13_cnn_image_classification.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "709a0612-ff8f-49b0-9f49-e01a8fc55d99",
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow.keras import datasets, layers, models\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "99a90151-3968-4217-9602-70c5ddebb27d",
   "metadata": {},
   "source": [
    "# Load the Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "c736ec0c-fb08-4d47-ac78-1ff52e136a4c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50000, 32, 32, 3)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(X_train, y_train), (X_test, y_test) = datasets.cifar10.load_data()\n",
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c900f22d-4324-4360-9e2e-4294627a0c0c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 32, 32, 3)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "bce1f120-264b-4149-8b55-5279af808517",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[6],\n",
       "       [9],\n",
       "       [9],\n",
       "       [4],\n",
       "       [1]], dtype=uint8)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "d646bcaa-e3b7-476a-8742-3987a29d1106",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([6, 9, 9, 4, 1], dtype=uint8)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train = y_train.reshape(-1,)\n",
    "y_train[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "aefde1a4-059b-47c7-9281-04e066d20e94",
   "metadata": {},
   "outputs": [],
   "source": [
    "classes = [\"airplane\",\"automobile\",\"bird\",\"cat\",\"deer\",\"dog\",\"frog\",\"horse\",\"ship\",\"truck\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "a46308e5-f850-49b7-a07b-3a73bf7a4ea6",
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_sample(X, y, index):\n",
    "    plt.figure(figsize=(15,2))\n",
    "    plt.imshow(X[index])\n",
    "    plt.xlabel(classes[y[index]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "ecd40d92-7e00-4550-9e43-ef8f4c77b321",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMkAAADcCAYAAADa3YUtAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8o6BhiAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAigElEQVR4nO2dfXBU5fn3r7NvZ1+y2bzubgJ5AwKCEUXejPJIqg/xoWpFf7/5+ciMAzOdDojwNOUPhDpTU6YmYFsrM1TbOo7a6VD6R7V12qrkVyWiFAV+Iu8gkECQLHkhye5m33fv5w/KJme/Nx6iCVn0+szsTM61955z79lc55zrvt4UIYQghmGuimG8J8Aw2Q4rCcPowErCMDqwkjCMDqwkDKMDKwnD6MBKwjA6sJIwjA6sJAyjAysJw+gwZkry4osvUlVVFVmtVpo9ezbt2rVrrA7FMGOKaSx2+qc//YkaGhroxRdfpLvuuot++9vf0uLFi+no0aNUXl7+pZ9NpVJ04cIFcjqdpCjKWEyPYUgIQYFAgEpLS8lg0LlXiDFg3rx5YuXKlRrZTTfdJNavX6/72Y6ODkFE/OLXdXl1dHTo/k+O+p0kFovR/v37af369Rp5fX097d69G8ZHo1GKRqPpbfHvoOTNr2wjq92ell84eQA+23P2BMiSSe1Xck+cCmMmVk0DWZ5nIsisNjw9p459DLJzZw5rthPBQRhjTOK+nHm5IDOpdpDNvuNOkE2aov1eEX8fjDl29CDIUqkYyOKJCMiOHzsKssBAL8iisahmOxE3wpi+S2GQBUN4zEQS51ZUlA+yvHyHZjslgrivBIgoEhbpv+PxBLW8+wE5nU4cmMGoK0lPTw8lk0nyeDwaucfjIZ/PB+Obm5vppz/9KcitdjvZ7EMnQ7VaYYzFYgFZppLIPmez4z+i3ZGDc5AoidVmA5mqqpptQywOY2RKkvk5IiKTFWV2hwNkORk/rimFx7Tbca6pFP4Tx+L4WKuqeG6jFjPIBKU02wrh/k0mnJvJJPnXU5IgMptxnCVjHknJtV72pJ5MCMk4/Uf6MTPcMw8uhJBOaMOGDTQwMJB+dXR0jNWUGOYrMep3kqKiIjIajXDX6OrqgrsL0eWrqeyKyjDZwqgricViodmzZ1NLSws9/PDDaXlLSws99NBD17yfQH8fxYfZKoV5BTBGFKPSCZP2Ob+kfBKMSUoeTQypEMhSIXywjfThc7kIa5+vJxS5YUx52RSQlU2pAFnpBLSN3G78nmaz9sKSyMNHyLKJXpAlEvjcH4mgzdDfh8/5PT2XQGayZDzOKvi4lV+IF0GrA485ILGrVCv+i6aE9ncxm3D//oF+kMWiQ49bibjEaLkKY7IEvHbtWnr88cdpzpw5VFtbS7/73e/o3LlztHLlyrE4HMOMKWOiJI8++ij19vbSxo0bqbOzk2pqaugf//gHVVTglZNhsp0xURIiolWrVtGqVavGavcMc93g2C2G0WHM7iRfm3icaNj6eiyKxnYohEZo5dQJmu3gIDr2YnF0ZBUUuUBmMuM1pLoanZN33jFHsz1B4ph0uYpBFjehX8Au8ZOYcHmflAxvWXgQDe1oXOI7saGBn5+HCw2TJ80A2bFj6LwlRXuMaBQXQFy56BA0oxuGBvwXQSYIf+NUSntC+vrwNw6HoiAbXjwrkbx2w53vJAyjAysJw+jASsIwOmStTZKIRCgxLIxFSeDzu2rB2KSBnh7NdqEX7YPym9Gx5y4rBZlZ9uCcwOf8zADB453ocAyd6cbPGfB5+8Shz0A2dzraB3fPm6vZFpJqtX7/AMjOnb0AMotZFheHwZdFxRNAdq7jc+3nrGjzBMNoM/j9PSAzmTFsKTcX9xcOa+0emXmRSKRApolHk9h5V4PvJAyjAysJw+jASsIwOrCSMIwOWWu4R8MhUsSQ8ZVjQ+MytwAddLffeptmu2xSNYwJSNLWTpzBPBZ/CB1jwf5+kPX2aw31Th9Gs+ZKnIlkQIfX3/70Z5CZ/wuvZQtrF2jHmHFBwevFxQgSaDD39wVA9j+fYlajyYyOTodTa+AnkmgRx4L9IDNKLs/FxRjpnZRkK/Ze0n4HA6FxL0vqyssbchjHJY7Wq8F3EobRgZWEYXRgJWEYHVhJGEaHrDXcVdVEqjpUFSNuxNIvYRtWOGnza9NCD3z4CYy51IsRs19cwAhUsxE9wGYDenKjGSmxkQgamyXFeKq7fGdBliupUhLo94PsZFubdv8lRThXSaWRkjJM6S2VyM75cCHjxCGUuUu0CxLt53BhgOJ4zlIxlCUlUdFWCy4WqCZttZRwBD+Xmysp1zQszVekrv3+wHcShtGBlYRhdGAlYRgdWEkYRoesNdxtNjfZhqWadvWjl/yUpNrj0SPaurwGifGalKQChwOS+r0SIz0cRSO6P6CVBSSptO3nj4HMYcPFiGmTsU4xSWplfbRrp2a7oqoKxkydhqnGhYWYpiyrbeXKRYPZkMDQ+8Go9jorS5sN96NHP5nEFGqrDcuoBv342dwML79qlZRulZSaDQ2LoIiPoO4W30kYRgdWEobRgZWEYXRgJWEYHbLWcM/LL9T0JznVcRLGdLa3gcxu1hqOA4MYth70d4FMSaGR3h9AA7w/jAanSdUanEUerGNlc6LBPKHyVpCVSYzQts/+BTKjojXm40n0Onf3YK79LbdMB9mUaiwqXlaCof05d8wC2cHj5zTb0QimNETNEo87oUc8sxA2EZHPJ8nJz+hC4MrH800kqcUVHorGYMOdYUYRVhKG0YGVhGF0yFqbpK1tv6bf4fHTp2DMhc7TIEtmOAWdLuw3OK26EmQ102tA1tmNjWbOduOzbrFX22SnYjI69pyF+Nx8UVLDVvSgnXXu7DmQdWekDEtKc9GiqWh/DAbxO6XQnCERQwfmkT1oG1VPu02z7ZmQB2P2fPIByHwX0SkrsxMiYZxHX0a6sS0Hj5kSaAcNhobOd0JSx+1q8J2EYXRgJWEYHVhJGEYHVhKG0SFrDfe9H71PpmERvCYPRsdOnn4LyGwZaaHTZ2DdrWlTsYh2MoJOPGFAI3eQZIWetQ40ozEPxsQTGFU7GMButq4YGq+yWlbnurROUmvOF7gvSfOcSZMrQSYk18pwP9YcO/7xAfxsWHu+a+77PzDmlpnorAzvQ8P99Kl2kNntmKLtyivMkKAR7pd08h3eYIgNd4YZRVhJGEaHESvJBx98QA8++CCVlpaSoij0l7/8RfO+EIIaGxuptLSUbDYb1dXV0ZEjR0Zrvgxz3RmxkgwODtKtt95KW7dulb7/3HPP0fPPP09bt26lvXv3ktfrpUWLFlEggBlmDHMjMGLDffHixbR48WLpe0IIeuGFF+jpp5+mRx55hIiIXn/9dfJ4PLRt2zZasWLFNR+n+4teMhqHjOlZt94PY1QVI1ULMuzvklKMNr0kSSftOIVGdCwlSWFV0OAzmrTGa1JgCislZGnEuDAgkugpznFhTa3eoNZbb7BgZEFK0v1K2uIJD0k5VjxvlaVlILMatfszEEZO31KDEQh5eXkgeyu8A2S+TjTAJ7i1hcCTCkZmy2qO+f1DiwWXvfsYWS5jVG2StrY28vl8VF9fn5apqkoLFy6k3bt3Sz8TjUbJ7/drXgyTTYyqkvh8PiIi8ni0sUwejyf9XibNzc3kcrnSr7IyvFoxzHgyJqtbiqItDyqEANkVNmzYQAMDA+lXh6QCCsOMJ6PqTPR6L9eU9fl8VFJSkpZ3dXXB3eUKqqqSquKzP8NkC6OqJFVVVeT1eqmlpYVmzbqc6hmLxai1tZU2b948on3ZHPmabkVmib3Z349puGpBnmY7JGlVHEE7j2z5WANLTUnufpLizCLjLEbi6K222vBUGxQMA08ZcFxOIXassgjtQoPRht51YcEogpSCc1OSaPQbjDgPswOLedtytLJEFBdFer/AYuSFDlx0eei794Fs32ftIAtmhM9Hotj+OxrGRZE8Z176b1ldrqsxYiUJBoN06tRQbkdbWxsdOHCACgoKqLy8nBoaGqipqYmqq6upurqampqayG6309KlS0d6KIbJCkasJPv27aPvfOc76e21a9cSEdGyZcvotddeo3Xr1lE4HKZVq1ZRX18fzZ8/n3bs2EFOJ16pGeZGYMRKUldXR0K6/n4ZRVGosbGRGhsbv868GCZr4NgthtEha0PlvWUVZDYPGYWKAfU5EkHH40W/9itZ8tBbHU+gAaqYsVhzOIje47jAeQzvoERElDDiap1d0nnJXdgPMnEJDc6YJPdbyejUZLPZYIwB7XZpbaukpGaXwSxJHZD0lQ4Oag11Wf0yVfLb+bvRmLfZsUX13bUzQXbitLZD2OGj6IML+rF+gGVYSgPX3WKYUYSVhGF0YCVhGB1YSRhGh6w13IViJKEMGY8yQyskyVFRMwzYgF8SAh/BUPaQpKOSWeJwdzrQKC/O1xqcuQXowS7OQ8M6acIi2mEVv+elCvS4R5OdWoHEy5+UdMhKSaIIkpKOXorEcM8rQK9+Kqk9blLyO7lc+N0tCroR+gP9IBNxXDy5bbq2pXaeE3+Tv/0Nw+67Lw7VJ+Acd4YZRVhJGEYHVhKG0SFrbRJKxIiGPT6bUvh87cJ+MVTm0j5z3zQpD8bkWPEZ2ajg9WLQ3w+ySAg70Noc2ojSadXoFCurwFpfBnMFyIL9eMyyYWkH6WO0aSOgcwvwZBTkowPTZEJHakoSZSQkjkirww6yRERrgxgk+zLLHMGEdmFhEdbYCobQ1hrs1zoPJxRjRPGSB+tB9pe//3f6b3YmMswowkrCMDqwkjCMDqwkDKND1hrud827jWzDDOxJM7BT7YUvsEj0hFKt0Ty1ejKM8RZj1ymjQCdbQOLcikqcdopB+9kcBzoTc3LQsDZacAHBLFmgCA9ieurtNVqjv3JqJYyJpzBFVVYcO5FCI1YY8XwYJbWs4hGtpZ6SGMQGEx5TsUo8tZJx0Th+B5NRG7GdjPXDmGLJIsCC/zU3/Xc4EqU333of5yCB7yQMowMrCcPowErCMDqwkjCMDllruM+6eSo5hhnAN89Cwz1cg0a5w6X1MktqQZOQVJM0GDF9t8DhBZkkexeuNClJCmtC5uGVGKVRSRHtyVPKQWbLKJAdHsRIACGp4UUKyoQkIldWbDspOW+pDHd9TFLvKpmS1PUySX4DyTU70IsLJWfbtFU+71owC8aE4hjVbR+2WKBIFmquBt9JGEYHVhKG0YGVhGF0YCVhGB2y1nC3OhxkG2a451gxRdNhl0zfpI3xloWBy9pAGGRGqUADPBWXyDKMXFmNsIRkCcEgsR2FJGQ/Jw9D7xMZHbGSKVmRLTyAkLRzNsgmkpSk+ZpwcUNkds6SpAwrKTymKpmvOYnf3SFrHX5RuzjQfQZreE2chqkJPYZhqcCymP6rwHcShtGBlYRhdGAlYRgdWEkYRoesNdxzcvPJmTMU7iwkHvFQFI1EEdXmTkclYwaDWEw5Fsdx0Sh6xBOSzlnxDM95XLKvkCRXOzSIXuGExFvvLMD6XE5XnmY7z4mFwa0WzGdPSkLxSZGEtxPKnE4M9+/tyug6FcY6WakU1utSSJJrn8S891xJTa2Kcm1rwXAIf08hCf93OYcWgsxGyULHVeA7CcPowErCMDqwkjCMDllrk/z9Hy1ktQ49AyfNu2BMXx86kYIDPZptmc9IZqdcvIj7Sko8kQWS1N/8okLNtirpXDt4qR9kJz8/BjK/pHFQWRXW5zJmNB3KdRbCmKoqjB6eWIaRzVWTJoCsQEVnotOKdmEqI+qaJM/68STaB0ZJqq5RckxPpcTWytXaKXGBzkojmjxUUDA0V1XF73I1+E7CMDqwkjCMDiNSkubmZpo7dy45nU5yu920ZMkSOnHihGaMEIIaGxuptLSUbDYb1dXV0ZEjR0Z10gxzPRmRkrS2ttKTTz5Je/bsoZaWFkokElRfX0+Dg0Pr1M899xw9//zztHXrVtq7dy95vV5atGgRBSS9RBjmRkARX9aUXYfu7m5yu93U2tpKd999NwkhqLS0lBoaGuipp54iIqJoNEoej4c2b95MK1as0N2n3+8nl8tF99z7MJmGRZ3mTZwGY0USjdxPd2trKVVMxGjQokI0ck+fPg2yhCR6derMGSArLNE6t/q+wEWAe+fVgky2qhCKRnCYpN5V2zltB9qTn+P8e3p7QJbnwnpU//GfD4Psrpungqz/c1w8UTNqe8UkhrsxXxJlLEkZlqXvqmZ0YCYzIrENkqjrlBEXZxI0VPA7OBimuu+to4GBAcqVdEbWzutrMDBwOa+6oOByKHdbWxv5fD6qrx+q6K2qKi1cuJB2794t3Uc0GiW/3695MUw28ZWVRAhBa9eupQULFlBNTQ0REfl8l0viezzaK6vH40m/l0lzczO5XK70q6ys7KtOiWHGhK+sJKtXr6aDBw/SH//4R3gvM6lJCCFNdCIi2rBhAw0MDKRfHR0d0nEMM158JWfimjVr6K233qIPPviAJg575vd6LzuqfD4flQxrPNPV1QV3lyuoqkqqikFsDJMtjEhJhBC0Zs0aevPNN2nnzp1UVVWleb+qqoq8Xi+1tLTQrFmXayHFYjFqbW2lzZs3j2hiS/7zMbLZhgwt1V0NY0IBfIT7/NBnmu0SLz6+yQw9mxWNt1gKa0hNrcF55JdovfChIox6fWDx/waZ3YkFswclhrskC5cSGanFkQR+rqsLOw+fbbuA87Djd/ed7wVZ+5HPQWaIaI97xtcFY+bVzwFZRSV2FJZ55g1WievcrF1QUSQRv6TgootFGTpnFvO1r1eNSEmefPJJ2rZtG/31r38lp9OZtjNcLhfZbDZSFIUaGhqoqamJqqurqbq6mpqamshut9PSpUtHciiGyRpGpCQvvfQSERHV1dVp5K+++iotX76ciIjWrVtH4XCYVq1aRX19fTR//nzasWMHOZ3OUZkww1xvRvy4pYeiKNTY2EiNjY1fdU4Mk1Vw7BbD6JC1ofKq2UCqZUiHTx4/DGP8A2i4Z97t4jH0vAYl6buyJWqrJJw6HsLwmoFu7TEvnsNl7LfffRtkfZJQnYEgFr52SjzCrnxtLS5HLq4Qnj+PRrq7CMPirbkY/r/r7zjfS58fBFkypvW4n/JhtMF5SZpy9XRcAHHlYgtsVz6mLtvsWi+8y4G/k9mKnn+7fegcxSRp2FeD7yQMowMrCcPowErCMDqwkjCMDllruAcuXaREeMgj/d5f/w5jOnznQWaIa73kBw9KooolRnoiIfPaonHX8rf3QGYxa43m22bdDmNiFvQT+aNYi+vMOfRY9/ZiLnwsop3bBV87jGlrx8/NmTUbZP/vybUg+2TPv0CWGEAvvD+jzlk4s4A2EZ3ZhwsZu/Z3gsxhwjpnZosk9D4jjMkpMdwnVlSC7KH/+L/pv0MhLpjNMKMGKwnD6MBKwjA6ZK1N4nV7yG4fqt1aXVkFY4SkMY7JoJUZpZ128dogJDW2LFbsGkuSdNLSUq2Dru6++2CM0y5xlFkxWvjo4c9AdvIUpuZ6J1RqtiOStsBGGx7z8MnjeMyTJ0Fmr5wOsgsXcL75eVqZW1J/2J6D0c6XfGdB1vvFKZB196BzMpLMcBhLwqQ7+/Ff+857h8aFw9x9l2FGDVYShtGBlYRhdGAlYRgdstZw7+vpo4htyFF1x/w7YcydCxeCTFW1zieTxEiX1mmSdNo1kqT4cwzTQsMxrVOw93wbjLkUQUfZpR5Mrz0jMdIvdGG0c447I/1VxQUFxYKGeyyBjXJaWj8EWcXkW0BWViCJIDZo/4XsZoxGjkYwCviMH6t65jgx2jkp0Mnr69PWWysqqoQxIUmX5PdaP0n/LWu0dDX4TsIwOrCSMIwOrCQMowMrCcPokLWGu92ukt02ZAT2+rGu1KcH94PM7dZ6gD1u7JSU2S2XiKivrx8nEcFjmlL42QlVWiO6LB8jfr84iVGvg0E0ot0e7ERlL8wDmTGjTlgojHMtKcFOV74LGDnd04spwyWlkhRnSSGQYGaHYhMa7nFJ4XHVhtEMqiQ6ItbbDTIyaKN+PRnRB0REMVlnZiH/Ww++kzCMDqwkDKMDKwnD6MBKwjA6ZK3hrppSpJqHvKbRSD+M2b37nyATca0Bm2vHMO14HL24kTAWxzZJriEVlViAu+YObferyeVYDLq/Aw1mXx92orLY0PCdXIjGfHe31ut8y7QaGHPzLdgdbPsffg8yE2F4e3wQFwJiMZSJRIZRbpW0o5Z0DaismgSyro4TICMDRj3YHNr9TZ+OXbkiIUmr72GFzaOSwuRXg+8kDKMDKwnD6MBKwjA6sJIwjA5Za7iHImGi4Q5YSXj7fYsfAFkqpvUUGyVGeiqJYdRC1lrZhAat1YHh575+rdEf6Mec8UthnIdixfD2EwfOgKz3X+h1nlSlNcrnTsEC1DGJF95mQSNaSCIQZB58gxH/XTLTy8MpSd0BSQeriolouEeCWNdrRi565j/Z/6lm+8JZNPjDgxgxIEJ96b9jHCrPMKMHKwnD6MBKwjA6sJIwjA5Za7g7HGay24cMZ5cktNlZjJ7WaEYBZ6vkOmBR0CAXNvTMq3Ycl4qgJzcQ0BblNkpaPrsn54Fssh097p+3YY47KbioYLZrDfAvOs/BmEJJq2yZLBZGIzcaxfD5QYkXPprh2Y5LioCbrLjY4SktBtnZTixEd/Ecno9IRjew00cOwJjCQty/GNYdTEhy4K8G30kYRocRKclLL71EM2fOpNzcXMrNzaXa2lp6++2h3npCCGpsbKTS0lKy2WxUV1dHR45gVQyGuZEYkZJMnDiRNm3aRPv27aN9+/bRPffcQw899FBaEZ577jl6/vnnaevWrbR3717yer20aNEiCkgaaDLMjcKIbJIHH3xQs/3ss8/SSy+9RHv27KEZM2bQCy+8QE8//TQ98sgjRET0+uuvk8fjoW3bttGKFStGNLFQ8BRRcpizLYX6bFZyQHbxovZ59fOj7TDGakL7w+LKA1mRG5/fS4uwG6wpw9FZ6CqEMRL/JUXCfSBzu9GemVBaALJOn7YW18mT2LCnMoZFxjNtNiKiQADtj1AI7QP/ADZEyrRJkjGMpjaq6BA8chjTqmUpt263B2QTZmojnt3FOKaoGCOnrcPmEbkeUcDJZJK2b99Og4ODVFtbS21tbeTz+ai+vj49RlVVWrhwIe3evfurHoZhxp0Rr24dOnSIamtrKRKJUE5ODr355ps0Y8aMtCJ4PFqt9ng8dPYsltm/QjQa1Vzd/H5J+zaGGUdGfCeZNm0aHThwgPbs2UNPPPEELVu2jI4ePZp+X8moeCGEANlwmpubyeVypV9lZZjUxDDjyYiVxGKx0JQpU2jOnDnU3NxMt956K23ZsoW83svPgL6MZ+Wuri64uwxnw4YNNDAwkH51dGATSoYZT762M1EIQdFolKqqqsjr9VJLSwvNmjWLiIhisRi1trbS5s2br/p5VVVJlaR3iliUUsN8aAaJPpvi6GTLNWst5P17WmGM7yI68RRJoed587BT7YLaOSAbGNAavgf/52MYMyip4XXyHF4QzrS3gywcQgedENq7szUXnWd+P64qBiQpw4N+XECQ3ftNRpS6nFpHYWkVLhbkF5aAzF2KhnXpLCzSXSCJArZkRGwbJRHcMgcsDesGZjJhx96rMSIl+fGPf0yLFy+msrIyCgQCtH37dtq5cye98847pCgKNTQ0UFNTE1VXV1N1dTU1NTWR3W6npUuXjuQwDJNVjEhJLl68SI8//jh1dnaSy+WimTNn0jvvvEOLFi0iIqJ169ZROBymVatWUV9fH82fP5927NhBTidWNGSYG4URKckrr7zype8rikKNjY3U2Nj4debEMFlF1gU4in8XaQ1HtE6vuMQmSQh87oxkfC4pyZRLSQrBKpImPvGEpPSQxBkXzXCCRWPoFIvFMPsvIdl/SjJfIZNl2CQpSb3dFKFMvq9rK4wrG5Y532QSjyn7nrImOjJHZySKv3HK8PVtkivOxGv57oq41jN0nTh//jwvAzPXjY6ODpo4ceKXjsk6JUmlUnThwgVyOp0UCASorKyMOjo6KDcXwzWYscXv939jz78QggKBAJWWlkrbAw4n6x63DAZDWrOvOCGvRB0z48M39fy7XBiHJ4PzSRhGB1YShtEhq5VEVVV65plnpB55Zuzh83+ZrDPcGSbbyOo7CcNkA6wkDKMDKwnD6MBKwjA6ZK2SvPjii1RVVUVWq5Vmz55Nu3btGu8pfSNpbm6muXPnktPpJLfbTUuWLKETJ7RV2r/1paJEFrJ9+3ZhNpvFyy+/LI4ePSp++MMfCofDIc6ePTveU/vGcd9994lXX31VHD58WBw4cEDcf//9ory8XASDwfSYTZs2CafTKf785z+LQ4cOiUcffVSUlJQIv98/jjO/fmSlksybN0+sXLlSI7vpppvE+vXrx2lG3x66uroEEYnW1lYhhBCpVEp4vV6xadOm9JhIJCJcLpf4zW9+M17TvK5k3eNWLBaj/fv3a0oTERHV19dzaaLrwJVU5IKCy7W+uFRUFtokPT09lEwmpaWJMotMMKOLEILWrl1LCxYsoJqaywXgrpzzb/PvkXVRwFcYaWki5uuzevVqOnjwIH344Yfw3rf598i6O0lRUREZjcYRlyZivh5r1qyht956i95//31NEtJXLRX1TSLrlMRisdDs2bOppaVFI29paaE777xznGb1zUUIQatXr6Y33niD3nvvParKKAk0vFTUFa6UivrW/B7ju24g58oS8CuvvCKOHj0qGhoahMPhEO3t7eM9tW8cTzzxhHC5XGLnzp2is7Mz/QqFQukxmzZtEi6XS7zxxhvi0KFD4rHHHuMl4Gzg17/+taioqBAWi0Xcfvvt6SVJZnQhIunr1VdfTY9JpVLimWeeEV6vV6iqKu6++25x6NCh8Zv0dYZD5RlGh6yzSRgm22AlYRgdWEkYRgdWEobRgZWEYXRgJWEYHVhJGEYHVpJvGe3t7aQoCh04cGC8p3LDwEqSJdTV1VFDQ8N4T4ORwEpygyCEkPb5YMYeVpIsYPny5dTa2kpbtmwhRVFIURR67bXXSFEUevfdd2nOnDmkqirt2rWLli9fTkuWLNF8vqGhgerq6tLbqVSKNm/eTFOmTCFVVam8vJyeffZZ6bFTqRT94Ac/oKlTp9LZs2fH8FveuGRt0tW3iS1bttDJkyeppqaGNm7cSESUrkaybt06+sUvfkGTJk2ivLy8a9rfhg0b6OWXX6Zf/epXtGDBAurs7KTjx4/DuFgsRkuXLqXTp0/Thx9+SG63e9S+0zcJVpIswOVykcViIbvdnk5yuvJPvXHjxnTj1mshEAjQli1baOvWrbRs2TIiIpo8eTItWLBAMy4YDNL9999P4XCYdu7cec29Or6N8ONWljNnDvaN/zKOHTtG0WiU7r333i8d99hjj1EwGKQdO3awgujASpLlOBwOzbbBYIBmmPH4UNNSm812Tfv97ne/SwcPHqQ9e/Z8/Ul+w2ElyRIsFou0c20mxcXF1NnZqZEN93lUV1eTzWajf/7zn1+6nyeeeII2bdpE3/ve96i1tfUrzfnbAtskWUJlZSV9/PHH1N7eTjk5OdJW1URE99xzD/385z+n3//+91RbW0t/+MMf6PDhwzRr1iwiIrJarfTUU0/RunXryGKx0F133UXd3d105MgR+v73v6/Z15o1ayiZTNIDDzxAb7/9NtgtzL8Z38RI5gonTpwQd9xxh7DZbOn0WSISfX19MPYnP/mJ8Hg8wuVyiR/96Edi9erVYuHChen3k8mk+NnPfiYqKiqE2WwW5eXloqmpSQghRFtbmyAi8emnn6bH//KXvxROp1N89NFHY/wtb0w4fZdhdGCbhGF0YCVhGB1YSRhGB1YShtGBlYRhdGAlYRgdWEkYRgdWEobRgZWEYXRgJWEYHVhJGEYHVhKG0eH/A7rW8sf6IhosAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1500x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_train, y_train, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "c284894e-9ccc-42d1-9651-41cfbcdb8db8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1500x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_train, y_train, 3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cd12f73e-fc7e-4b19-b253-b7db33c3b16a",
   "metadata": {},
   "source": [
    "### Normalizing the Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "bd3cec2a-3eba-430a-90fa-a9baf9ca91a7",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = X_train / 255\n",
    "X_test = X_test / 255"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec3a410f-b9b2-4aad-8a41-7e7e6fc9d8e1",
   "metadata": {},
   "source": [
    "### Build simple artificial neural network for image classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "f7a8eca9-89dc-4c99-91d9-d3b0f94b6ff3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Metal device set to: Apple M1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:15:37.362082: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:305] Could not identify NUMA node of platform GPU ID 0, defaulting to 0. Your kernel may not have been built with NUMA support.\n",
      "2023-01-21 09:15:37.362367: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:271] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 0 MB memory) -> physical PluggableDevice (device: 0, name: METAL, pci bus id: <undefined>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:15:38.593537: W tensorflow/core/platform/profile_utils/cpu_utils.cc:128] Failed to get CPU frequency: 0 Hz\n",
      "2023-01-21 09:15:38.712419: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1563/1563 [==============================] - 21s 13ms/step - loss: 1.8184 - accuracy: 0.3517\n",
      "Epoch 2/5\n",
      "1563/1563 [==============================] - 20s 13ms/step - loss: 1.6264 - accuracy: 0.4235\n",
      "Epoch 3/5\n",
      "1563/1563 [==============================] - 20s 13ms/step - loss: 1.5436 - accuracy: 0.4580\n",
      "Epoch 4/5\n",
      "1563/1563 [==============================] - 20s 13ms/step - loss: 1.4833 - accuracy: 0.4801\n",
      "Epoch 5/5\n",
      "1563/1563 [==============================] - 20s 13ms/step - loss: 1.4323 - accuracy: 0.4965\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x288c6b640>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ann = models.Sequential([\n",
    "    layers.Flatten(input_shape=(32,32,3)),\n",
    "    layers.Dense(3000, activation=\"relu\"),\n",
    "    layers.Dense(1000, activation=\"relu\"),\n",
    "    layers.Dense(10, activation=\"softmax\")\n",
    "])\n",
    "\n",
    "ann.compile(optimizer=\"SGD\",\n",
    "            loss = \"sparse_categorical_crossentropy\",\n",
    "            metrics=[\"accuracy\"])\n",
    "\n",
    "ann.fit(X_train, y_train, epochs=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "f901813c-0f28-4991-b896-37396dec8a49",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 16/313 [>.............................] - ETA: 2s - loss: 1.7663 - accuracy: 0.3848"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:17:35.047222: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 2s 7ms/step - loss: 1.6797 - accuracy: 0.4194\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[1.6796766519546509, 0.41940000653266907]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ann.evaluate(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "2937b156-f8bf-4c7d-a5db-1e12a5151e2d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 32/313 [==>...........................] - ETA: 1s"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:17:54.062161: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 2s 7ms/step\n",
      "Classification Report: \n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "           0       0.45      0.66      0.53      1000\n",
      "           1       0.45      0.74      0.56      1000\n",
      "           2       0.46      0.11      0.18      1000\n",
      "           3       0.49      0.10      0.16      1000\n",
      "           4       0.30      0.55      0.38      1000\n",
      "           5       0.51      0.20      0.29      1000\n",
      "           6       0.62      0.28      0.38      1000\n",
      "           7       0.34      0.74      0.47      1000\n",
      "           8       0.72      0.25      0.37      1000\n",
      "           9       0.46      0.57      0.51      1000\n",
      "\n",
      "    accuracy                           0.42     10000\n",
      "   macro avg       0.48      0.42      0.38     10000\n",
      "weighted avg       0.48      0.42      0.38     10000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix , classification_report\n",
    "import numpy as np\n",
    "y_pred = ann.predict(X_test)\n",
    "y_pred_classes = [np.argmax(element) for element in y_pred]\n",
    "\n",
    "print(\"Classification Report: \\n\", classification_report(y_test, y_pred_classes))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "05dbf45f-a4d9-43df-8cfd-e153c98b3957",
   "metadata": {},
   "source": [
    "# Build a Convolutional Neural Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "6431fc57-009a-41cd-a165-6f7e44482a3b",
   "metadata": {},
   "outputs": [],
   "source": [
    "cnn = models.Sequential([\n",
    "    layers.Conv2D(filters=32, kernel_size=(3,3), activation=\"relu\", input_shape=(32,32,3)),\n",
    "    layers.MaxPooling2D((2,2)),\n",
    "    \n",
    "    layers.Conv2D(filters=64, kernel_size=(3,3), activation=\"relu\"),\n",
    "    layers.MaxPooling2D((2,2)),\n",
    "    \n",
    "    layers.Flatten(),\n",
    "    layers.Dense(64, activation=\"relu\"),\n",
    "    layers.Dense(10, activation=\"softmax\")\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "d0921880-ad9b-4883-88ff-f2bb7676238d",
   "metadata": {},
   "outputs": [],
   "source": [
    "cnn.compile(optimizer=\"adam\",\n",
    "             loss = \"sparse_categorical_crossentropy\",\n",
    "             metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "30334463-e789-4c96-ab0a-44c1bb1be996",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:28:23.803752: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1563/1563 [==============================] - 17s 10ms/step - loss: 1.4735 - accuracy: 0.4699\n",
      "Epoch 2/10\n",
      "1563/1563 [==============================] - 16s 10ms/step - loss: 1.1218 - accuracy: 0.6102\n",
      "Epoch 3/10\n",
      "1563/1563 [==============================] - 16s 10ms/step - loss: 0.9936 - accuracy: 0.6551\n",
      "Epoch 4/10\n",
      "1563/1563 [==============================] - 15s 10ms/step - loss: 0.9079 - accuracy: 0.6851\n",
      "Epoch 5/10\n",
      "1563/1563 [==============================] - 16s 10ms/step - loss: 0.8450 - accuracy: 0.7067\n",
      "Epoch 6/10\n",
      "1563/1563 [==============================] - 16s 10ms/step - loss: 0.7841 - accuracy: 0.7282\n",
      "Epoch 7/10\n",
      "1563/1563 [==============================] - 16s 10ms/step - loss: 0.7373 - accuracy: 0.7433\n",
      "Epoch 8/10\n",
      "1563/1563 [==============================] - 15s 10ms/step - loss: 0.6913 - accuracy: 0.7594\n",
      "Epoch 9/10\n",
      "1563/1563 [==============================] - 16s 10ms/step - loss: 0.6484 - accuracy: 0.7754\n",
      "Epoch 10/10\n",
      "1563/1563 [==============================] - 16s 10ms/step - loss: 0.6109 - accuracy: 0.7872\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x294b2fc70>"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cnn.fit(X_train, y_train, epochs=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "f9031e38-5a9e-452f-89e5-4a2d606cd4b5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 20/313 [>.............................] - ETA: 1s - loss: 0.8980 - accuracy: 0.7188"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:31:00.696631: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 2s 6ms/step - loss: 0.9255 - accuracy: 0.6926\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.9255183935165405, 0.6926000118255615]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cnn.evaluate(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "8433853d-4eb8-43dd-9645-053191d2c408",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([3, 8, 8, 0, 6], dtype=uint8)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test = y_test.reshape(-1,)\n",
    "y_test[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "7ba32db2-fb76-4d5d-a5a9-91b6113a9609",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 51/313 [===>..........................] - ETA: 0s"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:31:02.738776: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 1s 4ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[2.3661340e-03, 1.1850568e-05, 1.6951181e-03, 8.9192492e-01,\n",
       "        3.1797383e-03, 9.9968374e-02, 2.9309970e-04, 2.8469622e-05,\n",
       "        3.3699552e-04, 1.9535908e-04],\n",
       "       [2.5817307e-02, 2.8783934e-02, 4.3217762e-05, 6.0636561e-05,\n",
       "        2.9851715e-06, 8.7217422e-06, 1.4230828e-06, 1.5654150e-06,\n",
       "        9.4463754e-01, 6.4263202e-04],\n",
       "       [2.2975350e-02, 4.0888418e-02, 8.3465833e-04, 3.5239451e-03,\n",
       "        5.4147845e-04, 9.2925265e-04, 5.1116542e-04, 2.3624969e-03,\n",
       "        8.9595377e-01, 3.1479362e-02],\n",
       "       [7.5032026e-01, 6.3317193e-04, 7.7499703e-02, 3.0824985e-02,\n",
       "        6.9867953e-04, 1.5720131e-04, 3.6276015e-03, 7.2679715e-04,\n",
       "        1.3543476e-01, 7.6870107e-05],\n",
       "       [5.4594657e-05, 1.3151638e-05, 7.8944080e-03, 5.2737720e-02,\n",
       "        8.0534071e-01, 3.4515146e-02, 9.9421300e-02, 8.2835477e-06,\n",
       "        8.5888587e-06, 6.1616292e-06]], dtype=float32)"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = cnn.predict(X_test)\n",
    "y_pred[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "b37c6f4c-24bd-46f5-99fc-c984005910d3",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[3, 8, 8, 0, 4]"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_classes = [np.argmax(element) for element in y_pred]\n",
    "y_classes[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "9c7e84c4-d888-4133-b7ca-25722da0a89e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([3, 8, 8, 0, 6], dtype=uint8)"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "daf8eef6-da17-4de5-b720-07fbb3fd8952",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1500x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_sample(X_test, y_test,3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "c8043bb8-badb-466e-a2e1-a969ad756ef5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'airplane'"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes[y_classes[3]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "3a04f5f8-fe1f-4cf1-a941-350c37ce1efa",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report: \n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "           0       0.80      0.68      0.73      1000\n",
      "           1       0.84      0.79      0.82      1000\n",
      "           2       0.64      0.55      0.59      1000\n",
      "           3       0.41      0.68      0.51      1000\n",
      "           4       0.72      0.54      0.62      1000\n",
      "           5       0.60      0.56      0.58      1000\n",
      "           6       0.85      0.69      0.76      1000\n",
      "           7       0.75      0.77      0.76      1000\n",
      "           8       0.79      0.85      0.82      1000\n",
      "           9       0.74      0.83      0.78      1000\n",
      "\n",
      "    accuracy                           0.69     10000\n",
      "   macro avg       0.71      0.69      0.70     10000\n",
      "weighted avg       0.71      0.69      0.70     10000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Classification Report: \\n\", classification_report(y_test, y_classes))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7fb7fa23-9e6e-42d3-8b37-3955f6d1d6f2",
   "metadata": {},
   "source": [
    "# Handwritten Digits Classification using CNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "b08761f4-40c6-4134-a896-cc1ddafe6b89",
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow import keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "7dcd19b9-086d-413c-98d8-03b6992e7c90",
   "metadata": {},
   "outputs": [],
   "source": [
    "(X_train, y_train), (X_test, y_test) = keras.datasets.mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "e795e750-b45f-4988-b3bf-d3190c154029",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(60000, 28, 28)"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "fd772b1f-1d1b-4474-86f1-d2b4da4f74f6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 28, 28)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "85dd3918-1a89-41f6-9150-5a7d5d884245",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x299aa5940>"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaMAAAGkCAYAAACckEpMAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8o6BhiAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAbhklEQVR4nO3df3DU953f8deaH2vgVntVsbSrICs6H5w9FiUNEECHQdCgQx0zxnJSbHcykCaMbQQ3VLi+YDpFl8khH1MYcpFNLlwOwwQOJjcYaKHGSkHCFHAxh2NKfEQ+RJDPklVksytkvCDx6R8qay/C4O96V2/t6vmY+U7Y7/f71vfNJ1/75Y/2u5/1OeecAAAwdJd1AwAAEEYAAHOEEQDAHGEEADBHGAEAzBFGAABzhBEAwBxhBAAwRxgBAMwRRgAAcxkVRi+99JKKi4t19913a+LEiXr99detW+pXNTU18vl8CVsoFLJuq18cPnxY8+bNU0FBgXw+n3bv3p1w3DmnmpoaFRQUaMSIESorK9OZM2dsmk2jO43DokWL+twjU6dOtWk2jWprazV58mQFAgHl5eVp/vz5Onv2bMI5g+Ge+CLjkCn3RMaE0c6dO7V8+XKtWrVKp06d0kMPPaSKigpduHDBurV+9eCDD6q1tTW+nT592rqlftHV1aUJEyaorq7ulsfXrl2r9evXq66uTidOnFAoFNKcOXPU2dnZz52m153GQZLmzp2bcI/s37+/HzvsH42NjaqqqtLx48dVX1+v7u5ulZeXq6urK37OYLgnvsg4SBlyT7gM8Y1vfMM9/fTTCfvuv/9+94Mf/MCoo/63evVqN2HCBOs2zElyr7zySvz19evXXSgUci+88EJ83yeffOKCwaD76U9/atBh/7h5HJxzbuHChe6RRx4x6cdSe3u7k+QaGxudc4P3nrh5HJzLnHsiI2ZGV69e1cmTJ1VeXp6wv7y8XEePHjXqykZTU5MKCgpUXFysxx9/XOfOnbNuyVxzc7Pa2toS7g+/36+ZM2cOuvtDkhoaGpSXl6dx48Zp8eLFam9vt24p7SKRiCQpNzdX0uC9J24ehxsy4Z7IiDC6ePGienp6lJ+fn7A/Pz9fbW1tRl31vylTpmjr1q06cOCANm3apLa2NpWWlqqjo8O6NVM37oHBfn9IUkVFhbZt26aDBw9q3bp1OnHihGbPnq1YLGbdWto451RdXa3p06erpKRE0uC8J241DlLm3BNDrRvwwufzJbx2zvXZl80qKirifx4/frymTZum++67T1u2bFF1dbVhZwPDYL8/JGnBggXxP5eUlGjSpEkqKirSvn37VFlZadhZ+ixdulRvv/22jhw50ufYYLonPm8cMuWeyIiZ0ejRozVkyJA+/0XT3t7e5798BpNRo0Zp/Pjxampqsm7F1I0nCrk/+gqHwyoqKsrae2TZsmXau3evDh06pDFjxsT3D7Z74vPG4VYG6j2REWE0fPhwTZw4UfX19Qn76+vrVVpaatSVvVgspnfeeUfhcNi6FVPFxcUKhUIJ98fVq1fV2Ng4qO8PSero6FBLS0vW3SPOOS1dulS7du3SwYMHVVxcnHB8sNwTdxqHWxmw94ThwxOe7Nixww0bNsz9/Oc/d7/5zW/c8uXL3ahRo9z58+etW+s3K1ascA0NDe7cuXPu+PHj7uGHH3aBQGBQjEFnZ6c7deqUO3XqlJPk1q9f706dOuV+97vfOeece+GFF1wwGHS7du1yp0+fdk888YQLh8MuGo0ad55atxuHzs5Ot2LFCnf06FHX3NzsDh065KZNm+a+8pWvZN04PPPMMy4YDLqGhgbX2toa3z7++OP4OYPhnrjTOGTSPZExYeSccy+++KIrKipyw4cPd1//+tcTHl8cDBYsWODC4bAbNmyYKygocJWVle7MmTPWbfWLQ4cOOUl9toULFzrneh/lXb16tQuFQs7v97sZM2a406dP2zadBrcbh48//tiVl5e7e+65xw0bNszde++9buHChe7ChQvWbafcrcZAktu8eXP8nMFwT9xpHDLpnvA551z/zcMAAOgrI94zAgBkN8IIAGCOMAIAmCOMAADmCCMAgDnCCABgLqPCKBaLqaamZsAt8GeBsejFOPRiHD7FWPTKtHHIqM8ZRaNRBYNBRSIR5eTkWLdjirHoxTj0Yhw+xVj0yrRxyKiZEQAgOxFGAABzA+77jK5fv673339fgUCgz/eORKPRhP8dzBiLXoxDL8bhU4xFr4EwDs45dXZ2qqCgQHfddfu5z4B7z+i9995TYWGhdRsAgBRpaWm54/csDbiZUSAQkCRN17/VUA0z7gYAkKxuXdMR7Y//e/12BlwY3fjV3FAN01AfYQQAGev//97ti3zVe9oeYHjppZdUXFysu+++WxMnTtTrr7+erksBADJcWsJo586dWr58uVatWqVTp07poYceUkVFhS5cuJCOywEAMlxawmj9+vX63ve+p+9///t64IEHtGHDBhUWFmrjxo3puBwAIMOlPIyuXr2qkydPqry8PGF/eXm5jh492uf8WCymaDSasAEABpeUh9HFixfV09Oj/Pz8hP35+flqa2vrc35tba2CwWB847FuABh80vYAw81PTzjnbvlExcqVKxWJROJbS0tLuloCAAxQKX+0e/To0RoyZEifWVB7e3uf2ZIk+f1++f3+VLcBAMggKZ8ZDR8+XBMnTlR9fX3C/vr6epWWlqb6cgCALJCWD71WV1frO9/5jiZNmqRp06bpZz/7mS5cuKCnn346HZcDAGS4tITRggUL1NHRoR/+8IdqbW1VSUmJ9u/fr6KionRcDgCQ4QbcQqk3vhCqTI+wHBAAZLBud00N2vOFvuCP7zMCAJgjjAAA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5ggjAIA5wggAYG6odQPAQOIbmtw/EkPuGZ3iTlLr7LNf9VzTM/K655qi+9o914xc4vNcI0lt64d7rvmHSTs911zs6fJcI0lTfrnCc80fVh9P6lrZgJkRAMAcYQQAMJfyMKqpqZHP50vYQqFQqi8DAMgiaXnP6MEHH9SvfvWr+OshQ4ak4zIAgCyRljAaOnQosyEAwBeWlveMmpqaVFBQoOLiYj3++OM6d+7c554bi8UUjUYTNgDA4JLyMJoyZYq2bt2qAwcOaNOmTWpra1Npaak6OjpueX5tba2CwWB8KywsTHVLAIABLuVhVFFRoccee0zjx4/XN7/5Te3bt0+StGXLlluev3LlSkUikfjW0tKS6pYAAANc2j/0OmrUKI0fP15NTU23PO73++X3+9PdBgBgAEv754xisZjeeecdhcPhdF8KAJChUh5Gzz77rBobG9Xc3Kw33nhD3/rWtxSNRrVw4cJUXwoAkCVS/mu69957T0888YQuXryoe+65R1OnTtXx48dVVFSU6ksBALJEysNox44dqf6RAIAsx6rdSNqQB8YmVef8wzzXvD/z9z3XXJnqfbXl3GByKzS/PsH7atDZ6H98HPBc85d1c5O61hvjt3uuab52xXPNCx/M8VwjSQWvu6TqBisWSgUAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOhVIhSeop+7rnmvUvv5jUtcYNG55UHfrXNdfjuea//GSR55qhXcktKDrtl0s91wT+udtzjf+i98VVJWnkm28kVTdYMTMCAJgjjAAA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjoVSIUnyn33fc83JTwqTuta4YR8kVZdtVrRO9Vxz7vLopK718n1/77kmct37Aqb5f3XUc81Al9wyrvCKmREAwBxhBAAwRxgBAMwRRgAAc4QRAMAcYQQAMEcYAQDMEUYAAHOEEQDAHGEEADBHGAEAzBFGAABzhBEAwByrdkOS1N3a5rnmJ3/57aSu9RdzuzzXDHn79zzX/HrJTzzXJOtHF/+V55p3vznSc03PpVbPNZL05LQlnmvO/6n36xTr196LADEzAgAMAIQRAMAcYQQAMEcYAQDMEUYAAHOEEQDAHGEEADBHGAEAzBFGAABzhBEAwBxhBAAwRxgBAMyxUCqSlrv5WFJ19/y3f+m5pqfjQ881D5b8B881Z2b8recaSdr7s5mea/IuHU3qWsnwHfO+gGlxcv/3AklhZgQAMEcYAQDMeQ6jw4cPa968eSooKJDP59Pu3bsTjjvnVFNTo4KCAo0YMUJlZWU6c+ZMqvoFAGQhz2HU1dWlCRMmqK6u7pbH165dq/Xr16uurk4nTpxQKBTSnDlz1NnZ+aWbBQBkJ88PMFRUVKiiouKWx5xz2rBhg1atWqXKykpJ0pYtW5Sfn6/t27frqaee+nLdAgCyUkrfM2publZbW5vKy8vj+/x+v2bOnKmjR2/95FAsFlM0Gk3YAACDS0rDqK2tTZKUn5+fsD8/Pz9+7Ga1tbUKBoPxrbCwMJUtAQAyQFqepvP5fAmvnXN99t2wcuVKRSKR+NbS0pKOlgAAA1hKP/QaCoUk9c6QwuFwfH97e3uf2dINfr9ffr8/lW0AADJMSmdGxcXFCoVCqq+vj++7evWqGhsbVVpamspLAQCyiOeZ0eXLl/Xuu+/GXzc3N+utt95Sbm6u7r33Xi1fvlxr1qzR2LFjNXbsWK1Zs0YjR47Uk08+mdLGAQDZw3MYvfnmm5o1a1b8dXV1tSRp4cKFevnll/Xcc8/pypUrWrJkiT766CNNmTJFr732mgKBQOq6BgBkFZ9zzlk38VnRaFTBYFBlekRDfcOs20EG++1fT/Ze8/BPk7rWd3/3bzzX/N/pSXwQ/HqP9xrASLe7pgbtUSQSUU5Ozm3PZW06AIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5lL65XrAQPLAn/3Wc813x3tf8FSSNhf9T881M79d5bkmsPO45xogEzAzAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5ggjAIA5wggAYI4wAgCYY9VuZK2eSxHPNR3PPJDUtS7sveK55gc/2uq5ZuW/e9RzjSS5U0HPNYV/cSyJCznvNYCYGQEABgDCCABgjjACAJgjjAAA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmWCgV+Izrv34nqbrH//w/ea7Ztvq/eq55a6r3xVUlSVO9lzw4aqnnmrGbWj3XdJ8777kG2YeZEQDAHGEEADBHGAEAzBFGAABzhBEAwBxhBAAwRxgBAMwRRgAAc4QRAMAcYQQAMEcYAQDMEUYAAHM+55yzbuKzotGogsGgyvSIhvqGWbcDpI374695rsl54b2krvV3f3AgqTqv7j/0fc81f/TnkaSu1dN0Lqk69J9ud00N2qNIJKKcnJzbnsvMCABgjjACAJjzHEaHDx/WvHnzVFBQIJ/Pp927dyccX7RokXw+X8I2dWoSX6YCABg0PIdRV1eXJkyYoLq6us89Z+7cuWptbY1v+/fv/1JNAgCym+dveq2oqFBFRcVtz/H7/QqFQkk3BQAYXNLynlFDQ4Py8vI0btw4LV68WO3t7Z97biwWUzQaTdgAAINLysOooqJC27Zt08GDB7Vu3TqdOHFCs2fPViwWu+X5tbW1CgaD8a2wsDDVLQEABjjPv6a7kwULFsT/XFJSokmTJqmoqEj79u1TZWVln/NXrlyp6urq+OtoNEogAcAgk/Iwulk4HFZRUZGamppuedzv98vv96e7DQDAAJb2zxl1dHSopaVF4XA43ZcCAGQozzOjy5cv6913342/bm5u1ltvvaXc3Fzl5uaqpqZGjz32mMLhsM6fP6/nn39eo0eP1qOPPprSxgEA2cNzGL355puaNWtW/PWN93sWLlyojRs36vTp09q6dasuXbqkcDisWbNmaefOnQoEAqnrGgCQVTyHUVlZmW63tuqBA/2zICMAIHuk/QEGALfm+19vea75+Ft5SV1r8oJlnmve+LMfe675x1l/47nm33+13HONJEWmJ1WGAYqFUgEA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJhjoVQgg/R80J5UXf5fea/75LluzzUjfcM912z66n/3XCNJDz+63HPNyFfeSOpaSD9mRgAAc4QRAMAcYQQAMEcYAQDMEUYAAHOEEQDAHGEEADBHGAEAzBFGAABzhBEAwBxhBAAwRxgBAMyxUCpg5Pr0r3mu+adv353UtUq+dt5zTTKLnibjJx/+66TqRu55M8WdwBIzIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5ggjAIA5wggAYI4wAgCYI4wAAOZYKBX4DN+kkqTqfvun3hcV3fTHWzzXzLj7quea/hRz1zzXHP+wOLmLXW9Nrg4DEjMjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5ggjAIA5Vu1GRhhaXOS55p++W+C5pmbBDs81kvTY711Mqm4ge/6DSZ5rGn881XPNv9hyzHMNsg8zIwCAOcIIAGDOUxjV1tZq8uTJCgQCysvL0/z583X27NmEc5xzqqmpUUFBgUaMGKGysjKdOXMmpU0DALKLpzBqbGxUVVWVjh8/rvr6enV3d6u8vFxdXV3xc9auXav169errq5OJ06cUCgU0pw5c9TZ2Zny5gEA2cHTAwyvvvpqwuvNmzcrLy9PJ0+e1IwZM+Sc04YNG7Rq1SpVVlZKkrZs2aL8/Hxt375dTz31VJ+fGYvFFIvF4q+j0Wgyfw8AQAb7Uu8ZRSIRSVJubq4kqbm5WW1tbSovL4+f4/f7NXPmTB09evSWP6O2tlbBYDC+FRYWfpmWAAAZKOkwcs6purpa06dPV0lJiSSpra1NkpSfn59wbn5+fvzYzVauXKlIJBLfWlpakm0JAJChkv6c0dKlS/X222/ryJEjfY75fL6E1865Pvtu8Pv98vv9ybYBAMgCSc2Mli1bpr179+rQoUMaM2ZMfH8oFJKkPrOg9vb2PrMlAABu8BRGzjktXbpUu3bt0sGDB1VcXJxwvLi4WKFQSPX19fF9V69eVWNjo0pLS1PTMQAg63j6NV1VVZW2b9+uPXv2KBAIxGdAwWBQI0aMkM/n0/Lly7VmzRqNHTtWY8eO1Zo1azRy5Eg9+eSTafkLAAAyn6cw2rhxoySprKwsYf/mzZu1aNEiSdJzzz2nK1euaMmSJfroo480ZcoUvfbaawoEAilpGACQfXzOOWfdxGdFo1EFg0GV6REN9Q2zbge3MfSr9yZVF5kY9lyz4Iev3vmkmzz9++c81wx0K1q9L0QqScde8r7oae7L/9v7ha73eK9B1up219SgPYpEIsrJybntuaxNBwAwRxgBAMwRRgAAc4QRAMAcYQQAMEcYAQDMEUYAAHOEEQDAHGEEADBHGAEAzBFGAABzhBEAwFzS3/SKgWtoOOS55sO/HeW55pniRs81kvRE4IOk6gaypf883XPNP2z8muea0X//fzzXSFJu57Gk6oD+wswIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOVbv7ydU/meS95j9+mNS1nv/D/Z5rykd0JXWtgeyDniuea2bsXZHUte7/z//ouSb3kveVtK97rgAyAzMjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJgjjAAA5lgotZ+cn+899387/pdp6CR1Xrx0X1J1P24s91zj6/F5rrn/R82ea8Z+8IbnGknqSaoKwA3MjAAA5ggjAIA5wggAYI4wAgCYI4wAAOYIIwCAOcIIAGCOMAIAmCOMAADmCCMAgDnCCABgjjACAJjzOeecdROfFY1GFQwGVaZHNNQ3zLodAECSut01NWiPIpGIcnJybnsuMyMAgDnCCABgzlMY1dbWavLkyQoEAsrLy9P8+fN19uzZhHMWLVokn8+XsE2dOjWlTQMAsounMGpsbFRVVZWOHz+u+vp6dXd3q7y8XF1dXQnnzZ07V62trfFt//79KW0aAJBdPH3T66uvvprwevPmzcrLy9PJkyc1Y8aM+H6/369QKJSaDgEAWe9LvWcUiUQkSbm5uQn7GxoalJeXp3Hjxmnx4sVqb2//3J8Ri8UUjUYTNgDA4JJ0GDnnVF1drenTp6ukpCS+v6KiQtu2bdPBgwe1bt06nThxQrNnz1YsFrvlz6mtrVUwGIxvhYWFybYEAMhQSX/OqKqqSvv27dORI0c0ZsyYzz2vtbVVRUVF2rFjhyorK/scj8ViCUEVjUZVWFjI54wAIMN5+ZyRp/eMbli2bJn27t2rw4cP3zaIJCkcDquoqEhNTU23PO73++X3+5NpAwCQJTyFkXNOy5Yt0yuvvKKGhgYVFxffsaajo0MtLS0Kh8NJNwkAyG6e3jOqqqrSL37xC23fvl2BQEBtbW1qa2vTlStXJEmXL1/Ws88+q2PHjun8+fNqaGjQvHnzNHr0aD366KNp+QsAADKfp5nRxo0bJUllZWUJ+zdv3qxFixZpyJAhOn36tLZu3apLly4pHA5r1qxZ2rlzpwKBQMqaBgBkF8+/prudESNG6MCBA1+qIQDA4MPadAAAc4QRAMAcYQQAMEcYAQDMEUYAAHOEEQDAHGEEADBHGAEAzBFGAABzhBEAwBxhBAAwRxgBAMwRRgAAc4QRAMAcYQQAMEcYAQDMEUYAAHOEEQDAHGEEADBHGAEAzBFGAABzhBEAwBxhBAAwRxgBAMwRRgAAc0OtG7iZc06S1K1rkjNuBgCQtG5dk/Tpv9dvZ8CFUWdnpyTpiPYbdwIASIXOzk4Fg8HbnuNzXySy+tH169f1/vvvKxAIyOfzJRyLRqMqLCxUS0uLcnJyjDocGBiLXoxDL8bhU4xFr4EwDs45dXZ2qqCgQHfddft3hQbczOiuu+7SmDFjbntOTk7OoL7JPoux6MU49GIcPsVY9LIehzvNiG7gAQYAgDnCCABgLqPCyO/3a/Xq1fL7/datmGMsejEOvRiHTzEWvTJtHAbcAwwAgMEno2ZGAIDsRBgBAMwRRgAAc4QRAMAcYQQAMEcYAQDMEUYAAHOEEQDA3P8DZ6yam7DUFooAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 480x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.matshow(X_train[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "be410088-24b6-4bfe-9859-ff9fc5c9070d",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = X_train / 255\n",
    "X_test = X_test / 255"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b1ad232b-381e-4301-a8c0-acd1eb997b47",
   "metadata": {},
   "source": [
    "### Using ANN for Classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "b6378696-30f5-44ed-9a88-7e3894fe1e71",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "  25/1875 [..............................] - ETA: 7s - loss: 1.7117 - accuracy: 0.5425"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:45:45.236515: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.2734 - accuracy: 0.9216\n",
      "Epoch 2/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.1238 - accuracy: 0.9635\n",
      "Epoch 3/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0880 - accuracy: 0.9736\n",
      "Epoch 4/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0661 - accuracy: 0.9803\n",
      "Epoch 5/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0536 - accuracy: 0.9837\n",
      "Epoch 6/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.0441 - accuracy: 0.9862\n",
      "Epoch 7/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0355 - accuracy: 0.9889\n",
      "Epoch 8/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.0284 - accuracy: 0.9915\n",
      "Epoch 9/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.0232 - accuracy: 0.9928\n",
      "Epoch 10/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.0199 - accuracy: 0.9938\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x29b44a910>"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ann = keras.Sequential([\n",
    "    keras.layers.Flatten(input_shape=(28,28)),\n",
    "    keras.layers.Dense(100, activation=\"relu\"),\n",
    "    keras.layers.Dense(10, activation=\"sigmoid\")\n",
    "])\n",
    "\n",
    "ann.compile(optimizer=\"adam\",\n",
    "            loss=\"sparse_categorical_crossentropy\",\n",
    "            metrics=[\"accuracy\"])\n",
    "\n",
    "ann.fit(X_train, y_train, epochs=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "c2d5cf2e-c30e-4425-81a0-f5a4e3c01535",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 40/313 [==>...........................] - ETA: 1s - loss: 0.0914 - accuracy: 0.9734"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:47:00.208713: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 1s 4ms/step - loss: 0.0809 - accuracy: 0.9776\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.08093953877687454, 0.9776000380516052]"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ann.evaluate(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "b1c42b5e-e2f8-4d10-a243-3e6660ecc8e6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(60000, 28, 28, 1)"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train = X_train.reshape(-1,28,28,1)\n",
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "146faee8-12c1-40dd-b9ec-997118db8318",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 28, 28, 1)"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test = X_test.reshape(-1,28,28,1)\n",
    "X_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c1fa96dc-bff8-45ca-b8b4-5a50bf6facc0",
   "metadata": {},
   "source": [
    "### Using CNN for Classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "c39d894e-e7c2-4d16-9c7b-8fa3114e9173",
   "metadata": {},
   "outputs": [],
   "source": [
    "cnn = keras.Sequential([\n",
    "    layers.Conv2D(30, (3,3), activation=\"relu\", input_shape=(28,28,1)),\n",
    "    layers.MaxPooling2D((2,2)),\n",
    "    \n",
    "    layers.Flatten(),\n",
    "    layers.Dense(100, activation=\"relu\"),\n",
    "    layers.Dense(10, activation=\"sigmoid\")\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "f61641a3-0397-4d00-b6b1-554415eab9e9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:51:16.932786: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1875/1875 [==============================] - 15s 7ms/step - loss: 0.1554 - accuracy: 0.9535\n",
      "Epoch 2/5\n",
      "1875/1875 [==============================] - 14s 7ms/step - loss: 0.0532 - accuracy: 0.9839\n",
      "Epoch 3/5\n",
      "1875/1875 [==============================] - 14s 7ms/step - loss: 0.0340 - accuracy: 0.9895\n",
      "Epoch 4/5\n",
      "1875/1875 [==============================] - 14s 8ms/step - loss: 0.0235 - accuracy: 0.9925\n",
      "Epoch 5/5\n",
      "1875/1875 [==============================] - 14s 8ms/step - loss: 0.0166 - accuracy: 0.9948\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x2d70d4af0>"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cnn.compile(optimizer=\"adam\",\n",
    "            loss=\"sparse_categorical_crossentropy\",\n",
    "            metrics=[\"accuracy\"])\n",
    "\n",
    "cnn.fit(X_train, y_train, epochs=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "19bb7a69-8861-4063-8b87-971322053636",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([5, 0, 4, 1, 9], dtype=uint8)"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "50e5fcc4-56b1-4865-8838-dcdab5288b88",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 31/313 [=>............................] - ETA: 1s - loss: 0.0427 - accuracy: 0.9839"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-21 09:52:27.840487: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:113] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 2s 5ms/step - loss: 0.0397 - accuracy: 0.9874\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.03973891958594322, 0.9874000549316406]"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cnn.evaluate(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "15cd0829-7979-4f93-a719-237cd4b61820",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.9.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}