{"metadata":{"colab":{"provenance":[]},"kernelspec":{"name":"python3","display_name":"Python 3","language":"python"},"language_info":{"name":"python","version":"3.10.13","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"none","dataSources":[{"sourceId":3004,"databundleVersionId":861823,"sourceType":"competition"}],"dockerImageVersionId":30747,"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":false}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# MNIST Digit Classification\nMNIST consists of 60,000 training images and 10,000 test\nimages. All are grayscale (one-channel) in other words one-dimension, and each image is 28 pixels high and 28 pixels wide.\n\nFigure shows some sample images from the MNIST\ndataset.\n\n","metadata":{"id":"cxnJ2mR2qWkl"}},{"cell_type":"markdown","source":"# Loading required libraries","metadata":{"id":"YNmUWRuOXzEh"}},{"cell_type":"code","source":"import pandas as pd\nimport numpy as np\nimport matplotlib\nimport matplotlib.pyplot as plt\n\nfrom sklearn.model_selection import train_test_split\nfrom keras.models import Sequential\nfrom keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout\nfrom keras.callbacks import EarlyStopping\nfrom keras.utils import to_categorical\n\nimport warnings\nwarnings.filterwarnings('ignore')","metadata":{"id":"CgUud25sXgVE","execution":{"iopub.status.busy":"2024-08-06T14:11:44.869357Z","iopub.execute_input":"2024-08-06T14:11:44.869789Z","iopub.status.idle":"2024-08-06T14:11:44.876301Z","shell.execute_reply.started":"2024-08-06T14:11:44.869752Z","shell.execute_reply":"2024-08-06T14:11:44.875376Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Loading the data set","metadata":{"id":"6x6-cGrLX46n"}},{"cell_type":"code","source":"mnist_train = pd.read_csv(\"/kaggle/input/digit-recognizer/train.csv\")\nmnist_test = pd.read_csv(\"/kaggle/input/digit-recognizer/test.csv\")\n\nX = mnist_train.drop(columns = [\"label\"])\ny = mnist_train[\"label\"]","metadata":{"id":"jldepNcQXt3Z","execution":{"iopub.status.busy":"2024-08-06T14:11:44.878134Z","iopub.execute_input":"2024-08-06T14:11:44.878922Z","iopub.status.idle":"2024-08-06T14:11:49.240092Z","shell.execute_reply.started":"2024-08-06T14:11:44.87889Z","shell.execute_reply":"2024-08-06T14:11:49.238971Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X","metadata":{"execution":{"iopub.status.busy":"2024-08-06T14:11:49.241327Z","iopub.execute_input":"2024-08-06T14:11:49.241657Z","iopub.status.idle":"2024-08-06T14:11:49.263117Z","shell.execute_reply.started":"2024-08-06T14:11:49.241632Z","shell.execute_reply":"2024-08-06T14:11:49.26225Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"y","metadata":{"execution":{"iopub.status.busy":"2024-08-06T14:11:49.265683Z","iopub.execute_input":"2024-08-06T14:11:49.266129Z","iopub.status.idle":"2024-08-06T14:11:49.272597Z","shell.execute_reply.started":"2024-08-06T14:11:49.266099Z","shell.execute_reply":"2024-08-06T14:11:49.27163Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Picture size\npicture_size = int(np.sqrt(X.shape[1]))\nprint(\"Picture size:\",picture_size, \"pixle\")","metadata":{"id":"EncuJlJzXwS-","outputId":"1a3110d0-fca1-4909-d4ba-3a399c5581dc","execution":{"iopub.status.busy":"2024-08-06T14:11:49.273764Z","iopub.execute_input":"2024-08-06T14:11:49.274025Z","iopub.status.idle":"2024-08-06T14:11:49.282332Z","shell.execute_reply.started":"2024-08-06T14:11:49.274003Z","shell.execute_reply":"2024-08-06T14:11:49.281493Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# train data shpe\nX.shape","metadata":{"execution":{"iopub.status.busy":"2024-08-06T14:11:49.28333Z","iopub.execute_input":"2024-08-06T14:11:49.283636Z","iopub.status.idle":"2024-08-06T14:11:49.293484Z","shell.execute_reply.started":"2024-08-06T14:11:49.283613Z","shell.execute_reply":"2024-08-06T14:11:49.292719Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# label data values\nnp.sort(y.unique())","metadata":{"execution":{"iopub.status.busy":"2024-08-06T14:11:49.294592Z","iopub.execute_input":"2024-08-06T14:11:49.294882Z","iopub.status.idle":"2024-08-06T14:11:49.304941Z","shell.execute_reply.started":"2024-08-06T14:11:49.294853Z","shell.execute_reply":"2024-08-06T14:11:49.304098Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Visualize a sample\nsome_digit = X.iloc[4000].values.reshape(28, 28)\nsome_digit_image = some_digit\nplt.figure(figsize=(3,3))\nplt.imshow(some_digit_image, cmap = matplotlib.cm.binary,interpolation=\"nearest\")\nplt.title(\"Sample Image\")\nplt.show()","metadata":{"id":"rsuEi1bRYWyg","outputId":"dcd2b2e0-450e-4236-88f3-c446ff2e3e52","execution":{"iopub.status.busy":"2024-08-06T14:11:49.306051Z","iopub.execute_input":"2024-08-06T14:11:49.306323Z","iopub.status.idle":"2024-08-06T14:11:49.465175Z","shell.execute_reply.started":"2024-08-06T14:11:49.306293Z","shell.execute_reply":"2024-08-06T14:11:49.4641Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Data preparation\n\n\n- Training set is used to train the model with.
\n- Validation set is used to test the train performance before testing it with the testing set, how ever it can be made by two ways:
\n \n - Manually by splitting a small portion from the train set
\n - or automatically while training the model
\n
\n- Test set is used to test the model perfomance i
\n
","metadata":{"id":"MbCp21ouYDxv"}},{"cell_type":"code","source":"# Splitting the data into 4 sets\nXtrain, Xtest, ytrain, ytest = train_test_split(X, y ,test_size=0.1, random_state=42)","metadata":{"id":"0DYy16oLYC-v","execution":{"iopub.status.busy":"2024-08-06T14:11:49.466763Z","iopub.execute_input":"2024-08-06T14:11:49.467053Z","iopub.status.idle":"2024-08-06T14:11:49.664609Z","shell.execute_reply.started":"2024-08-06T14:11:49.467027Z","shell.execute_reply":"2024-08-06T14:11:49.663779Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Data Normalization\nXtrain = Xtrain.astype('float32') / 255.\nXtest = Xtest.astype('float32') / 255.\nXtrain = Xtrain.values.reshape(-1,28,28,1)\nXtest = Xtest.values.reshape(-1,28,28,1)","metadata":{"id":"XJeMyRwpUUnV","execution":{"iopub.status.busy":"2024-08-06T14:11:49.668448Z","iopub.execute_input":"2024-08-06T14:11:49.66874Z","iopub.status.idle":"2024-08-06T14:11:49.775055Z","shell.execute_reply.started":"2024-08-06T14:11:49.668715Z","shell.execute_reply":"2024-08-06T14:11:49.774268Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"ytrain = to_categorical(ytrain, num_classes=10)\nytest = to_categorical(ytest, num_classes=10)","metadata":{"id":"Si4H6_dzZnLf","execution":{"iopub.status.busy":"2024-08-06T14:11:49.77635Z","iopub.execute_input":"2024-08-06T14:11:49.776705Z","iopub.status.idle":"2024-08-06T14:11:49.782455Z","shell.execute_reply.started":"2024-08-06T14:11:49.776669Z","shell.execute_reply":"2024-08-06T14:11:49.781643Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Modelling (Training/Testing)","metadata":{"id":"_8e56WqMZbH-"}},{"cell_type":"markdown","source":"# **CNN**\nis a type of artificial neural network\n(Deep Learning) used in image recognition and processing that is specifically\ndesigned to process pixel data.\n\nCNN like any neural network, contains some main parts:\n\n- Input layer: which is the images we want to train.
\n- Conv layer which act like a feature finder window that slides over\nthe image pixel by pixel to extract meaningful features that identify\nthe objects in the image.
\n\n- Feature maps: is the output of each covn layer, and input of the next one.
\n- Activation Function: also called \"Transfare function\" which is a function that transfare the data into other way of data.
\n- Pooling layer: also known as subsampling or down sampling, simple process where we reduce the size or dimensionality of the Feature Map.\n
\n
\n- The Fully Connected Layer: the activation function used to produce\nthese probabilities is the Soft Max Function\nas it turns the outputs of the FC layer (last\nlayer) into probabilities
\n
\n","metadata":{"id":"COwFtONpmMR1"}},{"cell_type":"markdown","source":"","metadata":{"id":"YquCNBi0mENU"}},{"cell_type":"markdown","source":"## Model Training\nModel contains:\n\n- 3 Conv2d Layers (Relu activation function)
\n\n- 3 MaxPooling2d Layers (Subsampling/Downscaling)
\n\n- 1 Dropout Layers (to turn off some units avoiding overfitting issues)
\n\n- 2 Dense Layers (1: Relu activation function, 1: Softmax activation function)
\n
\n","metadata":{"id":"zDF7G6wTZwh2"}},{"cell_type":"code","source":"model = Sequential()\n\nmodel.add(Conv2D(filters=32, kernel_size=4, padding='same', activation='relu', input_shape=(28, 28, 1)))\nmodel.add(MaxPooling2D(pool_size=2))\nmodel.add(Conv2D(filters=64, kernel_size=4, padding='same', activation='relu'))\nmodel.add(MaxPooling2D(pool_size=2))\nmodel.add(Conv2D(filters=128, kernel_size=4, padding='same', activation='relu'))\nmodel.add(MaxPooling2D(pool_size=2))\nmodel.add(Conv2D(filters=256, kernel_size=4, padding='same', activation='relu'))\nmodel.add(MaxPooling2D(pool_size=2))\nmodel.add(Dropout(0.3))\nmodel.add(Flatten())\nmodel.add(Dense(500, activation='relu'))\nmodel.add(Dropout(0.4))\nmodel.add(Dense(10, activation='softmax'))\n\n\nearly_stopping = EarlyStopping(monitor='val_loss', patience=3, restore_best_weights=True)\n\nmodel.compile(optimizer=\"adam\", loss=\"categorical_crossentropy\", metrics=[\"accuracy\"])","metadata":{"id":"EI2jrXfuZawn","outputId":"77989479-0a6e-4a61-c777-2470cff5707a","execution":{"iopub.status.busy":"2024-08-06T14:11:49.783566Z","iopub.execute_input":"2024-08-06T14:11:49.784354Z","iopub.status.idle":"2024-08-06T14:11:49.923034Z","shell.execute_reply.started":"2024-08-06T14:11:49.784298Z","shell.execute_reply":"2024-08-06T14:11:49.922272Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"history = model.fit(Xtrain, ytrain, epochs=30, validation_split=0.1, verbose=1, callbacks=[early_stopping])","metadata":{"execution":{"iopub.status.busy":"2024-08-06T14:11:49.924296Z","iopub.execute_input":"2024-08-06T14:11:49.924961Z","iopub.status.idle":"2024-08-06T14:12:28.289619Z","shell.execute_reply.started":"2024-08-06T14:11:49.924924Z","shell.execute_reply":"2024-08-06T14:12:28.288716Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Model Testing","metadata":{"id":"_3FA_c6zau5W"}},{"cell_type":"code","source":"evaluation = model.evaluate(Xtest, ytest)\nprint(f\"Test Loss: {evaluation[0]}, Test Accuracy: {evaluation[1]}\")\n","metadata":{"id":"wMZd9_l8at4u","execution":{"iopub.status.busy":"2024-08-06T14:12:28.290958Z","iopub.execute_input":"2024-08-06T14:12:28.291258Z","iopub.status.idle":"2024-08-06T14:12:29.087691Z","shell.execute_reply.started":"2024-08-06T14:12:28.291232Z","shell.execute_reply":"2024-08-06T14:12:29.086746Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Model Summary","metadata":{}},{"cell_type":"code","source":"model.summary()","metadata":{"execution":{"iopub.status.busy":"2024-08-06T14:12:29.088938Z","iopub.execute_input":"2024-08-06T14:12:29.089628Z","iopub.status.idle":"2024-08-06T14:12:29.119538Z","shell.execute_reply.started":"2024-08-06T14:12:29.089592Z","shell.execute_reply":"2024-08-06T14:12:29.118658Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.figure(figsize=(12, 4))\n\n\nplt.subplot(1, 2, 1)\nplt.plot(history.history['accuracy'])\nplt.plot(history.history['val_accuracy'])\nplt.title('Model Accuracy')\nplt.ylabel('Accuracy')\nplt.xlabel('Epoch')\nplt.legend(['Train', 'Validation'], loc='lower right')\n\nplt.subplot(1, 2, 2)\nplt.plot(history.history['loss'])\nplt.plot(history.history['val_loss'])\nplt.title('Model Loss')\nplt.ylabel('Loss')\nplt.xlabel('Epoch')\nplt.legend(['Train', 'Validation'], loc='upper right')\n\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2024-08-06T14:12:29.120723Z","iopub.execute_input":"2024-08-06T14:12:29.121072Z","iopub.status.idle":"2024-08-06T14:12:29.613053Z","shell.execute_reply.started":"2024-08-06T14:12:29.12104Z","shell.execute_reply":"2024-08-06T14:12:29.612147Z"},"trusted":true},"execution_count":null,"outputs":[]}]}