classification-binaire-de-la-base-mnist-nn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "564ffacf",
   "metadata": {
    "papermill": {
     "duration": 0.009212,
     "end_time": "2023-01-11T22:49:42.903451",
     "exception": false,
     "start_time": "2023-01-11T22:49:42.894239",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# DJEKOUN HOCINE OUASSIM - M1 SID"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a05a6370",
   "metadata": {
    "papermill": {
     "duration": 0.005078,
     "end_time": "2023-01-11T22:49:42.914556",
     "exception": false,
     "start_time": "2023-01-11T22:49:42.909478",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "L'objectif de ce tp est la création d'un modele de classification pour la base MNIST sur le chiffre 5 uniquement. En d'autres termes ce modele pourra nous aider a prédire si X image contient le chiffre 5 ou non.\n",
    "Voici les differentes étapes par lequelles on va passer\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8dcb9ec8",
   "metadata": {
    "papermill": {
     "duration": 0.004981,
     "end_time": "2023-01-11T22:49:42.924995",
     "exception": false,
     "start_time": "2023-01-11T22:49:42.920014",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# SOMMAIRE \n",
    "### 1. L'importation des librairies \n",
    "### 2. Charger la base de données MNIST \n",
    "### 3. L'analyse des données\n",
    "### 4. Transformations sur les données\n",
    "### 5. Visualisation des données\n",
    "### 6. Entrainement du modele\n",
    "### 7. Test de la prédiction\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "15c5de4c",
   "metadata": {
    "papermill": {
     "duration": 0.005037,
     "end_time": "2023-01-11T22:49:42.935304",
     "exception": false,
     "start_time": "2023-01-11T22:49:42.930267",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 1. L'importation des librairies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "f9e668fe",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-01-11T22:49:42.948053Z",
     "iopub.status.busy": "2023-01-11T22:49:42.947409Z",
     "iopub.status.idle": "2023-01-11T22:49:51.045991Z",
     "shell.execute_reply": "2023-01-11T22:49:51.044761Z"
    },
    "papermill": {
     "duration": 8.108659,
     "end_time": "2023-01-11T22:49:51.049164",
     "exception": false,
     "start_time": "2023-01-11T22:49:42.940505",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import numpy as np  \n",
    "import pandas as pd \n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "#Entrainement\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "#Modele\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout\n",
    "from tensorflow import keras\n",
    "\n",
    "import random\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0b680d16",
   "metadata": {
    "papermill": {
     "duration": 0.005202,
     "end_time": "2023-01-11T22:49:51.060399",
     "exception": false,
     "start_time": "2023-01-11T22:49:51.055197",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 2. Charger la base de données MNIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "f59a64ff",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-01-11T22:49:51.074595Z",
     "iopub.status.busy": "2023-01-11T22:49:51.073897Z",
     "iopub.status.idle": "2023-01-11T22:49:52.137000Z",
     "shell.execute_reply": "2023-01-11T22:49:52.135639Z"
    },
    "papermill": {
     "duration": 1.073954,
     "end_time": "2023-01-11T22:49:52.139858",
     "exception": false,
     "start_time": "2023-01-11T22:49:51.065904",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n",
      "11493376/11490434 [==============================] - 0s 0us/step\n",
      "11501568/11490434 [==============================] - 0s 0us/step\n"
     ]
    }
   ],
   "source": [
    "# La bibliothèque Keras nous offre la BDD MNIST (blanc sur noir)\n",
    "mnist = keras.datasets.mnist\n",
    "#On sépare notre base de données en deux partie \"test\" et \"train\" pour pouvoir mesurer les performances\n",
    "#de notre modele sur des données differentes de celles de l'entrainement\n",
    "(x_train, y_train), (x_test, y_test) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b886b6ec",
   "metadata": {
    "papermill": {
     "duration": 0.005483,
     "end_time": "2023-01-11T22:49:52.151330",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.145847",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 3.L'analyse des données"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "491a88bc",
   "metadata": {
    "papermill": {
     "duration": 0.00553,
     "end_time": "2023-01-11T22:49:52.164651",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.159121",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Un bloc optionel pour avoir une idée des données qu'on a entre les mains"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "46a2ffda",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-01-11T22:49:52.178753Z",
     "iopub.status.busy": "2023-01-11T22:49:52.178326Z",
     "iopub.status.idle": "2023-01-11T22:49:52.620288Z",
     "shell.execute_reply": "2023-01-11T22:49:52.618883Z"
    },
    "papermill": {
     "duration": 0.452636,
     "end_time": "2023-01-11T22:49:52.623155",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.170519",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Moyenne: 33.318421449829934\n",
      "Min: 0\n",
      "Max: 255\n",
      "Variance: 6172.850482291342\n",
      "Taille train: 60000\n",
      "Taille test: 10000\n"
     ]
    }
   ],
   "source": [
    "print(\"Moyenne:\", np.mean(x_train))\n",
    "print(\"Min:\", np.min(x_train))\n",
    "print(\"Max:\", np.max(x_train))\n",
    "print(\"Variance:\", np.var(x_train))\n",
    "print(\"Taille train:\", len(x_train)) \n",
    "print(\"Taille test:\", len(x_test)) "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b3ba2a0",
   "metadata": {
    "papermill": {
     "duration": 0.005859,
     "end_time": "2023-01-11T22:49:52.635153",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.629294",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 4.Transformations sur les données"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d5eb90dd",
   "metadata": {
    "papermill": {
     "duration": 0.005613,
     "end_time": "2023-01-11T22:49:52.646964",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.641351",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Voici toutes les modifications qu'on doit appliquer sur notre base de données pour atteindre l'objectif de ce TP"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6d7471ce",
   "metadata": {
    "papermill": {
     "duration": 0.00553,
     "end_time": "2023-01-11T22:49:52.658411",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.652881",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## 4.1 Diviser les images MNIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "e05ac5fd",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-01-11T22:49:52.673073Z",
     "iopub.status.busy": "2023-01-11T22:49:52.672598Z",
     "iopub.status.idle": "2023-01-11T22:49:52.908889Z",
     "shell.execute_reply": "2023-01-11T22:49:52.907675Z"
    },
    "papermill": {
     "duration": 0.246823,
     "end_time": "2023-01-11T22:49:52.911692",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.664869",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "#Comme on l'a remarqué dans la partie \"Analyse des données\" la valeur minimum de chaque cellule est 0 et la valeur maximal est 255. Ce n'est rien d'autre que le code de couleur de chaque pixel 255 etant blanc et 0 totalement noir \n",
    "#on souhaite prend le gris comme reference , chaque pixel dépassant la couleur gris est concidéré blanche, autrement elle est considéré noir.\n",
    "x_train[x_train < 127] = 1\n",
    "x_train[x_train >= 127] = 0\n",
    "\n",
    "x_test[x_test < 127] = 1\n",
    "x_test[x_test >= 127] = 0\n",
    "\n",
    "\n",
    "#Puis, on définit une fonction qui divise par 4 la matrice representant chaque image\n",
    "def premiere_division(image):\n",
    "    hauteur, largeur = image.shape\n",
    "    \n",
    "    diviseur = largeur // 2\n",
    "    coin_haut_droit = image[:diviseur, :diviseur] \n",
    "    coin_haut_gauche = image[:diviseur, diviseur:] \n",
    "    coin_bas_droit = image[diviseur:, :diviseur] \n",
    "    coin_bas_gauche = image[diviseur:, diviseur:]\n",
    "\n",
    "    return coin_haut_droit, coin_haut_gauche, coin_bas_droit, coin_bas_gauche\n",
    "\n",
    "\n",
    "#La dimension voulue étant toujours pas atteinte, on redéfinit une autre fonction qui s'occupera de la deuxieme division par 4\n",
    "def deuxieme_division(image):\n",
    "    coin_haut_droit, coin_haut_gauche, coin_bas_droit, coin_bas_gauche = premiere_division(image)\n",
    "    nouvelle_image = premiere_division(coin_haut_droit), premiere_division(coin_haut_gauche), premiere_division(coin_bas_droit), premiere_division(coin_bas_gauche)\n",
    "    nouvelle_image = np.asarray(nouvelle_image)\n",
    "    nouvelle_image = nouvelle_image.reshape(16, 7, 7)\n",
    "    return nouvelle_image\n",
    "\n",
    "#A ce stade, nous avons atteint le découpage que l'on souhaitait appliquer, qui est que chaque image est sous forme de 16 matrices, chacune étant de dimension 7x7"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "591b0964",
   "metadata": {
    "papermill": {
     "duration": 0.005638,
     "end_time": "2023-01-11T22:49:52.923751",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.918113",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## 4.2 Calcul du pourcentage de noir"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e2ddda85",
   "metadata": {
    "papermill": {
     "duration": 0.00564,
     "end_time": "2023-01-11T22:49:52.935726",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.930086",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "On calcule le pourcentage de pixels noirs dans la sous-image de notre \"nouvelle_image\" en divisant le nombre de pixels noirs par 49 (nombre total de pixels dans chaque sous-image) et en multipliant par 100."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f5f09f1f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-01-11T22:49:52.949432Z",
     "iopub.status.busy": "2023-01-11T22:49:52.948982Z",
     "iopub.status.idle": "2023-01-11T22:51:40.770920Z",
     "shell.execute_reply": "2023-01-11T22:51:40.769490Z"
    },
    "papermill": {
     "duration": 107.832377,
     "end_time": "2023-01-11T22:51:40.774067",
     "exception": false,
     "start_time": "2023-01-11T22:49:52.941690",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def mnist_pourcentage(m):\n",
    "    nouvelle_mnist = []\n",
    "    for img in m:\n",
    "        image = deuxieme_division(img)\n",
    "        image_pourcentage = []\n",
    "        for i in range(16):\n",
    "            noirs = 0\n",
    "            for j in range(7):\n",
    "                for k in range(7):\n",
    "                    if(image[i][j][k] == 0):\n",
    "                        noirs += 1\n",
    "            image_pourcentage.append((noirs / 49) * 100)\n",
    "        nouvelle_mnist.append(image_pourcentage)\n",
    "    return nouvelle_mnist\n",
    "\n",
    "\n",
    "\n",
    "# Maintenant on applique cette fonction sur notre split_test_train dont on aura besoin pour entrainer notre modele et tester sa précision\n",
    "new_x_train = mnist_pourcentage(x_train)\n",
    "new_x_test = mnist_pourcentage(x_test)\n",
    "new_x_train = np.array(new_x_train)\n",
    "new_x_test = np.array(new_x_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3435cbb2",
   "metadata": {
    "papermill": {
     "duration": 0.00559,
     "end_time": "2023-01-11T22:51:40.785893",
     "exception": false,
     "start_time": "2023-01-11T22:51:40.780303",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 5.Visualisation des données"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "732d150d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-01-11T22:51:40.799685Z",
     "iopub.status.busy": "2023-01-11T22:51:40.799294Z",
     "iopub.status.idle": "2023-01-11T22:51:42.694249Z",
     "shell.execute_reply": "2023-01-11T22:51:42.692824Z"
    },
    "papermill": {
     "duration": 1.90521,
     "end_time": "2023-01-11T22:51:42.697053",
     "exception": false,
     "start_time": "2023-01-11T22:51:40.791843",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/NK7nSAAAACXBIWXMAAAsTAAALEwEAmpwYAAAKrklEQVR4nO3dT6hc9RnG8edp1I26SJrhcomh10oohEKjDKFgEItVYjbRjZiFpCBcFwoKLip2UZehVKWLIlxrMC1WKaiYRWhNgyBCEUdJ88fQxoYrJlxzJ2RhXNno28WcyBhn7h3nnJlzmvf7gWFmzszNvAx+PfP/54gQgCvf9+oeAMB0EDuQBLEDSRA7kASxA0lcNc0bW79+fczNzU3zJoFUFhcXde7cOQ+6rFTstrdL+p2kNZL+EBF7Vrr+3NycOp1OmZsEsIJ2uz30srEfxtteI+n3ku6WtFnSLtubx/33AExWmefsWyV9FBGnIuILSa9I2lnNWACqVib2DZI+6Tt/utj2DbbnbXdsd7rdbombA1DGxF+Nj4iFiGhHRLvVak365gAMUSb2M5I29p2/odgGoIHKxP6epE22b7R9jaT7Je2vZiwAVRv7rbeIuGj7EUl/U++tt70RcbyyyQBUqtT77BFxQNKBimYBMEF8XBZIgtiBJIgdSILYgSSIHUiC2IEkiB1IgtiBJIgdSILYgSSIHUiC2IEkiB1IgtiBJIgdSILYgSSIHUiC2IEkiB1IgtiBJIgdSGKqSzYD/eyBKwt/LSKmNEkO7NmBJIgdSILYgSSIHUiC2IEkiB1IgtiBJIgdSKLUh2psL0q6IOlLSRcjol3FUACqV8Un6H4WEecq+HcATBAP44EkysYekt60/b7t+UFXsD1vu2O70+12S94cgHGVjX1bRNwi6W5JD9u+7fIrRMRCRLQjot1qtUreHIBxlYo9Is4Ux8uSXpe0tYqhAFRv7NhtX2v7+kunJd0l6VhVgwGoVplX42ckvV58J/kqSX+OiL9WMhWuGKt9Zx3TM3bsEXFK0k8qnAXABPHWG5AEsQNJEDuQBLEDSRA7kASxA0kQO5AEsQNJEDuQBLEDSRA7kASxA0kQO5AESzajNizJPF3s2YEkiB1IgtiBJIgdSILYgSSIHUiC2IEkeJ8dpfBT0f8/2LMDSRA7kASxA0kQO5AEsQNJEDuQBLEDSRA7kMSqsdvea3vZ9rG+betsH7R9sjheO9kxAZQ1yp79RUnbL9v2hKRDEbFJ0qHiPIAGWzX2iHhb0vnLNu+UtK84vU/SPdWOBaBq4z5nn4mIpeL0p5Jmhl3R9rztju1Ot9sd8+YAlFX6Bbro/Wrg0F8OjIiFiGhHRLvVapW9OQBjGjf2s7ZnJak4Xq5uJACTMG7s+yXtLk7vlvRGNeMAmJRR3np7WdI/JP3I9mnbD0raI+lO2ycl/bw4D6DBVv3xiojYNeSiOyqeBcAE8Qk6IAliB5IgdiAJYgeSIHYgCX5KGisq+1PRLMvcHOzZgSSIHUiC2IEkiB1IgtiBJIgdSILYgSSIHUiC2IEkiB1IgtiBJIgdSILYgSSIHUiC2IEkiB1IgtiBJIgdSILYgSSIHUiC2IEkiB1IgtiBJPjd+OT4Xfg8Rlmffa/tZdvH+rY9ZfuM7cPFYcdkxwRQ1igP41+UtH3A9mcjYktxOFDtWACqtmrsEfG2pPNTmAXABJV5ge4R20eKh/lrh13J9rztju1Ot9stcXMAyhg39uck3SRpi6QlSU8Pu2JELEREOyLarVZrzJsDUNZYsUfE2Yj4MiK+kvS8pK3VjgWgamPFbnu27+y9ko4Nuy6AZlj1fXbbL0u6XdJ626cl/VrS7ba3SApJi5IemtyIAKqwauwRsWvA5hcmMAuACeLjskASxA4kQexAEsQOJEHsQBLEDiRB7EASxA4kQexAEsQOJEHsQBLEDiRB7EAS/JT0Fa7sT0XjysGeHUiC2IEkiB1IgtiBJIgdSILYgSSIHUiC99mxIpZkvnKwZweSIHYgCWIHkiB2IAliB5IgdiAJYgeSIHYgiVVjt73R9lu2P7R93PajxfZ1tg/aPlkcr538uADGNcqe/aKkxyNis6SfSnrY9mZJT0g6FBGbJB0qzgNoqFVjj4iliPigOH1B0glJGyTtlLSvuNo+SfdMaEYAFfhOz9ltz0m6WdK7kmYiYqm46FNJM0P+Zt52x3an2+2WmRVACSPHbvs6Sa9KeiwiPuu/LHrflhj4jYmIWIiIdkS0W61WqWEBjG+k2G1frV7oL0XEa8Xms7Zni8tnJS1PZkQAVRjl1XhLekHSiYh4pu+i/ZJ2F6d3S3qj+vEwCttDD8Alo3yf/VZJD0g6avtwse1JSXsk/cX2g5I+lnTfRCYEUIlVY4+IdyQN20XcUe04ACaFT9ABSRA7kASxA0kQO5AEsQNJ8FPSyfFT0XmwZweSIHYgCWIHkiB2IAliB5IgdiAJYgeSIHYgCWIHkiB2IAliB5IgdiAJYgeSIHYgCWIHkuD77FcAvpOOUbBnB5IgdiAJYgeSIHYgCWIHkiB2IAliB5IYZX32jbbfsv2h7eO2Hy22P2X7jO3DxWHH5McFMK5RPlRzUdLjEfGB7eslvW/7YHHZsxHx28mNB6Aqo6zPviRpqTh9wfYJSRsmPRiAan2n5+y25yTdLOndYtMjto/Y3mt77ZC/mbfdsd3pdrvlpgUwtpFjt32dpFclPRYRn0l6TtJNkraot+d/etDfRcRCRLQjot1qtcpPDGAsI8Vu+2r1Qn8pIl6TpIg4GxFfRsRXkp6XtHVyYwIoa5RX4y3pBUknIuKZvu2zfVe7V9Kx6scDUJVRXo2/VdIDko7aPlxse1LSLttbJIWkRUkPTWA+ABUZ5dX4dyR5wEUHqh8HwKTwCTogCWIHkiB2IAliB5IgdiAJYgeSIHYgCWIHkiB2IAliB5IgdiAJYgeSIHYgCWIHkvA0l/u13ZX0cd+m9ZLOTW2A76apszV1LonZxlXlbD+IiIG//zbV2L9143YnItq1DbCCps7W1LkkZhvXtGbjYTyQBLEDSdQd+0LNt7+Sps7W1LkkZhvXVGar9Tk7gOmpe88OYEqIHUiilthtb7f9L9sf2X6ijhmGsb1o+2ixDHWn5ln22l62faxv2zrbB22fLI4HrrFX02yNWMZ7hWXGa73v6l7+fOrP2W2vkfRvSXdKOi3pPUm7IuLDqQ4yhO1FSe2IqP0DGLZvk/S5pD9GxI+Lbb+RdD4i9hT/o1wbEb9syGxPSfq87mW8i9WKZvuXGZd0j6RfqMb7boW57tMU7rc69uxbJX0UEaci4gtJr0jaWcMcjRcRb0s6f9nmnZL2Faf3qfcfy9QNma0RImIpIj4oTl+QdGmZ8VrvuxXmmoo6Yt8g6ZO+86fVrPXeQ9Kbtt+3PV/3MAPMRMRScfpTSTN1DjPAqst4T9Nly4w35r4bZ/nzsniB7tu2RcQtku6W9HDxcLWRovccrEnvnY60jPe0DFhm/Gt13nfjLn9eVh2xn5G0se/8DcW2RoiIM8XxsqTX1bylqM9eWkG3OF6ueZ6vNWkZ70HLjKsB912dy5/XEft7kjbZvtH2NZLul7S/hjm+xfa1xQsnsn2tpLvUvKWo90vaXZzeLemNGmf5hqYs4z1smXHVfN/Vvvx5REz9IGmHeq/I/0fSr+qYYchcP5T0z+JwvO7ZJL2s3sO6/6r32saDkr4v6ZCkk5L+Lmldg2b7k6Sjko6oF9ZsTbNtU+8h+hFJh4vDjrrvuxXmmsr9xsdlgSR4gQ5IgtiBJIgdSILYgSSIHUiC2IEkiB1I4n8vj15U+UU9PgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/NK7nSAAAACXBIWXMAAAsTAAALEwEAmpwYAAALGUlEQVR4nO3dT6hc5R3G8eep1Y26SJohhBh6rWQTCo0yhIIiFqnEbKIbMQtJQbguFBRcVOxCl6FUpYsixBpMi1UEFbMIrWkQxI04Spo/hjZWbjDhmjshC+PKRn9d3BO5xjt/nHPOnHPv7/uBYc6859w5P44+OWfed868jggBWP1+1HQBAKaDsANJEHYgCcIOJEHYgSR+PM2drVu3LmZmZqa5SyCVubk5nT9/3sutKxV229sl/VHSVZL+HBF7hm0/MzOjXq9XZpcAhuh2uwPXTXwZb/sqSX+SdLekLZJ22d4y6fsBqFeZz+zbJH0SEZ9GxFeSXpW0s5qyAFStTNg3SvpsyeszRdt32J613bPd6/f7JXYHoIzae+MjYm9EdCOi2+l06t4dgAHKhP2spE1LXt9QtAFooTJh/0DSZts32r5G0v2SDlRTFoCqTTz0FhGXbD8i6R9aHHrbFxEnKqsMQKVKjbNHxEFJByuqBUCN+LoskARhB5Ig7EAShB1IgrADSRB2IAnCDiRB2IEkCDuQBGEHkiDsQBKEHUiCsANJEHYgCcIOJEHYgSQIO5AEYQeSIOxAEoQdSIKwA0kQdiAJwg4kQdiBJAg7kARhB5Ig7EAShB1IgrADSRB2IIlSUzbbnpN0UdLXki5FRLeKogBUr1TYC7+KiPMVvA+AGnEZDyRRNuwh6W3bH9qeXW4D27O2e7Z7/X6/5O4ATKps2G+LiFsk3S3pYdu3X7lBROyNiG5EdDudTsndAZhUqbBHxNnieUHSm5K2VVEUgOpNHHbb19q+/vKypLskHa+qMADVKtMbv17Sm7Yvv8/fIuLvlVSFyhT/fSYWERVVgqZNHPaI+FTSLyqsBUCNGHoDkiDsQBKEHUiCsANJEHYgiSpuhEHDyg6vIQfO7EAShB1IgrADSRB2IAnCDiRB2IEkCDuQBOPsK0CT4+ij9s0tsCsHZ3YgCcIOJEHYgSQIO5AEYQeSIOxAEoQdSIJx9hZYyfejMw6/cnBmB5Ig7EAShB1IgrADSRB2IAnCDiRB2IEkGGdHrYaNwzMGP10jz+y299lesH18Sdta24dsnyqe19RbJoCyxrmMf0nS9ivanpB0OCI2SzpcvAbQYiPDHhHvSrpwRfNOSfuL5f2S7qm2LABVm7SDbn1EzBfLn0taP2hD27O2e7Z7/X5/wt0BKKt0b3ws9rIM7GmJiL0R0Y2IbqfTKbs7ABOaNOznbG+QpOJ5obqSANRh0rAfkLS7WN4t6a1qygFQl5Hj7LZfkXSHpHW2z0h6StIeSa/ZflDSaUn31VkkhqtzvLrOe+25F366RoY9InYNWHVnxbUAqBFflwWSIOxAEoQdSIKwA0kQdiAJbnFdARiCQhU4swNJEHYgCcIOJEHYgSQIO5AEYQeSIOxAEoyzT0HZ20S5FRRV4MwOJEHYgSQIO5AEYQeSIOxAEoQdSIKwA0kwzj4Fo8bB6/y55lGa3DemizM7kARhB5Ig7EAShB1IgrADSRB2IAnCDiTBOPsq0ORYeZ3fIeA+/mqNPLPb3md7wfbxJW1P2z5r+0jx2FFvmQDKGucy/iVJ25dpfy4ithaPg9WWBaBqI8MeEe9KujCFWgDUqEwH3SO2jxaX+WsGbWR71nbPdq/f75fYHYAyJg3785JukrRV0rykZwZtGBF7I6IbEd1OpzPh7gCUNVHYI+JcRHwdEd9IekHStmrLAlC1icJue8OSl/dKOj5oWwDtMHKc3fYrku6QtM72GUlPSbrD9lZJIWlO0kP1lbj6tfl+d8ayV4+RYY+IXcs0v1hDLQBqxNdlgSQIO5AEYQeSIOxAEoQdSIJbXFeAOofm6h5aG/b+TGU9XZzZgSQIO5AEYQeSIOxAEoQdSIKwA0kQdiAJxtlXgZU6ntzmW3tXI87sQBKEHUiCsANJEHYgCcIOJEHYgSQIO5AEYQeSIOxAEoQdSIKwA0kQdiAJwg4kQdiBJAg7kAT3s6O1yt7vPmz9Sv0NgDJGntltb7L9ju2PbZ+w/WjRvtb2Iduniuc19ZcLYFLjXMZfkvR4RGyR9EtJD9veIukJSYcjYrOkw8VrAC01MuwRMR8RHxXLFyWdlLRR0k5J+4vN9ku6p6YaAVTgB3XQ2Z6RdLOk9yWtj4j5YtXnktYP+JtZ2z3bvX6/X6ZWACWMHXbb10l6XdJjEfHF0nWx2NuxbI9HROyNiG5EdDudTqliAUxurLDbvlqLQX85It4oms/Z3lCs3yBpoZ4SAVRhnN54S3pR0smIeHbJqgOSdhfLuyW9VX15wGARMfQxjO2hj9VonHH2WyU9IOmY7SNF25OS9kh6zfaDkk5Luq+WCgFUYmTYI+I9SYP+qbuz2nIA1IWvywJJEHYgCcIOJEHYgSQIO5AEYQeSIOxAEoQdSIKwA0kQdiAJwg4kQdiBJAg7kAQ/JY1Va9g97WV+hnrUe7cVZ3YgCcIOJEHYgSQIO5AEYQeSIOxAEoQdSIKwA0kQdiAJwg4kQdiBJAg7kARhB5Ig7EAShB1IYpz52TfZfsf2x7ZP2H60aH/a9lnbR4rHjvrLBapRZm73lWqcH6+4JOnxiPjI9vWSPrR9qFj3XET8ob7yAFRlnPnZ5yXNF8sXbZ+UtLHuwgBU6wd9Zrc9I+lmSe8XTY/YPmp7n+01A/5m1nbPdq/f75erFsDExg677eskvS7psYj4QtLzkm6StFWLZ/5nlvu7iNgbEd2I6HY6nfIVA5jIWGG3fbUWg/5yRLwhSRFxLiK+johvJL0gaVt9ZQIoa5zeeEt6UdLJiHh2SfuGJZvdK+l49eUBqMo4vfG3SnpA0jHbR4q2JyXtsr1VUkiak/RQDfUBjViNw2/j9Ma/J2m5H9E+WH05AOrCN+iAJAg7kARhB5Ig7EAShB1IgrADSRB2IAnCDiRB2IEkCDuQBGEHkiDsQBKEHUiCsANJeJr37druSzq9pGmdpPNTK+CHaWttba1LorZJVVnbTyNi2d9/m2rYv7dzuxcR3cYKGKKttbW1LonaJjWt2riMB5Ig7EASTYd9b8P7H6attbW1LonaJjWV2hr9zA5gepo+swOYEsIOJNFI2G1vt/1v25/YfqKJGgaxPWf7WDENda/hWvbZXrB9fEnbWtuHbJ8qnpedY6+h2loxjfeQacYbPXZNT38+9c/stq+S9B9Jv5Z0RtIHknZFxMdTLWQA23OSuhHR+BcwbN8u6UtJf4mInxdtv5d0ISL2FP9QromI37aktqclfdn0NN7FbEUblk4zLukeSb9Rg8duSF33aQrHrYkz+zZJn0TEpxHxlaRXJe1soI7Wi4h3JV24onmnpP3F8n4t/s8ydQNqa4WImI+Ij4rli5IuTzPe6LEbUtdUNBH2jZI+W/L6jNo133tIetv2h7Znmy5mGesjYr5Y/lzS+iaLWcbIabyn6Yppxltz7CaZ/rwsOui+77aIuEXS3ZIeLi5XWykWP4O1aex0rGm8p2WZaca/1eSxm3T687KaCPtZSZuWvL6haGuFiDhbPC9IelPtm4r63OUZdIvnhYbr+VabpvFebppxteDYNTn9eRNh/0DSZts32r5G0v2SDjRQx/fYvrboOJHtayXdpfZNRX1A0u5iebektxqs5TvaMo33oGnG1fCxa3z684iY+kPSDi32yP9X0u+aqGFAXT+T9K/icaLp2iS9osXLuv9psW/jQUk/kXRY0ilJ/5S0tkW1/VXSMUlHtRisDQ3VdpsWL9GPSjpSPHY0feyG1DWV48bXZYEk6KADkiDsQBKEHUiCsANJEHYgCcIOJEHYgST+D9IkwQy/0BnHAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Une visualsation basique des données sur lequels on a travaillé pour que vous puissiez avoir un apperçu des modifications\n",
    "for i in range(10):\n",
    "    plt.imshow(x_test[i], cmap='gray')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5540db2b",
   "metadata": {
    "papermill": {
     "duration": 0.008267,
     "end_time": "2023-01-11T22:51:42.714486",
     "exception": false,
     "start_time": "2023-01-11T22:51:42.706219",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 6. Entrainement du modele"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6ed44982",
   "metadata": {
    "papermill": {
     "duration": 0.008402,
     "end_time": "2023-01-11T22:51:42.732769",
     "exception": false,
     "start_time": "2023-01-11T22:51:42.724367",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Le modele qu'on va utiliser est bien évidemment un reseau de neuronnes qui va prendre comme entrées les differents pourcentages de noir dans les 16 matrices representant chaque image de la base de données (16 entrées) et une seule sortie vu que notre classification et binaire ( elle va s'effectuer sur un seul numéro qui est le 5 )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "05c8d4df",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-01-11T22:51:42.753221Z",
     "iopub.status.busy": "2023-01-11T22:51:42.752256Z",
     "iopub.status.idle": "2023-01-11T22:51:42.759053Z",
     "shell.execute_reply": "2023-01-11T22:51:42.757787Z"
    },
    "papermill": {
     "duration": 0.020162,
     "end_time": "2023-01-11T22:51:42.761640",
     "exception": false,
     "start_time": "2023-01-11T22:51:42.741478",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# on limite la valeur à predire de notre training set et testing set à une seule (qui est 5)\n",
    "y_train_5= (y_train == 5) \n",
    "y_test_5= (y_test == 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "89dccdbe",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-01-11T22:51:42.782004Z",
     "iopub.status.busy": "2023-01-11T22:51:42.780490Z",
     "iopub.status.idle": "2023-01-11T22:52:04.266945Z",
     "shell.execute_reply": "2023-01-11T22:52:04.265451Z"
    },
    "papermill": {
     "duration": 21.499006,
     "end_time": "2023-01-11T22:52:04.269512",
     "exception": false,
     "start_time": "2023-01-11T22:51:42.770506",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-01-11 22:51:43.037408: I tensorflow/core/common_runtime/process_util.cc:146] Creating new thread pool with default inter op setting: 2. Tune using inter_op_parallelism_threads for best performance.\n",
      "2023-01-11 22:51:43.206060: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "1875/1875 [==============================] - 4s 2ms/step - loss: 0.4340 - accuracy: 0.9185\n",
      "Epoch 2/5\n",
      "1875/1875 [==============================] - 3s 2ms/step - loss: 0.1185 - accuracy: 0.9556\n",
      "Epoch 3/5\n",
      "1875/1875 [==============================] - 3s 2ms/step - loss: 0.1081 - accuracy: 0.9603\n",
      "Epoch 4/5\n",
      "1875/1875 [==============================] - 3s 2ms/step - loss: 0.1030 - accuracy: 0.9627\n",
      "Epoch 5/5\n",
      "1875/1875 [==============================] - 3s 2ms/step - loss: 0.0995 - accuracy: 0.9637\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f97139f2610>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Construction du modele\n",
    "#\"Flatten\" nous permet de redimensionner la matrice en un vecteur (16,)\n",
    "model = keras.Sequential([\n",
    "    keras.layers.Dense(32, input_shape=(16,), activation='relu'),\n",
    "    keras.layers.Dense(1, activation='sigmoid')\n",
    "])\n",
    "\n",
    "# Compilation de modele\n",
    "model.compile(optimizer='adam',\n",
    "              loss='BinaryCrossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "\n",
    "# Entrainement du modele\n",
    "model.fit(new_x_train, y_train_5, epochs=5)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "015e6f13",
   "metadata": {
    "papermill": {
     "duration": 0.026021,
     "end_time": "2023-01-11T22:52:04.322459",
     "exception": false,
     "start_time": "2023-01-11T22:52:04.296438",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 7.Test de la prédiction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "2190f355",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-01-11T22:52:04.377102Z",
     "iopub.status.busy": "2023-01-11T22:52:04.376659Z",
     "iopub.status.idle": "2023-01-11T22:52:04.953554Z",
     "shell.execute_reply": "2023-01-11T22:52:04.952612Z"
    },
    "papermill": {
     "duration": 0.607676,
     "end_time": "2023-01-11T22:52:04.956555",
     "exception": false,
     "start_time": "2023-01-11T22:52:04.348879",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cette image contient le numéro 5\n",
      "5\n"
     ]
    }
   ],
   "source": [
    "#Aprés une petite recherche on trouve que le 24 eme element de notre testing test est un 5(on peut le verifier grace a son libellé qui s'affiche en bas)\n",
    "#on test notre modele sur cette image\n",
    "prediction = model.predict(new_x_test)\n",
    "prediction[15]\n",
    "plt.imshow(x_test[15], cmap='gray')\n",
    "plt.show()\n",
    "if (prediction[15]> 0.5):\n",
    "     print(\"cette image contient le numéro 5\")\n",
    "else:\n",
    "     print(\"cette image ne contient pas le numéro 5\")\n",
    "print(y_test[15])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.12"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 154.328386,
   "end_time": "2023-01-11T22:52:07.911020",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2023-01-11T22:49:33.582634",
   "version": "2.3.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}