5GdnY2rr74ax44dU7uoREROvs7r0u253Xx2vL4VVpu9d4VScA0lvenvojy12WrDG2WVqG2yBnxNIjWpHl6eeuopzJkzB7NmzcLo0aOxdOlSxMTE4JVXXun2+GXLluHuu+/GhAkTkJeXh5deegkOhwOlpaVqF5WIKCDH6ltl28owsONoAy5avAY/enZ9UO/rumaRt+giSZLP4UsI4OF3d2LBWzsw48VNvSsgkUpUDS/t7e3YsmULioqKum6o06GoqAgbN2706RotLS3o6OhAcnJyt/utVissFovsi4iot/ypx/jD2zvk5yqSwrvlnTXH39U09bZY3gVpeYBPdlYBACqqG3t/QSIVqBpeamtrYbfbkZ6eLvs8PT0dVVVVPl3jwQcfRGZmpiwAuSopKUFCQoLzKzs7u9flJqL+yTV0+NMKc7qlo1f3ChSHRlN/FdGjjRYvXowVK1bg7bffhtls7vaYBQsWoKGhwfl15MiREJeSiPoK1yyg7LjqrSOrw88QsXZvDS54vBRf7Dvp13neeCuCf5PUscsuRT5Vw0tKSgr0ej2qq6tln1dXVyMjI8PruU8++SQWL16MTz/9FOPGjfN4nMlkQnx8vOyLiKjX/HiD2+zeJ7RThodZr32D2iYrbn+ldyMppSCNlXa9DIMLaYGq4cVoNGLSpEmyzrZnO98WFhZ6PO8vf/kLHnvsMaxatQr5+flqFpGIyEm4fe/bq1xZ8+K6GayA0R01mo3YEkVaYFD7BsXFxZg5cyby8/MxefJkPP3002hubsasWbMAALfffjuysrJQUlICAPjzn/+MhQsXYvny5cjNzXX2jYmNjUVsbKzaxSUiAtDzS9w1ONgcng8OZb8U5g7qL1QPL9OnT8fJkyexcOFCVFVVYcKECVi1apWzE29lZSV0uq4KoOeffx7t7e346U9/KrvOokWL8Mgjj6hdXCLqx7wFDW8ZxOHw3j9GzdoXT/eU3d+/CxFFPNXDCwDMnTsXc+fO7XbfunXrZNuHDh1Sv0BERD3wp9uqt5qXUAQXX3idBwau88UwvVDki+jRRkREamts68DCd3fim0N18j4vfs2oK5QfyIQqvnDoNPUXDC9E1K/99dN9+L+Nh/GzpfKJM3u1MGPvihTYPb1PsRuc6xBFCIYXIurXDp9qdn4vm+elh7e4t1oat1N7UfWyv6YRN/59A9ZVdL8eXDBwqDRpDcMLEfVrel0I+nv04rL3LNuGrZX1uOPVb9S8Tdc1WPVCGsDwQkT9mi7ADrXezgpmCKpraQ/OhRhKqA9heCGifk1W8+JXJ10/9KLZKBiZw5/biyDdk0hNDC9E1K/pdL692r290Hvs8+Knd8uP+bTukRoZg8GFtCAk87wQEUUqg4/hxR/u/XV9v0flqRbMW1EOADi0eFrwCuUjzvNCWsCaFyLq1/Qe+rz4M7lcMOehq2lsC+g8ITzXmvRUPta2kNYwvBBR/+Y6TDjAl3gwX/5hn5CXQYY0gOGFiMgHyuYU731glGsb+XMn5cHqpwnXZ2N2IS1geCEiOiNY/T3CMsPumf953k/UdzC8EBGd4W3UUIfdj9oUl0N3H7fI6lJ2HG3wWgY1mo166jAsn1k4+PcnCjaGFyIiH7xRVinb9vUl//u3d8i2f/3GVq/Hh7/LC9MLRT6GFyKiM9yGOLskiYMnmxEMVpvD9/L4UQ3SmxoT1ryQ1jC8EBF54Poitzn8CBJeai867N7Di+sQ7WAFibCPYCIKMoYXIqIzvNV0hKoWxDVnOITwb8kCD8eW7qnxPjpKMdqITUcU6RheiIg8cK2x8Kfmxdt1/Dm2uzu+ve0oZr1aBktbh8/B5rM91T7fn6tKkxYwvBARnaF8bbu+xx2Kl7rXYcl+rIPU9bn7jtMt7W7B5zcrt2NtxUn8fe0BRXkCx7xCWsPwQkQUZP6ubXSqyYpL/7IWf/20Qvb5/W9u9xgsGlrdg00wMgiDDGkBwwsR0RnKF7evzT3KWpOPd57w674vrT+Io6db8dya/bLPvzpwyss9GTSo/2J4ISLyQDaE2I/z3tp6zPM1u/nM4aE/jRC+d531p6/KnhMWL9fx+TJEYcPwQkR0lrcXd29GEPVUgxPAUObehIwH/r09aNciCgeGFyKiMwIdIny8oS3IJekk4DnX9GY4s83upbMxh0mTBjC8EBH5IGiLNnZTzVHf3OHhWM8VPnXdnRNAEastVvk8L+xLQxrA8EJEdIY/Q5yD+YJfufmI8/s/fbTHp3OUc7f0VBzXgKIc9t2bOWyIwoHhhYioG8GcUf8fXx1yfn+6pftalrO+/r4uiHfunjKr2F0+EOByAhT5GF6IiM7w2l+3F5UTljZb4CcHyNvoI7cJ92QLM/q3JAFRODC8EBH1AX4FDrcmMOFpF1FEYnghIjpD6+v6uPZrKT9S7/E496UOiLSF4YWIyAfhGkLsPVB53vfld7Ue90mKTi2uYUbj+Y36CYYXIqIz1OrzEhKK8ul1fvS6FR43iCISwwsR0RmBLgcQPvKA4lp+neR5n5LrrogPaURgeCEi8onynb56d3W3x4WW56Thz3BnZR8YokjH8EJEdIZ8plnhtUaiptEamkKFgOuzffldLaw2R/gKQ+QDhhciogjm86rSvWjocj3zjx/sDvg6RKHC8EJEdJaiz0soJpp99P1dPh/b2OZ9dt5A4wubjUhrGF6IiDwIxWwor2445POxmw+d9rjvlfUHAy8EswtpDMMLEdEZ3kYMR0LlhLJpqLW9q2/Ks2v2B3xd1ryQ1jC8EBFphDJj3PryJlWuSxTpGF6IqF+TXHq2yOd5ibwp9HsKGd5m4/V2KmteSGtCEl6WLFmC3NxcmM1mFBQUoKyszOOxu3btwk033YTc3FxIkoSnn346FEUkon4qXNP++yrQXOHPeT0dKoRAkzX0K2MTeaJ6eFm5ciWKi4uxaNEibN26FePHj8eUKVNQU1PT7fEtLS0YNmwYFi9ejIyMDLWLR0TkJJ/nRbEvAmonglWCg7XN8uv28Gxz/m8zxiz6BN+fbApSCYh6R/Xw8tRTT2HOnDmYNWsWRo8ejaVLlyImJgavvPJKt8dfcMEFeOKJJ3DzzTfDZDKpXTwiIidfp9APF7UC1PKyI173f7an8x+bz5Z+p8r9ifylanhpb2/Hli1bUFRU1HVDnQ5FRUXYuHFjUO5htVphsVhkX0REvRUBFS1uqi1tXvcHWuQv9p306bh3yo8HeAei4FI1vNTW1sJutyM9PV32eXp6OqqqqoJyj5KSEiQkJDi/srOzg3JdIup/IjCvyDz8ru8T2hH1ZZofbbRgwQI0NDQ4v44c8V79SUTkC2WQCVdNTKBNRZHeEZmoNwxqXjwlJQV6vR7V1fLVV6urq4PWGddkMrFvDBEFRaBDjSNFJDZ1EalB1ZoXo9GISZMmobS01PmZw+FAaWkpCgsL1bw1EVGvlHy8J9xFAABIUihWWCLSFlVrXgCguLgYM2fORH5+PiZPnoynn34azc3NmDVrFgDg9ttvR1ZWFkpKSgB0dvLdvXu38/tjx46hvLwcsbGxGDFihNrFJaJ+zLXm4vuTzUgeYHRuH1IMLw6VYI0wen7dgaBchygSqB5epk+fjpMnT2LhwoWoqqrChAkTsGrVKmcn3srKSuh0XRVAx48fx8SJE53bTz75JJ588klcfvnlWLdundrFJSLqVmVdS1juG2h0CVcTUmu7Ha9sOIiiUekYmRGHaksb3i0/hun5OUiIicLR0y3YV92IH4xMY60SBUz18AIAc+fOxdy5c7vdpwwkubm5ETEZFBGRli3fVBmW+y79/ACeKf0OT3xSgUOLp+GOV7/BnhMWlB2sw0szL8Alf14LAHjljnxcmZfew9WIuqf50UZERMHSl/7ddKy+NejXHJYyoNvP73ztG/xkyQYIIbD7hHyurT1nts9OdHfWpoN1QS8f9R8hqXkhIqIARVCg6q4ojW0dWLO3M5jsOm6ByeDbv4nPLoj52oaDcAjgzkuGBquY1A8wvBARncG5UfzXbLU7vzfoJRh9DC8A0GS14ZH3Owdo3DRpMBKio4JePuqb2GxERORBJPS/s0dAGc7y5efha82LgECHzeHcttrsXo4mkmN4ISI6I4JyglNLe+S81Lv78biGK7tDwKgP8LUSgT97ilwML0REHnAoL/CLf2z2ut/h6EodQgA6nY8/MwE4XIKPIwLDi90hYGnrCHcxqBsML0REZ7ivZxSBb9QQ+2xP1/IuZ38cpXuq8ddPK+BwCFkAsfuRQITieEcE/qyveeYLjHvkUxw9HZ45fsgzdtglIiKfnA0Ys8/UxpybHocxWQlu+30hhHBrcookn+2uxr7qJgDAxzuqMOeyYWEuEblizQsR0RnKmhY2G8kps8mfPtoTcO2JEIDNHrnh5bf/3u72WX1LO9pdOhlT+DC8EBGdwWYj/5xoaJP9jOx+vNeVzUZ2ISCEwF8/rcCnu6qCWMrARCk6HtdY2jDhj6txzTNfhKlE5IrNRkREFDC7CLzmRdls9NmeGjy3Zj8A4NDiaWeOE2GpAVP2O37/2xMAgAMnw7NAJ8mx5oWI6AxWtHinXHLgipGpcLjUtjj86rArFLU2AlUN8uvvr2lC/v98hpe+/D6wAveCXhGY/ufD3SEvA3nG8EJERAHRSZLX4c7emt2EkDcz2R3Cbb2jR9/fhVPN7fifD/cAAI6ebsFjH+zGkRCs8K2s7enuUbZVnsZxFdaQop6x2YiIyIlVL/6QIG8qUs4G3FNNlnKY9QdnmmZcP3N11+tbsPOYBaV7qrHugR8EVmgf6bz8076tw47Dp1pww9+/AtDVxEWhw/BCRHTGo+/LmwZOt3CCsp54G23krQ+MEMJr8OnOzmOdK1QfOqV+zYvOSz+bz/edxOnmdtXLQJ6x2YiI6Iwvv6sNdxE0x7VyRNnnxVsXGAF5zYyylsXhEOipn26T1YbXNhxUZfiyt/Cik6SIWrahP2J4ISKigHmbYVdZ8+LaB0YI7+fWNlu9BggAGLPoEzzy/m6Mf/RTAMD+mkYsWbsfrUEIFt5uPSjBjD9+wA684cTwQkT9jhACC976FkvW7g93UTTP4fDWYVdxrMu2gJA3OSlO1kkS2jp8CyGtZ44reuoLPPFJhfP3+t7247jt5U2yJh5vnYi/q27EM599h2arTTbaSEDguvGZPpWFQoN9Xoio3zlwsglvlB0BAFw/gS+lQEmS93lelNt2h7LmpWufzSFwVV4aSvfWdF4bwDeHTrvdz5fh7FsrO8+7941tAIC/rq7A/1w/Fg+9vQPrKk7iw3svQWKMEU1WG9Z/V4srRqbCHKXHD/9f5wR0p1va3Wp9TIauf+tH4jpM/Q1rXoio32nr6Ooj0RFh09JHOmXNhetmT+HFdVsormUXAhePSJFtKxk8rFit/FzZBFVtsQIAlm2qxLH6VizbVAkAuHvZVtz1+hY8+v4u2fHlR+plzUYSvA8Jp9BjeCGifo1LAPhH+eOSTfHfQ4ddh1ufF5d9ig66jm764Oo9hBebw3MNDwB0KNYtONsc9cW+kwDgrIU7S5LkHXYFhGwUPWtewo/hhYj6te5ekuSZskbE4aXZSBkM5aFC3ufF5hCyMGMXApedmyo73+Bt8hUXyjBzqLYZVltX/5kTDW041WSVHVNR1SjbVt5KGbwovBheiKhf82V+EeribUSRMgi617x0fS+EPNw4HMJtW9kc5KnmRUkZog6dapGtYP2frUdx9LR8ZtzNh+Wz+8pqXoSy7N5DGqmP4YWI+jV/1uMhZTCQZIGlc2VoT8d6nwfGrpy0ThFehHAPM564BhXX63ddCzBH6WX7Jciv7bY8gIdyny0rhRbDCxH1O0Lx0iTfKV/U8lCgqD3poYlJGVaUYSZK3/WKarc7fK556S5MCJeQlRpnkh0zcIBRFoz2VTW6rSrtrXlM2UxF6mN4IaJ+x654aZLvvC2+aHe4Nw25UtZ+KH8PDmWzkb4rQbTbHD7XvCgDaVqcSfbZ5KHJsntdNz5Tdq/mdrus7HtOWHwOZRQaDC9E1O94W4+HvFM2/bgO5FHWprh34HU9Vhl8BFbvru7aFkI2UZzV5oAuwJoXSVJ2uBWy2hK7Q8DgUstzVV6abIK8d8qPw9Jqkz1HnLlrmjTWvIQewwsR9TuuL7IN+0+FsSTa85s3y1223MOKQxFQXCkDhKy/jENgW2W9bNv1+Habo8flAs6yKXoOS5BkocvukI90sgshWx9p1KB4t9l9a11GJzmEQFKM0bnd2GYDhRbDCxH1O2wqCty6ipPO7x1CHkje2XYMltaulbhnv/aN7FyvHXSFwABjVydahwNw7Xdrswu3fiieKEc96STFvRXhRQgBu8tJNodAbsoA2TVca43c1mWyCzz+4W7kzv8QlSFY8ZoYXoioH+IIo+BYs7cGX3/fVXO1tbIeH+444dzeq5g7xfXFrgw+dofAz/Kzu7YV4abD4VAMX/b8O3SreZEkWZ8Xm7LmxSHcmr/S48yyawgoapgUNTcvfnkQAHDZE2ud5eMQavUwvBBRv8MRRsGjnJ3Wm1te/Nr5vbKJSdlMZHfIA4LNLmR9Xrz1M+muZk3ZbCS/l/y/CZtdyMKK8prKeV+OKeaMAYDZ/9iMa575EjY7Z0FUA8MLEfU7bDYKv8pTLWiydvUVcRttJOS1Ix12h6zZqLu5XFyv5VrrodO51/K4hh9lTYoyWAHuazi5hp2axja3MqzZW4O9VY3YddzisZwUOIYXIup3OMIo/HYca8Dv/v2tc7szrHTtV4YZm0PImo06vKzrYFOEk86FFeX7vXXgtTkcbsO8lQszCkVNjSvXa/nYx5j8xPBCRP0Oa/Ijz58+2otdxxuc2w6He82Lq55qXlz3S5L7ApLK0UbK/cr+Kg5lzYtr2RRBSllWCj6GFyLqd9hsFJm+PdoVXpQLNbqHF/m2sibFtdOuTpLc5pSRNRs5hNuEecr/QtyGebscsOXQadmxH357AqQuhhci6nfYbBT5lKONbHZlbYfnKfoF5DUzEnqYzVd0t8K1ouZF2WHXZfutbcdkx5Yd7Frk0XXuGgoehhci6ndY8xL5lM1GNodDFkA6bF6aaoS8KUdAPveL+1BpeRhxONw77Lre+8jpFq8BuKm9qyPyovd2eTyOAsfwQkT9DmteIl+HsqbFrhg6rehn0q4IM8o+LMp1k5RLRCjngVH2eXGtyfnTR3u9DrdvdhlFNTEn0fk9530JHoYXIup3WPMS+e56fQu2H613btvs8oDxfxsPy45/7MPdzu8F5B123UcuOWT/DRw42SSb/r+zw668PMp5ZVyvnz8kSbbPNbycbTbaeawBQxd8hH98dUj5qBQAhhci6jeqGtrwzrZjXEhPI9o6umpT6pqtsqYfZXh5a2tXv5MOu0DRU587t4/Vt+LeN7Y5tw+cbMYTn1Q4tw+fasHrX1c6tzs77Mr/G2lwWfYAkIeZ+Ogo2b6MhGi3Z/nRc+sBdDYjsQam9ww9H0JE1Ddcv2QDqixtGJ+dGO6ikJ8efte/viNWRTPSIcWaQ1UW94nlzrIr5pTpibLJauAAo2z7tQ0Huy3bM6XfYdZFuUiLly9FQD0LSc3LkiVLkJubC7PZjIKCApSVlXk9/l//+hfy8vJgNpsxduxYfPTRR6EoJhH1YR12h/OFtf1IfXgLQxFt4/ensGpXlc/HtyuGbbvOHAwAj7y/W7a9/Ug98h5ehefXHcDkP5XikzP3arba0NZhlzU7UfdUDy8rV65EcXExFi1ahK1bt2L8+PGYMmUKampquj3+q6++wi233ILZs2dj27ZtuP7663H99ddj586dahe1R2srapA7/0MsWbvfbbl0IopMLe022B0Ce05wmnbyTUu7f3/fXYdGA539W7yZ/sLXsu3//ucW3PrSJpy36BPkPbwKF/95Df76aQU+39e5gndDSwe2HD6Nu5dtQe78D3HFE2txurkdt79Shtz5H+LjHSdQ19yO1nY7vjpQi40HTnV32z5FEio3vhUUFOCCCy7A3/72NwCAw+FAdnY2fv3rX2P+/Plux0+fPh3Nzc344IMPnJ9deOGFmDBhApYuXdrj/SwWCxISEtDQ0ID4+PigPUdbhx15D69y+/zCYcm4enQGBidFY1hqLCrrmrG/pgnDUmKRNMCIHUfrUTBsIFLjTEiKMaKxrQM2h4DJoMPu4xZkJUUjPd4MvSShytKG6Cg9dJKENpsdQgCJMVHQSRJsDgcMOh1az6Ty+OgonGy04nh9K8YOTkBMlB6NbbbO4YGicw6EtDgTmq02SJKEHUcbkDcoDkJ0zjZ5+FQLRqTGItZsQG2TFXaHwACTAZbWDpxqbkdWYjSarDakxZkgSUCM0YCG1g5UVDUiI96M3ScsOC8zHpIEfL7vJCZmJyEz0YyE6ChUW6yIjtLDFKXDkboWJA0worHNhhP1rRg1KB4OIVBlaUNqrAlH61sxelA8jtS1YGCsCXpJgtVmh8mgh0MIJMZEYecxC67723rnz/zdey7GoVPNGJkRh/Q4M5IUVbSu/0m32x3osAsMMOohROeQyXabA5IEtLbbEW3s/LkZ9TqcarZigMmAWJMBdiFg7XCgsq4ZFVVN2Pj9KURH6XCy0Yq1FScxOCkaR0+34o6LcvGvzUewfM6FWLn5CMZmJcAcpUPuwAG44e9fOctxe+EQXHZOKlbvrsbKze4L2T364/PQ0NqBp1bvA9DZAfC2wiH48fhMWNpsMBl0zqGgtU3tSB5gxACjHjaHgCQBjW02tLbbMTDWCJNBj6Y2GyxtHWjtsCMpxojvTzZhVGY8qhraEGc24Cd/24CaRis2zL8S8WYDjAbdmU6NwIn6VjRZbRg1KB5Reh3qW9phitIjOkoPu0PgdEs7kmKMaLbasP1oPS7ITUZ0lB5N7Z0/x7M/922V9YjSS9BJEtLjzWjrsKOl3YYjda3ITo7B4KRoGPU6JERH4buaJiTFRMEUpYdBJ+FgbTNijHpsPnQaA0wGDE6KxtbK00iIjkJGvBlDUwfgREMbmtpseOKTCtx/9bmYkJ2IU83t+GLfSbR22DHn0mE456GPnT/jh64dhcc/2hPA//uJtCknOQaVdV3NZcNTB+DAyWZkJUZj6a2TYHM48OKX32PSkGTcMDEL5igd2jocMBp0KDt4CtlJMUiMMSLObECUXodTTVbYHAKZie59eoLBn/e3quGlvb0dMTEx+Pe//43rr7/e+fnMmTNRX1+Pd9991+2cnJwcFBcX47777nN+tmjRIrzzzjvYvn272/FWqxVWa1cvcYvFguzs7KCHl71VFkx9+sugXY+IiEirFt84FjdPzgnqNf0JL6o2G9XW1sJutyM9PV32eXp6Oqqqum9PrKqq8uv4kpISJCQkOL+ys7ODU3gFg44Ds4iIiABg/ls7UNfcHrb7a3600YIFC1BcXOzcPlvzEmzDUgZ0+3leRhzmXXUONh8+jaSYKIzMiIdOAjYdrMP5OUlIjTPCHKXHp7uqAQCZiWZUNViRmxKDJqsNFw4bCEtrBzrsArVNVuh1EvSShNMt7Wi22jA4KQY6HRCl10EnSYgzGyBBQlZSNGx2B+LMUThwsglReh2arDYkRBuQmRiNPScsOFLXivhoA85Ji8Ox+lbodRKio/Q42WjFOemxaLc5UN/agaEDB6DD7kCj1YY4kwE1jVYMT41FS7sNe040IsakR7PVhqY2G/IGxeNQbTPizAbkpgxAW4cdiTFGnG5uR4xRj2qLFUaDDjoJ2FvViJzkGAxKMMOg16GhpQON1g60ttsxalA89lU3YlhqLNburcGoQfEQEDha11nOEWmx+HjnCRj1OuytasSKb47gP7+6COfnJMJqc8AcpUdrux17qiw4UteCxjYbthw+jcJhAzEpNwlGvc7Z9DBk4ABY2jpw+FQzDtW2oM1mx9HTrcjLiMMPRqYhSt/ZJGNp68Dw1FgcqWuBJHW2e6fHm9Fhd+CSP6+V/d4z4s3Q6yTcMDELS9btx4/HZ+Ld8uO4IDcJhcMGIjdlAIrf7KopjDMb0NjWfSe8p6dPQGZiNI6ebkHxm9sxbewgFAxLxsJ3dyEvIw6P/Pg8bK08jZzkGGTEmyFJgKXVhna7A1abA5kJZsSZo9BhdyBKr0O7zYEhKTFo6+hsfjvblPNdTSMyE6NhtTlw7HQr/ut/N2LgACNuvXAIRmfGo665Heemx+Lc9DhYbQ4cPtWCnOQYfLanGhcPT8H1f9+AeLMBT988ERI6q6RjTHqUV9bjwx0nkBjT2ZSVFGPE1wdP4eeTc5CbMgBVDW1wCAGDTgeHENh9woIbJmbhYG0zmq02lB+pR15GPHJTYnCqqR1H6lpwrL4Vpig9bpiYhU93VUGvk3B+ThL0Ogn/3nIUk4cmIzEmClsP12NkRiyMer2zafHeK0fAbNTjL6s6h8Kuuf9yXPnXz7v92RP1BQOMelw+MhU/GJmG9789gYToKLy//TiAzr9VV45Kw9bDp2Ey6PDSzAuwr7oRM17ahLuvGA6jQYflmyrxkwmZePHLrlFRf75pLN7bfhynmztwsLYZbTY7hiTHOEdtvTwzH8mKJvtQ0nyzkZJafV7O+uFTn2P/ySbsfnQqoo36oF+f+g6HQ0Cncx9u6elzUs91z63HjmMN+MmETHy2uxrNfnbIJApEdnI0jtS19uoaRr0Oex+bisY2Gz7/7iSuzEvDnhMWbNhfi8m5yWi02lA0Kh16xd8Uq82OGosV2ckxvbp/KEVMnxegs8Pu5MmT8dxzzwHo7LCbk5ODuXPneuyw29LSgvfff9/52UUXXYRx48aFtcMuEfUNFy9eg2P1vXuhEPni9sIhbpPpebPk5+fjyrw0fL6vBm9vO4aHrh2NnIHaCR+95c/7W/Vmo+LiYsycORP5+fmYPHkynn76aTQ3N2PWrFkAgNtvvx1ZWVkoKSkBAMybNw+XX345/vrXv2LatGlYsWIFNm/ejBdeeEHtohJRP5AYE8XwQqqYfclQvLy+q+klI8F98rlJQ5Kw5fBp5/aBP13rVmsydcwgTB0zSL2C9gGqh5fp06fj5MmTWLhwIaqqqjBhwgSsWrXK2Sm3srISOpfOsBdddBGWL1+OP/zhD/j973+Pc845B++88w7GjBmjdlGJqB9IiglfOz31bRmKmXKPu4TklFgjvvzdldhX3Yj5b+3AtLEZmDomwy24kG9UbzYKNTYbEZE3c5dvxQffngh3MagPenr6BNy3sty5/cod+bjztc0AgOVzCnDR8JQwlUwbIqrZiIgokriuBkwUTNFGPS4/N9U5M25CdBQ+K74MR+paGVyCjOGFiPqVIf2oAySFlkEnYZBLPxdzlB4j0uIwIi0ujKXqmzjzGhH1KzMKhuD8nEQ8dj370VFwGfQ6mKO6ptCIjuJ0GmpheCGifiVnYAzeuvti3HyBOrNxU/C9eHt+wOfeeH4WUmJNzu3/N308bpyY5fH4H45O97ivJwadJJv/i3OBqYfhhYj6Jb3EUR6R6vbCIbJtZQ3G8jkFzu9vOn8wHrlutHP7yrw0zLo417k9ITsRPy/oWoMnKcaIGJPnUPHCbZMCLTYMOgkdNodzOyaKPTPUwvBCRP0Ss0vkumSEvHOrQS//ZZ1d8gIAkmKiZLPIXn5uKs7LTHBuDzAaEOUyHNkcpYdR7zm8SL34D8Og1+GVDV3zvJiNfMWqhT9ZIuqXevOSInVdmZcm21b+poyGrldXQnQUTAa9bNu1psZqc2BvdaNzO0qvkwXXxTeODU6h0Vnz4nAZzOYasii4+JMlIqKIYlC89F3Dysj0OES57E+IiZLt1+kkRLvUeAxKMGPnsQbnttVmh9XWtbbVTycNRm6AI9AuOzdVUW4Jr826wLnNgKwehhciIopoowZ1TVg2dnCCLKzEm6NgctneV9UoCzfjBicgITrKuZ0Sa5LV1Bj0Otgcgc39c/XodDw4Nc+5HaXX4YqRaShf+EMcLLk2oGuSbxheiIgoork2v8SaDLLtOLMBDa0dzu0rR6XB2tHVaTYhOgpDBg5wbp+bHoccxUrL9gDDi14nIcqlP87Zqf4TY4ysdVEZwwsREYXV8jkFXtf40bns21tlkdWsxBgNGJ+d6Nwem5WAhJiumhaDXoc/39TZr+XsKKb/ys/GxSMG4oEpIwEg4JoXvSTJyh2l4ys1VDiOi4iIwio11uSx9uPdey6WbXcolndIjTMiIToKH8+79ExNiA75Q5Iw9wcjcE56LIDOgHNo8TTnOdFGPZb94sJel1uvk2T9c/R61raECsMLEREF3bnpsdhX3eTcvuvy4Vj6+QEAncPUb5iYhbe2HgMAWbPOWYcWT4PDIZy1LpOHJqPsYB1enXUBDC61HTnJnee69ouRJAm/PVOr4otA5/zR6yRZWaK4QnTIsI6LiIh84jr5GwA8dO0o5/fTxg6S7Vv5y0LZ9tQxGc7vhQAcLjUtrh1wXbk2F73534U4tHga4s1RiDEa8PG8S/Hpby7zeK4/nr1lIkwGHR798Xl+nafXSWht7xq5pBwlRerhT5qIiHxy/QT5tPoO0RVAlDUdyqnxXWsonvzZeHQE2M/krFGD4nFuenAWPJw8NBk7H52CmRfl+nWeXifh1a+6JqVTTqZH6mF4ISIin0QpahaaXWodlHOlmBVT+ru+2C/ITYLN7kAkUT6bL/Q6CR02lxok1ryEDH/SRETkE6NBXrPQ1tEVXrwNDb73yhGyfiXRRr1bx1st0ksSHrymq8bJwD4vIcPwQkREPnGtnYiO0qPDx9qTKMVEcDFGA4anyjvpzv3BCADAvKvOCUJJg2fprec7vy+5cSxmuiwaqddJyE6KkW1TaHC0ERER+cQ1vLw0Mx/bKk/7dJ4pSod2l9WWo6P0mH5BNr45dBqzLxkKALj/6nNx4/lZGJriPvIonCYNSXZ+nx5vwjlpmfjHxsMAOsNKjktzGSemCx2GFyKifuz2wiH4vzMv4564hpeMBDPuuHgoNh2sk40kAiCbjh/oHM481KWmRa+TMCItDu+4zOEiSRKGpcYG8giqch3NZI7Sy2pX9DoJaXFmvHPPxYg1eV6pmoKP4YWIqB/TBVhbEGc2INZkwD9nFzg/W3rr+Xjsgz149paJAIC//XwidhxtwNWj06HTSSh76Cq3jryRzrUTbnfhBQAmuMzwS6HB8EJE1I/5k13izF2vDGXtCgBMHTMIU8d0zffyo3GZ+NG4TOd2Wpw5sEKGkWvNS3SUXjY8nH1cwofhhYiojxmUYMaJhrZu9yXFROF0S9dChv7UvJij9Pis+DJIkiRbmbmvGTIwBodPtWBEWiz0Ogk/mZCJuuZ2jEyPw76aRudxgdZaUe8xvBAR9THXjBmEVzYc7HafslOpv5UHI9KCMzFcJHt9dgFeXn/Q2Zn4mZsnOve5Dvl27YRMocWh0kRE/Ygyq3irPbhxYpbHfX1ZdnIMHvnxechOjnHb5zoxcLWl+9otUh/DCxH1W+/PvSTcRQgJ19oVZc2Lt+G9F49IcX7fX4OMkuvcNpOGJIWxJP0bwwsR9VtjByeEuwgh4Vq7oswqvnbbGNxNLUR/5Nphd3BSdBhL0r8xvBBRv7bylxeGuwhBJyCfet81vCj7uHjr8yK7itD+dP7BMHpQPM7LjMcPR6dzUrowYoddIurXCoYNDHcR1OfyjlX2ceGIGf8Y9Dp88OtLGFzCjDUvRER9jKTolit5+B7w3udFuNS2JMYYg1CyvoHBJfxY80JE1MdJXjrs6nt4Ef+/6eOxZu9J/LwgR42iEQWENS9ERBowoxfhwbUmRplVeurzcsPEwXjulomam9af+jaGFyIiDQhW3xS38OJzj12iyMHwQkSkAdeMla/cnDvQ96HLsmYjt14vRNrD8EJEFOF+NG4QohXNNjHGwLosehtGTaQVDC9ERBGuu9Et/rTouAYUh0O5z/N5yqBDFCkYXoiIIpwE9Ybneqt54bx0FKkYXoiINMCf6OJPzvF2LLMLRSqGFyIiDehNxYvrqUL43ueFNS8UqRheiIg0IFijhJR5xFufF6JIxfBCRBThJKl3NS+ulLUp3uZ5YYddilSqhZe6ujrMmDED8fHxSExMxOzZs9HU1OT1nBdeeAFXXHEF4uPjIUkS6uvr1SoeEVFEWf6LAo/7ep1bXC6gDCTe1zbq7Y2J1KFaeJkxYwZ27dqF1atX44MPPsAXX3yBX/7yl17PaWlpwdSpU/H73/9erWIREUWk8dmJGJYywON+ZcZQ9l3xxrVfi/uq0j5fhihiqBJe9uzZg1WrVuGll15CQUEBLrnkEjz33HNYsWIFjh8/7vG8++67D/Pnz8eFF16oRrGIiLzKSoyWbedlxPX6mimxJp+OU4aK12ZdINvuTZ+XWFPXhHbK8njtsBvwHYnUpUp42bhxIxITE5Gfn+/8rKioCDqdDps2bQrqvaxWKywWi+yLiCgQv51yrmz7bz+fqOr9fn9tnvN7ZYa4YmSayz5Jtv+ZmycErQxeJ6ljuxFFKFXCS1VVFdLS0mSfGQwGJCcno6qqKqj3KikpQUJCgvMrOzs7qNcnIlJD0ag05OcmB3TuZeekBnxfdsKlvsCv8DJ//vwz/wLw/LV37161ytqtBQsWoKGhwfl15MiRkN6fiPoO96aZ0HUICdbEckDwSs2KF4pUfq3sdf/99+OOO+7wesywYcOQkZGBmpoa2ec2mw11dXXIyMjwcGZgTCYTTCbf2pSJiCLFnRcPlW1769PiHqm8x5NgZY6cZN9XriYKJb/CS2pqKlJTe66uLCwsRH19PbZs2YJJkyYBANasWQOHw4GCAs/DAYmI+rpnb5mIC4cmIy3ejK2Vp52f6yQ/QocUvFoRb0HoipGBN08RqUmVPi+jRo3C1KlTMWfOHJSVlWHDhg2YO3cubr75ZmRmZgIAjh07hry8PJSVlTnPq6qqQnl5Ofbv3w8A2LFjB8rLy1FXV6dGMYmIZJTNMa7b08YNCso1DToJafHmbo6TYHd4SCSKSep6ajZqtzk87vMn9Ki1GCRRb6k2z8uyZcuQl5eHq666Ctdeey0uueQSvPDCC879HR0dqKioQEtLi/OzpUuXYuLEiZgzZw4A4LLLLsPEiRPx3nvvqVVMIiKfBOs17nodm70rSegkwOFjsuipLFab3f+CAbgyL63ng4gigF/NRv5ITk7G8uXLPe7Pzc11G4b3yCOP4JFHHlGrSERE3UqJNaK2qR2Fwwaqfi/XyozUOJPL5xIcnmpeFLzNzQK41664/q31lo/++rPxmPjYap/KQBROqoUXIiKtWP/glWi22jBQMYGbOo0mXVcdmjIAz9w8Aaln7mv3tealh4L51R3G5VoxJr0/ZxKFDcMLEfV75ig9zFHqvbiTBxhR19wOwD14/GRClvN7u4euKspOtcFaYVqppxodokjBVaWJiHzQm86rH957ifN7bwHBU58XSZIHFknyPtncBblJAZSS4YW0g+GFiMgD18DSm9f6oISuNZO8XcdTeOmpNenlmV1LsQxLHYC0OPfRTM5rebkOF2kkrWB4ISIKIW+VG7dMzgEAXDxC3nF4fHaCbFtZQ3LVqHTn92lxJozPTpTtt7oMnfa2XhGHRpNWsM8LEZEPgvVez00Z4HFf8Q/PReGwgZg0pLPZ57Piy7HxQC2mX5CDg7XNsrJ4yiA6SULyACOuG5+J97cfBwCcOtPfpvNcZf8ZIu1heCEi8iDQF/t/flWIm57fKPvsw3svQU2jFcNTYz2eF6XX4bJzu2a1HZEWixFp7sd7K9fZWpmBA4we9su3uXwRaRGbjYiIfNBTkLn1whzn95OGuK8WfV5mAn4wMvBJ4OQz7HpZB6mnRRuV+5leSIMYXoiIfNBTfxBPw5fVaJaR4DlzdFfOcYO7+swoy6kctXTX5cMBAL+6YnivykikJjYbERF5EEn9V4cMjIFBJyE+Oqrbcl2Vl4bSvTWYel6G2z7XDr49PdPvpozETydleW3eIgo3hhciIg0wGfTY+egU6CSp29qVv/x0HL491oArznVfCVqv8zzkW9nxV6eTMCItLhhFJlINwwsRkQ+8VVj8dNJgj5PGBbNLibdZgAfGmjz2qdHLal6UzUZE2sM+L0REvfSzSYPDXQSvXNcsWnBNnmyfj8spEUUUhhciIl/0NIonhB12feE6Gd0ffzwGw1IH4JHrRqNAsXK2t2UGiCIVwwsRUQA2zL9Sth1pIcC1NDkDY7Dm/itwx8VD3Y4bnBQTukIRBQn7vBAReSBbDFFRh+JpEjgltRY79DbNf+d+364zKiMOT/3XeGQmRvd8MFGEYM0LEZEHrrUp8dGe/63Xm0njAnXTmX42Y7Liu93va02QAHDj+YNxoaI5iSiSseaFiMgHd148FK9uOOT3eWr1efnvy4ZjbFYCJigWYTyLHXGpL2N4ISLywLWpaIApsD+Xaq3UrNdJuPQc9zld/MWQQ1rEZiMiIg9cm14CjSDhmqXX10wSZ+a/YUl7GF6IiFw8+bPxAIDHrh/j8zneAkq4wstED81Jri4aPjDgGiWicOJ/tURELn46aTCmjR2EaKMeh081ezzO11Ci1mijntx0fmeH3vOHJLntizUZ0GS14b6ic0NdLKKgYHghIlKINrpPw+9vBhmZHoeK6kZcNy4zSKXyj04n4Wf52d3uW//gD3CkrhVjXVabJtIShhciIhWs/O8L8fX3dbgyr/v1hsIpMcaIxBjf5qkhikQML0REvSTBfdROYowRU8dkhKU8RH0dO+wSEQXA01pGRKQ+hhciIh/0FFbCNaqIqD9ieCEiCgJO9kYUOgwvREREpCkML0REvlA0C7GZiCh8GF6IiIhIUxheiIh6ibUwRKHF8EJE5AMGFKLIwfBCRBQAZZbhYCOi0GF4ISIiIk1heCEi8kEPU9Rxvl2iEGJ4ISIKAjYbEYUOwwsRUQAk9uAlChuGFyIiD1yn/GdYIYocDC9ERD7wFl2Ya4hCi+GFiMgD134sOiYUooihanipq6vDjBkzEB8fj8TERMyePRtNTU1ej//1r3+NkSNHIjo6Gjk5Obj33nvR0NCgZjGJiLrlcGk3UmYXRhmi8FE1vMyYMQO7du3C6tWr8cEHH+CLL77AL3/5S4/HHz9+HMePH8eTTz6JnTt34rXXXsOqVaswe/ZsNYtJRNSjnipeBIcbEYWMQa0L79mzB6tWrcI333yD/Px8AMBzzz2Ha6+9Fk8++SQyMzPdzhkzZgz+85//OLeHDx+Oxx9/HLfeeitsNhsMBtWKS0TkRtZhl3UtRBFDtZqXjRs3IjEx0RlcAKCoqAg6nQ6bNm3y+ToNDQ2Ij4/3GFysVissFovsi4goODw3G7mSethPRMGlWnipqqpCWlqa7DODwYDk5GRUVVX5dI3a2lo89thjXpuaSkpKkJCQ4PzKzs7uVbmJiM5yrXlRdthVhhU2GxGFjt/hZf78+ZAkyevX3r17e10wi8WCadOmYfTo0XjkkUc8HrdgwQI0NDQ4v44cOdLrexMRAfLRRqxYIYocfnciuf/++3HHHXd4PWbYsGHIyMhATU2N7HObzYa6ujpkZGR4Pb+xsRFTp05FXFwc3n77bURFRXk81mQywWQy+Vx+IiJfySepC185iEjO7/CSmpqK1NTUHo8rLCxEfX09tmzZgkmTJgEA1qxZA4fDgYKCAo/nWSwWTJkyBSaTCe+99x7MZrO/RSQiCgq9riuxeJthl7PvEoWWan1eRo0ahalTp2LOnDkoKyvDhg0bMHfuXNx8883OkUbHjh1DXl4eysrKAHQGl6uvvhrNzc14+eWXYbFYUFVVhaqqKtjtdrWKSkTUreGpAzBt3CDcXjjEbZ97YGGnF6JQUXXs8bJlyzB37lxcddVV0Ol0uOmmm/Dss88693d0dKCiogItLS0AgK1btzpHIo0YMUJ2rYMHDyI3N1fN4hIRyUiShCU/P9+nY9lhlyh0VA0vycnJWL58ucf9ubm5EC7/j7/iiitk20REWsE/XUShw7WNiIiCQLDZiChkGF6IiIKANS9EocPwQkQUBMwuRKHD8EJE1EsSWPNCFEoML0REQcA+L0Shw/BCRBQMzC5EIcPwQkQUBMwuRKHD8EJE1EuSBM5RRRRCDC9EREEQY1J1zk8icsH/txERBUHxD8/FnhMW/Fd+driLQtTnMbwQEQVBSqwJb999cbiLQdQvsNmIiKiXJChXmCYiNTG8EBH1ksTsQhRSbDYiIgrQLZOzUdXQhvMy48NdFKJ+heGFiChAJTeOC3cRiPolNhsRERGRpjC8EBERkaYwvBAREZGmMLwQERGRpjC8EBERkaYwvBAREZGmMLwQERGRpjC8EBERkaYwvBAREZGmMLwQEfmJaxkRhRfDCxGRj0YN6lzDaOp5GWEuCVH/xrWNiIh89H93TsaqnSfwk4lZ4S4KUb/G8EJE5KPUOBNuK8wNdzGI+j02GxEREZGmMLwQERGRpjC8EBERkaYwvBAREZGmMLwQERGRpjC8EBERkaYwvBAREZGmMLwQERGRpjC8EBERkaYwvBAREZGmMLwQERGRpjC8EBERkaYwvBAREZGm9LlVpYUQAACLxRLmkhAREZGvzr63z77Hvelz4aWxsREAkJ2dHeaSEBERkb8aGxuRkJDg9RhJ+BJxNMThcOD48eOIi4uDJElBvbbFYkF2djaOHDmC+Pj4oF47EvT15wP6/jPy+bSvrz8jn0/71HpGIQQaGxuRmZkJnc57r5Y+V/Oi0+kwePBgVe8RHx/fZ/+jBPr+8wF9/xn5fNrX15+Rz6d9ajxjTzUuZ7HDLhEREWkKwwsRERFpCsOLH0wmExYtWgSTyRTuoqiirz8f0Pefkc+nfX39Gfl82hcJz9jnOuwSERFR38aaFyIiItIUhhciIiLSFIYXIiIi0hSGFyIiItIUhhcfLVmyBLm5uTCbzSgoKEBZWVm4i9StL774Atdddx0yMzMhSRLeeecd2X4hBBYuXIhBgwYhOjoaRUVF+O6772TH1NXVYcaMGYiPj0diYiJmz56NpqYm2THffvstLr30UpjNZmRnZ+Mvf/mL2o8GACgpKcEFF1yAuLg4pKWl4frrr0dFRYXsmLa2Ntxzzz0YOHAgYmNjcdNNN6G6ulp2TGVlJaZNm4aYmBikpaXhgQcegM1mkx2zbt06nH/++TCZTBgxYgRee+01tR8PAPD8889j3LhxzgmgCgsL8fHHHzv3a/35lBYvXgxJknDfffc5P9PyMz7yyCOQJEn2lZeX1yeezdWxY8dw6623YuDAgYiOjsbYsWOxefNm534t/63Jzc11+x1KkoR77rkHgPZ/h3a7HQ8//DCGDh2K6OhoDB8+HI899phsTaGI//0J6tGKFSuE0WgUr7zyiti1a5eYM2eOSExMFNXV1eEumpuPPvpIPPTQQ+Ktt94SAMTbb78t27948WKRkJAg3nnnHbF9+3bx4x//WAwdOlS0trY6j5k6daoYP368+Prrr8WXX34pRowYIW655Rbn/oaGBpGeni5mzJghdu7cKd544w0RHR0t/vd//1f155syZYp49dVXxc6dO0V5ebm49tprRU5OjmhqanIec9ddd4ns7GxRWloqNm/eLC688EJx0UUXOffbbDYxZswYUVRUJLZt2yY++ugjkZKSIhYsWOA85vvvvxcxMTGiuLhY7N69Wzz33HNCr9eLVatWqf6M7733nvjwww/Fvn37REVFhfj9738voqKixM6dO/vE87kqKysTubm5Yty4cWLevHnOz7X8jIsWLRLnnXeeOHHihPPr5MmTfeLZzqqrqxNDhgwRd9xxh9i0aZP4/vvvxSeffCL279/vPEbLf2tqampkv7/Vq1cLAGLt2rVCCO3/Dh9//HExcOBA8cEHH4iDBw+Kf/3rXyI2NlY888wzzmMi/ffH8OKDyZMni3vuuce5bbfbRWZmpigpKQljqXqmDC8Oh0NkZGSIJ554wvlZfX29MJlM4o033hBCCLF7924BQHzzzTfOYz7++GMhSZI4duyYEEKIv//97yIpKUlYrVbnMQ8++KAYOXKkyk/krqamRgAQn3/+uRCi83mioqLEv/71L+cxe/bsEQDExo0bhRCdAU+n04mqqirnMc8//7yIj493PtPvfvc7cd5558nuNX36dDFlyhS1H6lbSUlJ4qWXXupTz9fY2CjOOeccsXr1anH55Zc7w4vWn3HRokVi/Pjx3e7T+rOd9eCDD4pLLrnE4/6+9rdm3rx5Yvjw4cLhcPSJ3+G0adPEnXfeKfvsxhtvFDNmzBBCaOP3x2ajHrS3t2PLli0oKipyfqbT6VBUVISNGzeGsWT+O3jwIKqqqmTPkpCQgIKCAuezbNy4EYmJicjPz3ceU1RUBJ1Oh02bNjmPueyyy2A0Gp3HTJkyBRUVFTh9+nSInqZTQ0MDACA5ORkAsGXLFnR0dMieMS8vDzk5ObJnHDt2LNLT053HTJkyBRaLBbt27XIe43qNs8eE+ndut9uxYsUKNDc3o7CwsE893z333INp06a5laMvPON3332HzMxMDBs2DDNmzEBlZSWAvvFsAPDee+8hPz8fP/vZz5CWloaJEyfixRdfdO7vS39r2tvb8frrr+POO++EJEl94nd40UUXobS0FPv27QMAbN++HevXr8c111wDQBu/P4aXHtTW1sJut8v+IwSA9PR0VFVVhalUgTlbXm/PUlVVhbS0NNl+g8GA5ORk2THdXcP1HqHgcDhw33334eKLL8aYMWOc9zcajUhMTHQrnz/l93SMxWJBa2urGo8js2PHDsTGxsJkMuGuu+7C22+/jdGjR/eZ51uxYgW2bt2KkpISt31af8aCggK89tprWLVqFZ5//nkcPHgQl156KRobGzX/bGd9//33eP7553HOOefgk08+wa9+9Svce++9+Mc//iErZ1/4W/POO++gvr4ed9xxh/O+Wv8dzp8/HzfffDPy8vIQFRWFiRMn4r777sOMGTNkZYzk31+fW1Wa+o977rkHO3fuxPr168NdlKAbOXIkysvL0dDQgH//+9+YOXMmPv/883AXKyiOHDmCefPmYfXq1TCbzeEuTtCd/dcrAIwbNw4FBQUYMmQI3nzzTURHR4exZMHjcDiQn5+PP/3pTwCAiRMnYufOnVi6dClmzpwZ5tIF18svv4xrrrkGmZmZ4S5K0Lz55ptYtmwZli9fjvPOOw/l5eW47777kJmZqZnfH2teepCSkgK9Xu/Wk7y6uhoZGRlhKlVgzpbX27NkZGSgpqZGtt9ms6Gurk52THfXcL2H2ubOnYsPPvgAa9euxeDBg52fZ2RkoL29HfX19W7l86f8no6Jj48PyQvIaDRixIgRmDRpEkpKSjB+/Hg888wzfeL5tmzZgpqaGpx//vkwGAwwGAz4/PPP8eyzz8JgMCA9PV3zz+gqMTER5557Lvbv398nfn8AMGjQIIwePVr22ahRo5zNY33lb83hw4fx2Wef4Re/+IXzs77wO3zggQectS9jx47Fbbfdht/85jfOmlAt/P4YXnpgNBoxadIklJaWOj9zOBwoLS1FYWFhGEvmv6FDhyIjI0P2LBaLBZs2bXI+S2FhIerr67FlyxbnMWvWrIHD4UBBQYHzmC+++AIdHR3OY1avXo2RI0ciKSlJ1WcQQmDu3Ll4++23sWbNGgwdOlS2f9KkSYiKipI9Y0VFBSorK2XPuGPHDtn/8VavXo34+HjnH+TCwkLZNc4eE67fucPhgNVq7RPPd9VVV2HHjh0oLy93fuXn52PGjBnO77X+jK6amppw4MABDBo0qE/8/gDg4osvdpuiYN++fRgyZAiAvvG3BgBeffVVpKWlYdq0ac7P+sLvsKWlBTqd/PWv1+vhcDgAaOT31+suv/3AihUrhMlkEq+99prYvXu3+OUvfykSExNlPckjRWNjo9i2bZvYtm2bACCeeuopsW3bNnH48GEhROfwt8TERPHuu++Kb7/9VvzkJz/pdvjbxIkTxaZNm8T69evFOeecIxv+Vl9fL9LT08Vtt90mdu7cKVasWCFiYmJCMlT6V7/6lUhISBDr1q2TDWVsaWlxHnPXXXeJnJwcsWbNGrF582ZRWFgoCgsLnfvPDmO8+uqrRXl5uVi1apVITU3tdhjjAw88IPbs2SOWLFkSsmGM8+fPF59//rk4ePCg+Pbbb8X8+fOFJEni008/7RPP1x3X0UZCaPsZ77//frFu3Tpx8OBBsWHDBlFUVCRSUlJETU2N5p/trLKyMmEwGMTjjz8uvvvuO7Fs2TIRExMjXn/9decxWv9bY7fbRU5OjnjwwQfd9mn9dzhz5kyRlZXlHCr91ltviZSUFPG73/3OeUyk//4YXnz03HPPiZycHGE0GsXkyZPF119/He4idWvt2rUCgNvXzJkzhRCdQ+AefvhhkZ6eLkwmk7jqqqtERUWF7BqnTp0St9xyi4iNjRXx8fFi1qxZorGxUXbM9u3bxSWXXCJMJpPIysoSixcvDsnzdfdsAMSrr77qPKa1tVXcfffdIikpScTExIgbbrhBnDhxQnadQ4cOiWuuuUZER0eLlJQUcf/994uOjg7ZMWvXrhUTJkwQRqNRDBs2THYPNd15551iyJAhwmg0itTUVHHVVVc5g4sQ2n++7ijDi5afcfr06WLQoEHCaDSKrKwsMX36dNn8J1p+Nlfvv/++GDNmjDCZTCIvL0+88MILsv1a/1vzySefCABuZRZC+79Di8Ui5s2bJ3JycoTZbBbDhg0TDz30kGxIc6T//iQhXKbUIyIiIopw7PNCREREmsLwQkRERJrC8EJERESawvBCREREmsLwQkRERJrC8EJERESawvBCREREmsLwQkRERJrC8EJERESawvBCREREmsLwQkRERJrC8EJERESa8v8Bp5XeecaGbewAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Shape of waveform: {}\".format(sample.size()))\n",
    "print(\"Sample rate of waveform: {}\".format(sample_rate))\n",
    "\n",
    "plt.plot(sample.squeeze().numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "                <audio  controls=\"controls\" >\n",
       "                    <source src=\"data:audio/wav;base64,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\" type=\"audio/wav\" />\n",
       "                    Your browser does not support the audio element.\n",
       "                </audio>\n",
       "              "
      ],
      "text/plain": [
       "<IPython.lib.display.Audio object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(ipd.Audio(sample, rate=sample_rate))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 64\n",
    "\n",
    "train_loader = torch.utils.data.DataLoader(\n",
    "    torch.utils.data.TensorDataset(train_audios, train_labels),\n",
    "    batch_size=batch_size,\n",
    "    shuffle=True,\n",
    ")\n",
    "\n",
    "test_loader = torch.utils.data.DataLoader(\n",
    "    torch.utils.data.TensorDataset(test_audios, test_labels),\n",
    "    batch_size=batch_size,\n",
    "    shuffle=False,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### CNN architecture: convolutions, non linearities and a readout layer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In Lab 1B we processed audio with a convolutional neural network (CNN). Our CNN was such that if the input signal contains $N$ entries, the output signal contains $N$ entries as well. This is unsuitable for a classification task  in which we want the output to be a vector of scores $s$ containing as many entries as classes.\n",
    "\n",
    "To sort this out we add a readout layer to match the dimensionalities of the input and output. A readout layer is just a fully connected layer at the output.  Instead of a convolutional filter, the last layer is the fully connected layer,\n",
    "\n",
    "$$\n",
    "   \\Phi(x; \\mathcal{H}) \n",
    "      ~=~ x_L \n",
    "      ~=~ \\sigma \\Big(\\, A_L x_{L-1} \\,\\Big) .\n",
    "$$\n",
    "\n",
    "If we have input signals $x=x_0$ of dimension $N$ and we have $C$ classes, the matrix $A_L$ is of dimension $C \\times N$. This results in an output $\\Phi(x; \\mathcal{H}) = x_L$ containing $C$ entries that we can equate to the predicted score $s$. \n",
    "\n",
    "In \\eqref{eqn_readout} we consider a CNN with layers that have a single channel. In practice, we have already seen that we have to use CNNs with multiple channels. This is not a problem but it requires that we reshape the matrix $x_{L-1}$ to turn it into a vector. We do that by stacking the columns of $x_{L-1}$ on top of each other, \n",
    "\n",
    "$$\n",
    "   \\text{vec}(x_{L-1}) \n",
    "      = \\Big[\\, x_{L-1}^1;\\, x_{L-1}^2;\\, \\ldots;\\, x_{L-1}^{F_{l-1}} \\,\\Big] .\n",
    "$$\n",
    "\n",
    "This operation has no conceptual significance. It is just a reshaping of data. We need to do this to write the readout layer as the matrix multiplication with $x_{L-1} = \\text{vec}(x_{L-1})$.I"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Modify the CNN code of Lab2B to endow it with a method that implements a single convolutional layer. Add another method to implement a readout layer. Use these two methods to modify the CNN class to implement a CNN with readout. \n",
    "\n",
    "This class receives as initialization parameters the number of layers $L$ and the dimensions $N$ and $C$ of the input features $x$ and the output scores $s$. To specify the $L-1$ convolutional layers we also accept the vectors $[K_1, \\ldots, K_{L-1}]$ and $[F_0, F_1, \\ldots, F_{L-1}]$ containing the number of taps $K_\\ell$ of the filters used at each layer and the number of features $F_\\ell$ at the output of each layer. \n",
    "\n",
    "\n",
    "The forward method of this class takes a matrix $x$ of dimension $N\\times F_0$ as an input and returns the output $\\Phi(x; \\mathcal{H})$ of a CNN with $L-1$ convolutional layers and a readout layer. The dimension of this output vector is $C$."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will use PyTorch's implementation of 1-D convolutions: **nn.Conv1D()**. This function takes the same arguments and performs the same operation as the 1-D convolution function we created in lab 1B. However, PyTorch's implemenation is faster and more memory-efficient since it uses a low-level programming language. You can read more about this implemenation in the following link: https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def convolution_1D(signal, conv_filter):\n",
    "    taps = conv_filter.shape[0]\n",
    "    out_features = conv_filter.shape[1]\n",
    "    N = signal.shape[-1]\n",
    "    z = torch.zeros((out_features, N))\n",
    "    for k in range(taps):\n",
    "        z = z + torch.matmul(conv_filter[k] , signal)\n",
    "        signal = signal.roll(1,)\n",
    "        signal[:,0] = 0\n",
    "    return z\n",
    "\n",
    "class convolutional_layer(nn.Module):\n",
    "    def __init__(self, in_features, out_features, kernel_size):\n",
    "        super().__init__()\n",
    "        self.filter = torch.nn.Parameter(torch.empty(kernel_size, out_features, in_features))\n",
    "        nn.init.xavier_normal_(self.filter)\n",
    "        \n",
    "    def forward(self,x):\n",
    "        return convolution_1D(x, self.filter)\n",
    "    \n",
    "class readout_layer(nn.Module):\n",
    "    def __init__(self, input_length, classes):\n",
    "        '''\n",
    "        inputs:\n",
    "            L (int): number of layers\n",
    "            taps (list[int]): list with number of taps for each layer\n",
    "        '''\n",
    "        super().__init__()\n",
    "        self.filter = torch.nn.Parameter(torch.empty((classes, input_length)))\n",
    "        nn.init.xavier_normal_(self.filter)\n",
    "    \n",
    "    def forward(self,x):\n",
    "        return torch.matmul(x,self.filter.T)\n",
    "\n",
    "class CNN(nn.Module):\n",
    "    def __init__(self, n_input=1, n_output=3, n_channel=[8, 8], kernel_sizes = [80, 3]):\n",
    "        super().__init__()\n",
    "        input_length = 8000\n",
    "        emb_dim = n_channel[1]*input_length\n",
    "        self.conv1 = nn.Conv1d(n_input, n_channel[0], kernel_size=kernel_sizes[0], padding='same') \n",
    "        # Setting the argument padding='same' implies that output features have the same dimension as input features\n",
    "        self.conv2 = nn.Conv1d(n_channel[0], n_channel[1], kernel_size=kernel_sizes[1], padding='same')\n",
    "        self.fc1 = readout_layer(emb_dim, n_output)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.conv1(x)\n",
    "        x = F.relu(x)\n",
    "        x = self.conv2(x)\n",
    "        x = F.relu(x)\n",
    "        x = torch.flatten(x, start_dim=1) #Vectorization operation\n",
    "        x = self.fc1(x)\n",
    "        return x\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Task 2\n",
    "Load the data from which contains audio samples in which people are speaking digits from $0$ to $2$. Train a CNN with a readout layer to classify the audio signals into the different possible spoken digits. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(model, epoch, log_interval):\n",
    "    cross_entropy = nn.CrossEntropyLoss()\n",
    "    model.train()\n",
    "    for batch_idx, (data, target) in enumerate(train_loader):\n",
    "\n",
    "        data = data.to(device)\n",
    "        target = target.to(device)\n",
    "\n",
    "        output = model(data).squeeze()\n",
    "\n",
    "        loss = cross_entropy(output, target.type(torch.LongTensor).to(device))\n",
    "\n",
    "        optimizer.zero_grad()\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        # print training stats\n",
    "        if batch_idx % log_interval == 0:\n",
    "            print(f\"Train Epoch: {epoch} \\tLoss: {loss.item():.3f}\")\n",
    "\n",
    "        # record loss\n",
    "        losses.append(loss.item())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def number_of_correct(pred, target):\n",
    "    num_correct = torch.sum(pred == target)\n",
    "    return num_correct\n",
    "\n",
    "def get_likely_index(tensor):\n",
    "    # find most likely label index for each element in the batch\n",
    "    return tensor.argmax(dim=-1)\n",
    "\n",
    "def test(model, epoch):\n",
    "    model.eval()\n",
    "    correct = 0\n",
    "    for data, target in test_loader:\n",
    "\n",
    "        data = data.to(device)\n",
    "        target = target.to(device)\n",
    "        output = model(data)\n",
    "\n",
    "        pred = get_likely_index(output)\n",
    "        correct += number_of_correct(pred, target)\n",
    "\n",
    "    print(f\"\\nTest Epoch: {epoch}\\tAccuracy: {correct}/{len(test_loader.dataset)} ({100. * correct / len(test_loader.dataset):.0f}%)\\n\")\n",
    "    return correct/len(test_loader.dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CNN(\n",
      "  (conv1): Conv1d(1, 8, kernel_size=(80,), stride=(1,), padding=same)\n",
      "  (conv2): Conv1d(8, 8, kernel_size=(3,), stride=(1,), padding=same)\n",
      "  (fc1): readout_layer()\n",
      ")\n",
      "Number of parameters: 192848\n",
      "Train Epoch: 1 \tLoss: 1.098\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/elenter/miniconda3/envs/dcase/lib/python3.10/site-packages/torch/nn/modules/conv.py:309: UserWarning: Using padding='same' with even kernel lengths and odd dilation may require a zero-padded copy of the input be created (Triggered internally at ../aten/src/ATen/native/Convolution.cpp:895.)\n",
      "  return F.conv1d(input, weight, bias, self.stride,\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 1 \tLoss: 1.105\n",
      "Train Epoch: 1 \tLoss: 1.100\n",
      "\n",
      "Test Epoch: 1\tAccuracy: 418/1241 (34%)\n",
      "\n",
      "Train Epoch: 2 \tLoss: 1.093\n",
      "Train Epoch: 2 \tLoss: 1.089\n",
      "Train Epoch: 2 \tLoss: 1.093\n",
      "\n",
      "Test Epoch: 2\tAccuracy: 430/1241 (35%)\n",
      "\n",
      "Train Epoch: 3 \tLoss: 1.082\n",
      "Train Epoch: 3 \tLoss: 1.092\n",
      "Train Epoch: 3 \tLoss: 1.080\n",
      "\n",
      "Test Epoch: 3\tAccuracy: 586/1241 (47%)\n",
      "\n",
      "Train Epoch: 4 \tLoss: 1.093\n",
      "Train Epoch: 4 \tLoss: 1.065\n",
      "Train Epoch: 4 \tLoss: 1.075\n",
      "\n",
      "Test Epoch: 4\tAccuracy: 596/1241 (48%)\n",
      "\n",
      "Train Epoch: 5 \tLoss: 1.073\n",
      "Train Epoch: 5 \tLoss: 1.044\n",
      "Train Epoch: 5 \tLoss: 1.022\n",
      "\n",
      "Test Epoch: 5\tAccuracy: 584/1241 (47%)\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'training loss')"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "log_interval = 50\n",
    "n_epoch = 5\n",
    "\n",
    "model = CNN(n_input=1, n_output=3, n_channel=[8, 8], kernel_sizes = [80, 3])\n",
    "model.to(device)\n",
    "print(model)\n",
    "\n",
    "def count_parameters(model):\n",
    "    return sum(p.numel() for p in model.parameters() if p.requires_grad)\n",
    "\n",
    "n = count_parameters(model)\n",
    "print(\"Number of parameters: %s\" % n)\n",
    "\n",
    "optimizer = optim.SGD(model.parameters(), lr=0.01)\n",
    "\n",
    "losses = []\n",
    "test_acc = []\n",
    "\n",
    "for epoch in range(1, n_epoch + 1):\n",
    "    train(model, epoch, log_interval)\n",
    "    test_acc.append(test(model, epoch))\n",
    "\n",
    "# Let's plot the batch training loss versus the number of iteration.\n",
    "plt.plot(losses)\n",
    "plt.title(\"training loss\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f701536ea40>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(test_acc, label = 'test acc')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Pooling operators reduce the dimension of the input signal while retaining the structure of time. They do so by introducing a reduction factor $\\Delta$, dividing the time line in windows of width $\\Delta$, and extracting information from every window (Figure \\ref{fig_pooling}). \n",
    "\n",
    "The simplest form of pooling is sampling. Let $w=[w_0; w_1;\\ldots;w_{N-1}]$ be a signal with $N$ components. We define the sampled signal as the signal $x=[x_0; x_1;\\ldots;x_{N_s}]$ in which the component $x_m$ is given by\n",
    "$$\n",
    "   x_m ~=~ w_{m\\Delta} ~.\n",
    "$$\n",
    "Thus, the sampled signal $x$ copies one out of $\\Delta$ entries of $w$ and ignores the rest. The rationale for ignoring entries is that the values of entries in a window, say $w_{m\\Delta}$ and $w_{m\\Delta+1}$, are not very different and we loose little information by discarding some of them.\n",
    "\n",
    "In general, computing some aggregate summary of the entries in a pooling window is more effective than plain sampling. Average pooling replaces the entries in each pooling window by their average,\n",
    "\n",
    "$$\n",
    "   x_m \n",
    "      ~=~ \\frac{1}{\\Delta}\n",
    "             \\Big( \\,\n",
    "                w_{m\\Delta} +  \n",
    "                   w_{m\\Delta+1} + \n",
    "                      \\ldots + \n",
    "                         w_{m\\Delta+(\\Delta-1)}\n",
    "                             \\,\\Big) ~.\n",
    "$$\n",
    "\n",
    "Another common approach to summarizing entries in a pooling window is to compute the maximum value,\n",
    "%\n",
    "$$\n",
    "   x_m \n",
    "      ~=~ \\max\n",
    "             \\Big( \\,\n",
    "                w_{m\\Delta} ,\\,  \n",
    "                   w_{m\\Delta+1} ,\\, \n",
    "                      \\ldots ,\\, \n",
    "                         w_{m\\Delta+(\\Delta-1)}\n",
    "                             \\,\\Big) ~.\n",
    "$$\n",
    "This is called max pooling. Max pooling is more common than average pooling and both are more common than sampling. In any case, pooling is effective when the signal $w$ does not change much within the pooling window. When this is the case, sampling, average pooling, and max pooling all produce similar summary signals $x$.\n",
    "\n",
    "The pooling operations in \\eqref{eqn_sampling}-\\eqref{eqn_max_pooling} apply to individual vectors. In CNNs we want to pool multiple features. This is done by pooling each feature individually. Given the matrix feature $w=[w^1; w^2;\\ldots;w^{F}]$ the pooled matrix feature $x=[x^1; x^2;\\ldots;x^{F}]$ contains the same number of features $F$ and is such that each of the individual vector features $x^f$ is pooled separately. Thus, Component $m$ of Feature $f$ is given by\n",
    "$$ \n",
    "   x_m^f \n",
    "      ~=~ \\text{pool}\n",
    "             \\Big( \\,\n",
    "                w^f_{k\\Delta} ,\\,  \n",
    "                   w^f_{k\\Delta+1} ,\\, \n",
    "                      \\ldots ,\\, \n",
    "                         w^f_{k\\Delta+(\\Delta-1)}\n",
    "                             \\,\\Big) ~,\n",
    "$$\n",
    "where the operation $\\text{pool}(\\cdot)$ stands in for either sampling, average pooling or max pooling."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Task\n",
    "Modify the CNN to incorporate pooling. This can be done by modifying the method that implements convolutional layers to incorporate the pooling operation. \n",
    "\n",
    "As in any CNN the initialization parameters include the number of layers $L$ along with vectors $[K_1, \\ldots, K_{L-1}]$ and $[F_0, F_1, \\ldots, F_{L-1}]$. These vectors contain the number of taps $K_\\ell$ of the filters used at each layer and the number of features $F_\\ell$ at the output of each layer. \n",
    "\n",
    "Since we are incorporating pooling the initialization parameters must also include the vecor $[N_0, N_1, \\ldots, N_{L}]$ containing the dimension $N_\\ell$ of the features at the output of each layer. Notice that $N_0$ matches the dimension of the input signal and $N_L=C$ matches the number of classes.\n",
    "\n",
    "The forward method of this class takes a matrix $x$ of dimension $N_0\\times F_0$ as an input and returns the output $\\Phi(x; \\mathcal{H})$ of a CNN with $L-1$ convolutional layers and a readout layer. The dimension of this output is $N_L \\times 1$.\n",
    "\n",
    "Use relu nonlinearities in all layers. Use max pooling in all convolutional layers where pooling is implemented.  $\\blacksquare$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Naive pooling implementations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def average_pooling_naive(signal = torch.rand(32, 2, 80), poolsize = 4):\n",
    "    output_length = int(signal.shape[-1]/poolsize)\n",
    "    pooled_signal = torch.empty((signal.shape[0], signal.shape[1], output_length))\n",
    "    for i in range(output_length):\n",
    "        start = i*poolsize\n",
    "        end = start + poolsize\n",
    "        pooled_signal[:,:,i] = torch.mean(signal[:,:,start:end],dim=2).values\n",
    "    return pooled_signal\n",
    "\n",
    "def max_pooling_naive(signal = torch.rand(32, 2, 80), poolsize = 4):\n",
    "    output_length = int(signal.shape[-1]/poolsize)\n",
    "    pooled_signal = torch.empty((signal.shape[0], signal.shape[1], output_length))\n",
    "    for i in range(output_length):\n",
    "        start = i*poolsize\n",
    "        end = start + poolsize\n",
    "        pooled_signal[:,:,i] = torch.max(signal[:,:,start:end],dim=2).values\n",
    "    return pooled_signal"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### CNN with Pooling, ReLU non-linearities and a readout layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "class max_pool_1d(nn.Module):\n",
    "    # Implements max pooling in a vectorized, faster fashion.\n",
    "    def __init__(self, pool_size, device='cpu'):\n",
    "        super().__init__()\n",
    "        self.poolsize = pool_size\n",
    "        self.device = device\n",
    "\n",
    "    def forward(self, signal):\n",
    "        output_length = int(signal.shape[-1]/self.poolsize)\n",
    "        signal = signal.reshape((signal.shape[0], signal.shape[1], output_length, -1)).to(self.device)\n",
    "        pooled_signal = torch.max(signal,dim=-1).values\n",
    "        return pooled_signal\n",
    "\n",
    "class average_pool_1d(nn.Module):\n",
    "    # Implements the average pooling in a vectorized, faster fashion.\n",
    "    def __init__(self, pool_size, device='cpu'):\n",
    "        super().__init__()\n",
    "        self.poolsize = pool_size\n",
    "        self.device = device\n",
    "\n",
    "    def forward(self, signal):\n",
    "        output_length = int(signal.shape[-1]/self.poolsize)\n",
    "        signal = signal.reshape((signal.shape[0],signal.shape[1],output_length,-1)).to(self.device)\n",
    "        pooled_signal = torch.mean(signal,dim=-1)\n",
    "\n",
    "        return pooled_signal\n",
    "\n",
    "class CNN(nn.Module):\n",
    "    def __init__(self, n_input=1, n_output=3, n_channel=[8, 8], kernel_sizes = [80, 3], pooling_windows = [8, 8]):\n",
    "        super().__init__()\n",
    "        input_length = 8000\n",
    "        emb_dim = int(n_channel[1]*input_length/(8*8))\n",
    "        self.conv1 = nn.Conv1d(n_input, n_channel[0], kernel_size=kernel_sizes[0], padding='same')\n",
    "        self.pool1 = max_pool_1d(pooling_windows[0])\n",
    "        self.conv2 = nn.Conv1d(n_channel[0], n_channel[1], kernel_size=kernel_sizes[1], padding='same')\n",
    "        self.pool2 = max_pool_1d(pooling_windows[1])\n",
    "        self.fc1 = readout_layer(emb_dim, n_output)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.conv1(x)\n",
    "        x = F.relu(x)\n",
    "        x = self.pool1(x)\n",
    "        x = self.conv2(x)\n",
    "        x = F.relu(x)\n",
    "        x = self.pool2(x)\n",
    "        x = torch.flatten(x, start_dim=1) # Vectorization operation\n",
    "        x = self.fc1(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CNN(\n",
      "  (conv1): Conv1d(1, 8, kernel_size=(80,), stride=(1,), padding=same)\n",
      "  (pool1): max_pool_1d()\n",
      "  (conv2): Conv1d(8, 8, kernel_size=(3,), stride=(1,), padding=same)\n",
      "  (pool2): max_pool_1d()\n",
      "  (fc1): readout_layer()\n",
      ")\n",
      "Train Epoch: 1 \tLoss: 1.085\n",
      "Train Epoch: 1 \tLoss: 1.094\n",
      "Train Epoch: 1 \tLoss: 1.107\n",
      "\n",
      "Test Epoch: 1\tAccuracy: 442/1241 (36%)\n",
      "\n",
      "Train Epoch: 2 \tLoss: 1.092\n",
      "Train Epoch: 2 \tLoss: 1.085\n",
      "Train Epoch: 2 \tLoss: 1.101\n",
      "\n",
      "Test Epoch: 2\tAccuracy: 534/1241 (43%)\n",
      "\n",
      "Train Epoch: 3 \tLoss: 1.076\n",
      "Train Epoch: 3 \tLoss: 1.067\n",
      "Train Epoch: 3 \tLoss: 1.110\n",
      "\n",
      "Test Epoch: 3\tAccuracy: 598/1241 (48%)\n",
      "\n",
      "Train Epoch: 4 \tLoss: 1.005\n",
      "Train Epoch: 4 \tLoss: 1.070\n",
      "Train Epoch: 4 \tLoss: 1.101\n",
      "\n",
      "Test Epoch: 4\tAccuracy: 548/1241 (44%)\n",
      "\n",
      "Train Epoch: 5 \tLoss: 1.004\n",
      "Train Epoch: 5 \tLoss: 0.995\n",
      "Train Epoch: 5 \tLoss: 0.895\n",
      "\n",
      "Test Epoch: 5\tAccuracy: 761/1241 (61%)\n",
      "\n",
      "Train Epoch: 6 \tLoss: 0.958\n",
      "Train Epoch: 6 \tLoss: 0.804\n",
      "Train Epoch: 6 \tLoss: 0.871\n",
      "\n",
      "Test Epoch: 6\tAccuracy: 866/1241 (70%)\n",
      "\n",
      "Train Epoch: 7 \tLoss: 0.803\n",
      "Train Epoch: 7 \tLoss: 0.672\n",
      "Train Epoch: 7 \tLoss: 0.940\n",
      "\n",
      "Test Epoch: 7\tAccuracy: 746/1241 (60%)\n",
      "\n",
      "Train Epoch: 8 \tLoss: 0.803\n",
      "Train Epoch: 8 \tLoss: 0.725\n",
      "Train Epoch: 8 \tLoss: 0.748\n",
      "\n",
      "Test Epoch: 8\tAccuracy: 908/1241 (73%)\n",
      "\n",
      "Train Epoch: 9 \tLoss: 0.667\n",
      "Train Epoch: 9 \tLoss: 0.781\n",
      "Train Epoch: 9 \tLoss: 0.573\n",
      "\n",
      "Test Epoch: 9\tAccuracy: 900/1241 (73%)\n",
      "\n",
      "Train Epoch: 10 \tLoss: 0.720\n",
      "Train Epoch: 10 \tLoss: 0.548\n",
      "Train Epoch: 10 \tLoss: 0.774\n",
      "\n",
      "Test Epoch: 10\tAccuracy: 919/1241 (74%)\n",
      "\n",
      "Train Epoch: 11 \tLoss: 0.703\n",
      "Train Epoch: 11 \tLoss: 0.641\n",
      "Train Epoch: 11 \tLoss: 0.657\n",
      "\n",
      "Test Epoch: 11\tAccuracy: 875/1241 (71%)\n",
      "\n",
      "Train Epoch: 12 \tLoss: 0.818\n",
      "Train Epoch: 12 \tLoss: 0.709\n",
      "Train Epoch: 12 \tLoss: 0.516\n",
      "\n",
      "Test Epoch: 12\tAccuracy: 945/1241 (76%)\n",
      "\n",
      "Train Epoch: 13 \tLoss: 0.536\n",
      "Train Epoch: 13 \tLoss: 0.798\n",
      "Train Epoch: 13 \tLoss: 0.563\n",
      "\n",
      "Test Epoch: 13\tAccuracy: 972/1241 (78%)\n",
      "\n",
      "Train Epoch: 14 \tLoss: 0.747\n",
      "Train Epoch: 14 \tLoss: 0.730\n",
      "Train Epoch: 14 \tLoss: 0.634\n",
      "\n",
      "Test Epoch: 14\tAccuracy: 927/1241 (75%)\n",
      "\n",
      "Train Epoch: 15 \tLoss: 0.605\n",
      "Train Epoch: 15 \tLoss: 0.676\n",
      "Train Epoch: 15 \tLoss: 0.661\n",
      "\n",
      "Test Epoch: 15\tAccuracy: 691/1241 (56%)\n",
      "\n",
      "Train Epoch: 16 \tLoss: 0.917\n",
      "Train Epoch: 16 \tLoss: 0.675\n",
      "Train Epoch: 16 \tLoss: 0.573\n",
      "\n",
      "Test Epoch: 16\tAccuracy: 958/1241 (77%)\n",
      "\n",
      "Train Epoch: 17 \tLoss: 0.479\n",
      "Train Epoch: 17 \tLoss: 0.737\n",
      "Train Epoch: 17 \tLoss: 0.440\n",
      "\n",
      "Test Epoch: 17\tAccuracy: 985/1241 (79%)\n",
      "\n",
      "Train Epoch: 18 \tLoss: 0.493\n",
      "Train Epoch: 18 \tLoss: 0.554\n",
      "Train Epoch: 18 \tLoss: 0.345\n",
      "\n",
      "Test Epoch: 18\tAccuracy: 979/1241 (79%)\n",
      "\n",
      "Train Epoch: 19 \tLoss: 0.428\n",
      "Train Epoch: 19 \tLoss: 0.565\n",
      "Train Epoch: 19 \tLoss: 0.629\n",
      "\n",
      "Test Epoch: 19\tAccuracy: 995/1241 (80%)\n",
      "\n",
      "Train Epoch: 20 \tLoss: 0.914\n",
      "Train Epoch: 20 \tLoss: 0.552\n",
      "Train Epoch: 20 \tLoss: 1.239\n",
      "\n",
      "Test Epoch: 20\tAccuracy: 985/1241 (79%)\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'training loss')"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "log_interval = 50\n",
    "n_epoch = 20\n",
    "\n",
    "model = CNN(n_input=1, n_output=3, n_channel=[8, 8], kernel_sizes = [80, 3])\n",
    "model.to(device)\n",
    "print(model)\n",
    "\n",
    "def count_parameters(model):\n",
    "    return sum(p.numel() for p in model.parameters() if p.requires_grad)\n",
    "\n",
    "optimizer = optim.SGD(model.parameters(), lr=0.03)\n",
    "\n",
    "losses = []\n",
    "test_acc = []\n",
    "\n",
    "for epoch in range(1, n_epoch + 1):\n",
    "    train(model, epoch, log_interval)\n",
    "    test_acc.append(test(model, epoch))\n",
    "\n",
    "# Let's plot the training loss versus the number of iteration.\n",
    "plt.plot(losses)\n",
    "plt.title(\"training loss\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f7008215930>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(test_acc, label = 'test acc')\n",
    "plt.legend()"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "01fa574d3c0e3eceb65f8ef9ae5ed31808805181d1762be2f8a6cdf4ab51f08c"
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}