MNIST_Deep_Learning.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "MNIST_Deep_Learning.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "2NyYjeAc1btr"
      },
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn import datasets\n",
        "import keras\n",
        "from keras.datasets import mnist\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense \n",
        "from keras.optimizers import Adam\n",
        "from keras.utils.np_utils import to_categorical\n",
        "import random"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gft291sc2Bcv"
      },
      "source": [
        "np.random.seed(0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "J3idoCzA2BQp",
        "outputId": "0421aadf-b04a-485e-ca12-79385e4635cf"
      },
      "source": [
        "(X_train, y_train), (X_test, y_test) = mnist.load_data()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n",
            "11493376/11490434 [==============================] - 0s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "N4LVR4PD2BQE",
        "outputId": "5400cbe8-ba0b-4550-aa07-05393f1ed13f"
      },
      "source": [
        "print(X_train.shape)\n",
        "print(X_test.shape)\n",
        "# we want to make sure our training set has the same amount of images as it has labels \n",
        "print(y_train.shape[0])\n",
        "# as seen, we have 60000 labels for 60000 images "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(60000, 28, 28)\n",
            "(10000, 28, 28)\n",
            "60000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Aa6wrIsZ2BJq"
      },
      "source": [
        "# assert; it takes a single argument ,True of False, if the condition is met then the code will run smoothly with no issues,   \n",
        "# if the condition is not met, the code will stop running and notify the user that something went wrong \n",
        "# this function helps u to debug ur code \n",
        "assert(X_train.shape[0] == y_train.shape[0]), \"The number of images is not equal to number of labels\"\n",
        "assert(X_test.shape[0] == y_test.shape[0]), \"The number of images is not equal to number of labels\"\n",
        "assert(X_train.shape[1:] == (28,28)), \"The dimensions of the images are not 28x28\"\n",
        "assert(X_test.shape[1:] == (28,28)), \"The dimensions of the images are not 28x28\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6HC_KnMu2A-q"
      },
      "source": [
        "# now we will visulize our data \n",
        "num_of_samples = []\n",
        "\n",
        "# make grid\n",
        "cols = 5\n",
        "num_classes = 10\n",
        "fig, axs = plt.subplots(nrows=num_classes, ncols=cols, figsize=(5,10))\n",
        "fig.tight_layout()\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 718
        },
        "id": "x6rZe2gk2A5R",
        "outputId": "60266cb2-b817-4f6b-b627-74db1f721891"
      },
      "source": [
        "# what we will do loop through every single column and for each column itteration we wil itterate through every row (every class)\n",
        "# we will use nested for loops arrangment \n",
        "# we will repeat the grid \n",
        "cols = 5\n",
        "num_classes = 10\n",
        "fig, axs = plt.subplots(nrows=num_classes, ncols=cols, figsize=(5,10))\n",
        "fig.tight_layout()\n",
        "\n",
        "for i in range(cols):\n",
        "  for j in range (num_classes):\n",
        "    # we have to define an array that will filter out images that dont belong to the class\n",
        "    x_selected = X_train[y_train == j] # it will only keep the images will label J\n",
        "    # we want to reach our x_selected array and print a randomly selected image which corresponds\n",
        "    # to the current itteration (current class)\n",
        "    # and since we are using training data to populate x_selected which has dimension of (60000,28,28)\n",
        "    # then x_selected will have the exact same dimesnion so for each j we want to access a random image from x_selected[0]\n",
        "    axs[j][i].imshow(x_selected[random.randint(0, len(x_selected - 1)), :, :], cmap=plt.get_cmap(\"gray\")) # notice we chose grayscale color map\n",
        "    # we want to turn off the axes values \n",
        "    axs[j][i].axis(\"off\")\n",
        "    # now we want to use labels:\n",
        "    if i == 2:\n",
        "      axs[j][i].set_title(str(j))\n",
        "      # so far we are shoing 5 samples of each class but there are thousands of them \n",
        "      num_of_samples.append(len(x_selected))\n",
        "    \n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 360x720 with 50 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 329
        },
        "id": "ZrJOIYil2A4p",
        "outputId": "d19020b3-33f5-4366-9e97-1c1f95d7873a"
      },
      "source": [
        "print(num_of_samples) # this will show you the amount of images that belong to each class\n",
        "plt.figure(figsize=(12,4))\n",
        "plt.bar(range(0,num_classes), num_of_samples)\n",
        "plt.title(\"Distribution of the training dataset\")\n",
        "plt.xlabel(\"Class number\")\n",
        "plt.ylabel(\"Number of images\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[5923, 6742, 5958, 6131, 5842, 5421, 5918, 6265, 5851, 5949]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0, 0.5, 'Number of images')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-PtnmvRz2A2K"
      },
      "source": [
        "# now we will prepare our data so we can use to train our network\n",
        "# first : one_hot encoding \n",
        "y_train = to_categorical(y_train, 10)\n",
        "y_test = to_categorical(y_test, 10)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qHK09eYy2Axs",
        "outputId": "697a9358-8c81-4e00-c154-c4a0a6e7ed90"
      },
      "source": [
        "# next step is to normalize our data \n",
        "# normalizing : from 0 to 255 >>>> make it from 0 to 1 \n",
        "X_train = X_train/255\n",
        "X_test = X_test/255\n",
        "\n",
        "# now we want to flatten our images \n",
        "# we currently have images that is 2D array 28x28 (784 pixels in total)\n",
        "# however due to the structure of our neural network our input values are going to be multiplied by \n",
        "# the weight matrix connecting our input layer with the 1st hidden layer \n",
        "# to conduct matrix multiplication we must make our images one dimensional array \n",
        "# so instead of 28x28 we will flatten our image into a single row with 784 pixels\n",
        "\n",
        "num_pixels = 784\n",
        "X_train = X_train.reshape(X_train.shape[0], num_pixels)\n",
        "X_test = X_test.reshape(X_test.shape[0], num_pixels)\n",
        "\n",
        "print(X_train.shape)\n",
        "print(X_test.shape)\n",
        "\n",
        "# now that our data is propably formated  we can finally move on towards creating our neural network\n",
        "# and use it to classify mnist images"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(60000, 784)\n",
            "(10000, 784)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xG6rHyeHnu40"
      },
      "source": [
        "def creat_model():\n",
        "  model = Sequential()\n",
        "  model.add(Dense(10,input_dim=num_pixels, activation='relu'))\n",
        "  model.add(Dense(10, activation='relu'))\n",
        "  model.add(Dense(num_classes, activation='softmax'))\n",
        "  model.compile(Adam(lr=0.1), loss='categorical_crossentropy', metrics=['accuracy'])\n",
        "  return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qtgOsHiHnu8A",
        "outputId": "6e092079-bd0b-436c-da0f-e90022800440"
      },
      "source": [
        "model = creat_model()\n",
        "print(model.summary())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense (Dense)                (None, 10)                7850      \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 10)                110       \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 10)                110       \n",
            "=================================================================\n",
            "Total params: 8,070\n",
            "Trainable params: 8,070\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kblHODNYnu_1",
        "outputId": "c6ff53ae-e827-407f-88c1-1b5699ebdf4c"
      },
      "source": [
        "history = model.fit(X_train, y_train, validation_split=0.1, epochs = 10, batch_size = 200, verbose = 1, shuffle = 1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "270/270 [==============================] - 2s 4ms/step - loss: 1.2573 - accuracy: 0.5469 - val_loss: 0.7365 - val_accuracy: 0.7517\n",
            "Epoch 2/10\n",
            "270/270 [==============================] - 1s 2ms/step - loss: 0.8156 - accuracy: 0.7308 - val_loss: 0.7242 - val_accuracy: 0.7672\n",
            "Epoch 3/10\n",
            "270/270 [==============================] - 1s 2ms/step - loss: 0.7735 - accuracy: 0.7495 - val_loss: 0.7730 - val_accuracy: 0.7482\n",
            "Epoch 4/10\n",
            "270/270 [==============================] - 1s 2ms/step - loss: 0.7867 - accuracy: 0.7520 - val_loss: 0.7768 - val_accuracy: 0.7538\n",
            "Epoch 5/10\n",
            "270/270 [==============================] - 1s 2ms/step - loss: 0.7882 - accuracy: 0.7547 - val_loss: 0.6918 - val_accuracy: 0.7882\n",
            "Epoch 6/10\n",
            "270/270 [==============================] - 1s 2ms/step - loss: 0.7684 - accuracy: 0.7620 - val_loss: 0.7180 - val_accuracy: 0.7788\n",
            "Epoch 7/10\n",
            "270/270 [==============================] - 1s 2ms/step - loss: 0.7658 - accuracy: 0.7572 - val_loss: 0.6895 - val_accuracy: 0.7988\n",
            "Epoch 8/10\n",
            "270/270 [==============================] - 1s 2ms/step - loss: 0.7547 - accuracy: 0.7725 - val_loss: 0.6694 - val_accuracy: 0.7838\n",
            "Epoch 9/10\n",
            "270/270 [==============================] - 1s 2ms/step - loss: 0.7439 - accuracy: 0.7702 - val_loss: 0.6974 - val_accuracy: 0.7780\n",
            "Epoch 10/10\n",
            "270/270 [==============================] - 1s 2ms/step - loss: 0.7497 - accuracy: 0.7682 - val_loss: 0.7131 - val_accuracy: 0.7672\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 312
        },
        "id": "dIXi5r7fnvCh",
        "outputId": "87ead9a0-effd-4e6d-a915-33ea0dc3b5c3"
      },
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.legend(['loss', 'val_loss'])\n",
        "plt.title('loss')\n",
        "plt.xlabel('epoch')\n",
        "\n",
        "# notice that the validation loss is consistantly lower than the training loss which makes sense \n",
        "# when the validation loss is higher .. this means overfitting"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 0, 'epoch')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fCAqNp9r29dW",
        "outputId": "fcdcfa3b-2536-4643-bc3f-1292c0c0dbc2"
      },
      "source": [
        "# now we will test our model and see how it performs with images that were not seen before\n",
        "\n",
        "score = model.evaluate(X_test, y_test, verbose=0)\n",
        "print(type(score))\n",
        "print('Test score: ', score[0])\n",
        "print('Test Accuracy: ', score[1])\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'list'>\n",
            "Test score:  0.758537769317627\n",
            "Test Accuracy:  0.7594000101089478\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Omgubkd-4Gul"
      },
      "source": [
        "# we will try our model on an image from the web\n",
        "import requests\n",
        "from PIL import Image\n",
        "url = 'https://www.researchgate.net/profile/Jose_Sempere/publication/221258631/figure/fig1/AS:305526891139075@1449854695342/Handwritten-digit-2.png'\n",
        "response = requests.get(url, stream=True)\n",
        "print(response)\n",
        "img = Image.open(response.raw)\n",
        "plt.imshow(img)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6gehOVSM4G7_"
      },
      "source": [
        "import cv2\n",
        "img_array = np.asarray(img)\n",
        "resized = cv2.resize(img_array, (28,28))\n",
        "print(resized.shape) # notice we still have a depth of 4 \n",
        "gray_scale = cv2.cvtColor (resized, cv2.COLOR_BGR2GRAY)\n",
        "print(gray_scale.shape)\n",
        "# we want the image to have black background with white number \n",
        "image = cv2.bitwise_not(gray_scale)\n",
        "plt.imshow(image,cmap=plt.get_cmap(\"gray\"))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GnjYRJuN8Kw4"
      },
      "source": [
        "image = image/255\n",
        "image = image.reshape(1,784)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MULPdxsL4G_S"
      },
      "source": [
        "prediction = model.predict_classes(image)\n",
        "print(\"predicted digit: \", str(prediction))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6fO3U4Nz8YnZ"
      },
      "source": [
        "# to make a more accurate model, we can change the model hyperparameters"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}