{ "cells": [ { "cell_type": "markdown", "metadata": { "id": "GD7iTp3GnPQa" }, "source": [ "\n", "# **MNIST Digit Recognition (CNN)**\n", "\n", "## πŸ“Œ **Project Workflow**\n", "\n", "This project follows an end-to-end **machine learning workflow** to build, evaluate, and deploy a handwritten digit recognition system using a Convolutional Neural Network (CNN).\n", "\n", " \n", "#### 1. **Dataset Loading**\n", "\n", "* Load the MNIST digit dataset provided in CSV format from Kaggle.\n", "* Separate image pixel values and digit labels.\n", "* Inspect dataset shape and structure.\n", "\n", " \n", "\n", "#### 2. **Data Preprocessing**\n", "\n", "\n", "* Convert pixel values from integers to floating-point format.\n", "* Normalize pixel values to the range **[0, 1]**.\n", "* Reshape flattened pixel vectors into **28Γ—28 grayscale images**.\n", "* One-hot encode class labels for multi-class classification.\n", "\n", "\n", " \n", "\n", "#### 3. **Data Splitting**\n", "\n", "* Split the dataset into training and validation sets.\n", "* Use stratified sampling to preserve class distribution.\n", "\n", "\n", " \n", "#### 4. **Visualization**\n", "\n", "* Visualize sample handwritten digits from the training set.\n", "* Verify correct label–image correspondence.\n", "\n", "\n", " \n", "\n", "#### 5. **Model Architecture Design**\n", "\n", "* Design a Convolutional Neural Network using Keras.\n", "* Stack convolution, batch normalization, pooling, and dense layers.\n", "* Use softmax activation for multi-class digit classification.\n", "\n", "\n", " \n", "\n", "#### 6. **Model Training**\n", "\n", "* Train the CNN on the training dataset.\n", "* Monitor validation accuracy and loss during training.\n", "* Apply regularization (Dropout) to reduce overfitting.\n", "\n", "\n", "\n", " \n", "#### 7. **Model Evaluation**\n", "\n", "* Evaluate model performance using validation accuracy.\n", "* Generate classification metrics to assess prediction quality.\n", "\n", " \n", "\n", "\n", "#### 8. **Prediction**\n", "\n", "* Use the trained model to predict labels for unseen test data.\n", "\n", "\n", " \n", "\n", "#### 9. **Model Saving**\n", "\n", "* Save the trained CNN model for deployment.\n", "* Prepare the model for integration into a web application.\n", "\n", " \n", "\n", "#### 9. **Conclusion**\n", "\n", "* Successfully implemented an **end-to-end handwritten digit recognition system** using a CNN.\n", "* With a **web application** that allows users to draw digits on a canvas for real-time predictions\n", "\n", "\n", " \n", "\n", "\n", "#### **Project Summary**\n", "\n", "* Implemented an end-to-end handwritten digit recognition system using CNNs.\n", "* Processed MNIST data provided in CSV format.\n", "* Applied normalization, reshaping, and categorical encoding.\n", "* Designed and trained a deep learning model using Keras.\n", "* Achieved high validation accuracy on unseen data.\n", "* Generated Kaggle submission predictions.\n", "* Saved the trained model for deployment in a web application.\n", "\n" ] }, { "cell_type": "markdown", "metadata": { "id": "wIAIev0Eo-e8" }, "source": [ "## **Importing Libraries**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "ZJHHUqYjhL-p" }, "outputs": [], "source": [ "import numpy as np\n", "import pandas as pd\n", "import matplotlib.pyplot as plt\n", "\n", "\n", "from tensorflow.keras.models import Sequential\n", "from tensorflow.keras.layers import Conv2D, MaxPooling2D, Dense, Flatten, Dropout, BatchNormalization\n", "from tensorflow.keras.utils import to_categorical\n", "from tensorflow.keras.optimizers import Adam\n", "\n", "from sklearn.model_selection import train_test_split\n", "from sklearn.metrics import accuracy_score, classification_report, confusion_matrix\n", "\n" ] }, { "cell_type": "markdown", "metadata": { "id": "tHlSelZnpdOx" }, "source": [ "\n", "## **Dataset Loading**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "hBkbehS1iplY" }, "outputs": [], "source": [ "train_df = pd.read_csv(\"/content/train.csv\")\n", "test_df = pd.read_csv(\"/content/test.csv\")\n", "\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 255 }, "id": "UN65rcRYk5z_", "outputId": "95c6312c-fa77-4614-8128-8f84b9a08cfe" }, "outputs": [ { "data": { "application/vnd.google.colaboratory.intrinsic+json": { "type": "dataframe", "variable_name": "train_df" }, "text/html": [ "\n", "
\n", "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
labelpixel0pixel1pixel2pixel3pixel4pixel5pixel6pixel7pixel8...pixel774pixel775pixel776pixel777pixel778pixel779pixel780pixel781pixel782pixel783
01000000000...0000000000
10000000000...0000000000
21000000000...0000000000
34000000000...0000000000
40000000000...0000000000
\n", "

5 rows Γ— 785 columns

\n", "
\n", "
\n", "\n", "
\n", " \n", "\n", " \n", "\n", " \n", "
\n", "\n", "\n", "
\n", " \n", "\n", "\n", "\n", " \n", "
\n", "\n", "
\n", "
\n" ], "text/plain": [ " label pixel0 pixel1 pixel2 pixel3 pixel4 pixel5 pixel6 pixel7 \\\n", "0 1 0 0 0 0 0 0 0 0 \n", "1 0 0 0 0 0 0 0 0 0 \n", "2 1 0 0 0 0 0 0 0 0 \n", "3 4 0 0 0 0 0 0 0 0 \n", "4 0 0 0 0 0 0 0 0 0 \n", "\n", " pixel8 ... pixel774 pixel775 pixel776 pixel777 pixel778 pixel779 \\\n", "0 0 ... 0 0 0 0 0 0 \n", "1 0 ... 0 0 0 0 0 0 \n", "2 0 ... 0 0 0 0 0 0 \n", "3 0 ... 0 0 0 0 0 0 \n", "4 0 ... 0 0 0 0 0 0 \n", "\n", " pixel780 pixel781 pixel782 pixel783 \n", "0 0 0 0 0 \n", "1 0 0 0 0 \n", "2 0 0 0 0 \n", "3 0 0 0 0 \n", "4 0 0 0 0 \n", "\n", "[5 rows x 785 columns]" ] }, "execution_count": 38, "metadata": {}, "output_type": "execute_result" } ], "source": [ "train_df.head()" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 255 }, "id": "yD8lwcF-k9tM", "outputId": "7fa3b961-4ea5-46e5-f917-76d864098e8b" }, "outputs": [ { "data": { "application/vnd.google.colaboratory.intrinsic+json": { "type": "dataframe", "variable_name": "test_df" }, "text/html": [ "\n", "
\n", "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
pixel0pixel1pixel2pixel3pixel4pixel5pixel6pixel7pixel8pixel9...pixel774pixel775pixel776pixel777pixel778pixel779pixel780pixel781pixel782pixel783
00000000000...0000000000
10000000000...0000000000
20000000000...0000000000
30000000000...0000000000
40000000000...0000000000
\n", "

5 rows Γ— 784 columns

\n", "
\n", "
\n", "\n", "
\n", " \n", "\n", " \n", "\n", " \n", "
\n", "\n", "\n", "
\n", " \n", "\n", "\n", "\n", " \n", "
\n", "\n", "
\n", "
\n" ], "text/plain": [ " pixel0 pixel1 pixel2 pixel3 pixel4 pixel5 pixel6 pixel7 pixel8 \\\n", "0 0 0 0 0 0 0 0 0 0 \n", "1 0 0 0 0 0 0 0 0 0 \n", "2 0 0 0 0 0 0 0 0 0 \n", "3 0 0 0 0 0 0 0 0 0 \n", "4 0 0 0 0 0 0 0 0 0 \n", "\n", " pixel9 ... pixel774 pixel775 pixel776 pixel777 pixel778 pixel779 \\\n", "0 0 ... 0 0 0 0 0 0 \n", "1 0 ... 0 0 0 0 0 0 \n", "2 0 ... 0 0 0 0 0 0 \n", "3 0 ... 0 0 0 0 0 0 \n", "4 0 ... 0 0 0 0 0 0 \n", "\n", " pixel780 pixel781 pixel782 pixel783 \n", "0 0 0 0 0 \n", "1 0 0 0 0 \n", "2 0 0 0 0 \n", "3 0 0 0 0 \n", "4 0 0 0 0 \n", "\n", "[5 rows x 784 columns]" ] }, "execution_count": 39, "metadata": {}, "output_type": "execute_result" } ], "source": [ "test_df.head()" ] }, { "cell_type": "markdown", "metadata": { "id": "yPS01GDhpt48" }, "source": [ "## **Data Splitting**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "EG4f7bPBmNiU" }, "outputs": [], "source": [ "X = train_df.drop(\"label\", axis=1).values\n", "y = train_df[\"label\"].values\n", "\n", "\n", "X_test = test_df.values\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "h9LnngrWmOol", "outputId": "dcd86df6-21f0-463e-9207-56c99c2e100c" }, "outputs": [ { "data": { "text/plain": [ "array([[0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " ...,\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0]])" ] }, "execution_count": 41, "metadata": {}, "output_type": "execute_result" } ], "source": [ "X" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "5Pbm5AMhmP3H", "outputId": "a7e2ed9b-74aa-4ce3-8aa8-d2def21bda7b" }, "outputs": [ { "data": { "text/plain": [ "array([1, 0, 1, ..., 7, 6, 9])" ] }, "execution_count": 42, "metadata": {}, "output_type": "execute_result" } ], "source": [ "y" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "3dlXo--zpyfV", "outputId": "df7acff9-45fa-43ca-d823-2cea82bfe899" }, "outputs": [ { "data": { "text/plain": [ "array([[0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " ...,\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0]])" ] }, "execution_count": 43, "metadata": {}, "output_type": "execute_result" } ], "source": [ "X_test" ] }, { "cell_type": "markdown", "metadata": { "id": "DcUiD9UKp4FK" }, "source": [ "## **Data Preprocessing**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "BUFJSDwtmUVK" }, "outputs": [], "source": [ "X = X.astype(\"float32\") / 255.0\n", "X = X.reshape(-1, 28, 28, 1)\n", "\n", "X_test_final = test_df.values.astype(\"float32\") / 255.0\n", "X_test_final = X_test_final.reshape(-1, 28, 28, 1)\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "YCXWW7-amYAD", "outputId": "807735b2-18ff-416f-db53-ec1b25da5c04" }, "outputs": [ { "data": { "text/plain": [ "array([[[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.03921569],\n", " [0.06666667],\n", " [0.06666667],\n", " [0.06666667],\n", " [0.06666667],\n", " [0.31764707],\n", " [0.7058824 ],\n", " [0.7058824 ],\n", " [0.13725491],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.54509807],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.1882353 ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.23529412],\n", " [0.89411765],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.8117647 ],\n", " [0.77254903],\n", " [0.18039216],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.8352941 ],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.8745098 ],\n", " [0.20392157],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.25882354],\n", " [0.90588236],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.42352942],\n", " [0.15686275],\n", " [0.15686275],\n", " [0.4509804 ],\n", " [0.95686275],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.5254902 ],\n", " [0.01176471],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.24705882],\n", " [0.44705883],\n", " [0.44705883],\n", " [0.44705883],\n", " [0.14509805],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.8039216 ],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.05882353],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.22352941],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.05882353],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.16470589],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.05882353],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.37254903],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.05882353],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.8039216 ],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.05882353],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.23921569],\n", " [0.3882353 ],\n", " [0.3764706 ],\n", " [0. ],\n", " [0. ],\n", " [0.1764706 ],\n", " [0.8784314 ],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.7647059 ],\n", " [0.03921569],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.04313726],\n", " [0.09803922],\n", " [0.4117647 ],\n", " [0.3254902 ],\n", " [0.7411765 ],\n", " [0.7411765 ],\n", " [0.89411765],\n", " [0.99215686],\n", " [0.9843137 ],\n", " [0.7411765 ],\n", " [0.7411765 ],\n", " [0.85490197],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.8235294 ],\n", " [0.10588235],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.16470589],\n", " [0.45490196],\n", " [0.6784314 ],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.8666667 ],\n", " [0.45490196],\n", " [0.02745098],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.4627451 ],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.9607843 ],\n", " [0.83137256],\n", " [0.87058824],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.627451 ],\n", " [0.05882353],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.99607843],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.7411765 ],\n", " [0.3882353 ],\n", " [0. ],\n", " [0.1254902 ],\n", " [0.7921569 ],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.9411765 ],\n", " [0.47843137],\n", " [0.47843137],\n", " [0.74509805],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.68235296],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [1. ],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.93333334],\n", " [0.87058824],\n", " [0.87058824],\n", " [0.87058824],\n", " [0.94509804],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.9019608 ],\n", " [0.27450982],\n", " [0. ],\n", " [0. ],\n", " [0.06666667],\n", " [0.6862745 ],\n", " [0.8980392 ],\n", " [0.99215686],\n", " [0.99215686],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.61960787],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.99215686],\n", " [0.8039216 ],\n", " [0.41568628],\n", " [0.25490198],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.24313726],\n", " [0.95686275],\n", " [0.6156863 ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.02352941],\n", " [0.10196079],\n", " [0.7019608 ],\n", " [0.7019608 ],\n", " [0.7019608 ],\n", " [0.7019608 ],\n", " [0.7019608 ],\n", " [0.11764706],\n", " [0.05882353],\n", " [0.03921569],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0.05490196],\n", " [0.02352941],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]],\n", "\n", " [[0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ],\n", " [0. ]]], dtype=float32)" ] }, "execution_count": 45, "metadata": {}, "output_type": "execute_result" } ], "source": [ "X_test_final[0]" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "sf434SJfmfMU" }, "outputs": [], "source": [ "X_train, X_val, y_train, y_val = train_test_split(\n", " X, y,\n", " test_size=0.2,\n", " random_state=42,\n", " stratify=y\n", ")\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "hS5uVp_Jp_gJ", "outputId": "1b92b0ce-01ce-4adf-aa24-9341c42bea7a" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "X_train Shape : (33600, 28, 28, 1)\n", "y_train Shape : (33600,)\n", "X_test Shape : (28000, 784)\n", "y_test Shape : (8400,)\n" ] } ], "source": [ "print(\"X_train Shape :\", X_train.shape)\n", "print(\"y_train Shape :\", y_train.shape)\n", "print(\"X_test Shape :\", X_test.shape)\n", "print(\"y_test Shape :\", y_val.shape)" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "OmA9soj9qLHT" }, "outputs": [], "source": [ "\n", "\n", "# One-hot encode\n", "y_train = to_categorical(y_train, num_classes=10)\n", "y_val = to_categorical(y_val, num_classes=10)" ] }, { "cell_type": "markdown", "metadata": { "id": "HT4EuEXPqnXx" }, "source": [ "## **Visualization**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "icTJ0YXomi7B" }, "outputs": [], "source": [ "def show_samples(images, labels, title):\n", " plt.figure(figsize=(15,2))\n", " for i in range(10):\n", " plt.subplot(1,10,i+1)\n", " plt.imshow(images[i].reshape(28,28), cmap=\"gray\")\n", " plt.title(np.argmax(labels[i]))\n", " plt.axis(\"off\")\n", " plt.suptitle(title, fontsize=14, fontweight=\"bold\")\n", " plt.show()\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 160 }, "id": "S7P5ZWYxmkro", "outputId": "6af790e0-be96-4a53-d615-bd414f185778" }, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "show_samples(X_train, y_train, \"MNIST Training Samples\")\n" ] }, { "cell_type": "markdown", "metadata": { "id": "wrfNiaLMqvmy" }, "source": [ "## **Model Architecture Design**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "AmtWbi12mmp8" }, "outputs": [], "source": [ "def build_mnist_cnn():\n", " model = Sequential([\n", " Conv2D(32, (3,3), activation=\"relu\", padding=\"same\", input_shape=(28,28,1)),\n", " BatchNormalization(),\n", " MaxPooling2D(),\n", "\n", " Conv2D(64, (3,3), activation=\"relu\", padding=\"same\"),\n", " BatchNormalization(),\n", " MaxPooling2D(),\n", "\n", " Conv2D(128, (3,3), activation=\"relu\", padding=\"same\"),\n", " BatchNormalization(),\n", " MaxPooling2D(),\n", "\n", " Flatten(),\n", " Dense(128, activation=\"relu\"),\n", " Dropout(0.4),\n", "\n", " Dense(10, activation=\"softmax\")\n", " ])\n", "\n", " model.compile(\n", " optimizer=Adam(learning_rate=0.001),\n", " loss=\"categorical_crossentropy\",\n", " metrics=[\"accuracy\"]\n", " )\n", "\n", " return model\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 648 }, "id": "Pz3RDvlimoXA", "outputId": "08f9454a-6d38-4827-aeed-f180caf172ab" }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/usr/local/lib/python3.12/dist-packages/keras/src/layers/convolutional/base_conv.py:113: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n", " super().__init__(activity_regularizer=activity_regularizer, **kwargs)\n" ] }, { "data": { "text/html": [ "
Model: \"sequential\"\n",
              "
\n" ], "text/plain": [ "\u001b[1mModel: \"sequential\"\u001b[0m\n" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n",
              "┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃\n",
              "┑━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n",
              "β”‚ conv2d (Conv2D)                 β”‚ (None, 28, 28, 32)     β”‚           320 β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ batch_normalization             β”‚ (None, 28, 28, 32)     β”‚           128 β”‚\n",
              "β”‚ (BatchNormalization)            β”‚                        β”‚               β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ max_pooling2d (MaxPooling2D)    β”‚ (None, 14, 14, 32)     β”‚             0 β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ conv2d_1 (Conv2D)               β”‚ (None, 14, 14, 64)     β”‚        18,496 β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ batch_normalization_1           β”‚ (None, 14, 14, 64)     β”‚           256 β”‚\n",
              "β”‚ (BatchNormalization)            β”‚                        β”‚               β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ max_pooling2d_1 (MaxPooling2D)  β”‚ (None, 7, 7, 64)       β”‚             0 β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ conv2d_2 (Conv2D)               β”‚ (None, 7, 7, 128)      β”‚        73,856 β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ batch_normalization_2           β”‚ (None, 7, 7, 128)      β”‚           512 β”‚\n",
              "β”‚ (BatchNormalization)            β”‚                        β”‚               β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ max_pooling2d_2 (MaxPooling2D)  β”‚ (None, 3, 3, 128)      β”‚             0 β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ flatten (Flatten)               β”‚ (None, 1152)           β”‚             0 β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ dense (Dense)                   β”‚ (None, 128)            β”‚       147,584 β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ dropout (Dropout)               β”‚ (None, 128)            β”‚             0 β”‚\n",
              "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n",
              "β”‚ dense_1 (Dense)                 β”‚ (None, 10)             β”‚         1,290 β”‚\n",
              "β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜\n",
              "
\n" ], "text/plain": [ "┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n", "┃\u001b[1m \u001b[0m\u001b[1mLayer (type) \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1mOutput Shape \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1m Param #\u001b[0m\u001b[1m \u001b[0m┃\n", "┑━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n", "β”‚ conv2d (\u001b[38;5;33mConv2D\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m28\u001b[0m, \u001b[38;5;34m28\u001b[0m, \u001b[38;5;34m32\u001b[0m) β”‚ \u001b[38;5;34m320\u001b[0m β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ batch_normalization β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m28\u001b[0m, \u001b[38;5;34m28\u001b[0m, \u001b[38;5;34m32\u001b[0m) β”‚ \u001b[38;5;34m128\u001b[0m β”‚\n", "β”‚ (\u001b[38;5;33mBatchNormalization\u001b[0m) β”‚ β”‚ β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ max_pooling2d (\u001b[38;5;33mMaxPooling2D\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m32\u001b[0m) β”‚ \u001b[38;5;34m0\u001b[0m β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ conv2d_1 (\u001b[38;5;33mConv2D\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m64\u001b[0m) β”‚ \u001b[38;5;34m18,496\u001b[0m β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ batch_normalization_1 β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m14\u001b[0m, \u001b[38;5;34m64\u001b[0m) β”‚ \u001b[38;5;34m256\u001b[0m β”‚\n", "β”‚ (\u001b[38;5;33mBatchNormalization\u001b[0m) β”‚ β”‚ β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ max_pooling2d_1 (\u001b[38;5;33mMaxPooling2D\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m64\u001b[0m) β”‚ \u001b[38;5;34m0\u001b[0m β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ conv2d_2 (\u001b[38;5;33mConv2D\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m128\u001b[0m) β”‚ \u001b[38;5;34m73,856\u001b[0m β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ batch_normalization_2 β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m7\u001b[0m, \u001b[38;5;34m128\u001b[0m) β”‚ \u001b[38;5;34m512\u001b[0m β”‚\n", "β”‚ (\u001b[38;5;33mBatchNormalization\u001b[0m) β”‚ β”‚ β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ max_pooling2d_2 (\u001b[38;5;33mMaxPooling2D\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m3\u001b[0m, \u001b[38;5;34m3\u001b[0m, \u001b[38;5;34m128\u001b[0m) β”‚ \u001b[38;5;34m0\u001b[0m β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ flatten (\u001b[38;5;33mFlatten\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m1152\u001b[0m) β”‚ \u001b[38;5;34m0\u001b[0m β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ dense (\u001b[38;5;33mDense\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m) β”‚ \u001b[38;5;34m147,584\u001b[0m β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ dropout (\u001b[38;5;33mDropout\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m) β”‚ \u001b[38;5;34m0\u001b[0m β”‚\n", "β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€\n", "β”‚ dense_1 (\u001b[38;5;33mDense\u001b[0m) β”‚ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m10\u001b[0m) β”‚ \u001b[38;5;34m1,290\u001b[0m β”‚\n", "β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜\n" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
 Total params: 242,442 (947.04 KB)\n",
              "
\n" ], "text/plain": [ "\u001b[1m Total params: \u001b[0m\u001b[38;5;34m242,442\u001b[0m (947.04 KB)\n" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
 Trainable params: 241,994 (945.29 KB)\n",
              "
\n" ], "text/plain": [ "\u001b[1m Trainable params: \u001b[0m\u001b[38;5;34m241,994\u001b[0m (945.29 KB)\n" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "
 Non-trainable params: 448 (1.75 KB)\n",
              "
\n" ], "text/plain": [ "\u001b[1m Non-trainable params: \u001b[0m\u001b[38;5;34m448\u001b[0m (1.75 KB)\n" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "model = build_mnist_cnn()\n", "model.summary()\n" ] }, { "cell_type": "markdown", "metadata": { "id": "gUcQsS0gq1sT" }, "source": [ "## **Model Training**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "89xyYcHZmqAf", "outputId": "68042079-a0bf-4de1-cde2-93ade8f9f299" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Epoch 1/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m89s\u001b[0m 162ms/step - accuracy: 0.8449 - loss: 0.5536 - val_accuracy: 0.9702 - val_loss: 0.0932\n", "Epoch 2/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m80s\u001b[0m 152ms/step - accuracy: 0.9758 - loss: 0.0804 - val_accuracy: 0.9869 - val_loss: 0.0374\n", "Epoch 3/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m81s\u001b[0m 151ms/step - accuracy: 0.9849 - loss: 0.0524 - val_accuracy: 0.9876 - val_loss: 0.0447\n", "Epoch 4/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m81s\u001b[0m 149ms/step - accuracy: 0.9860 - loss: 0.0453 - val_accuracy: 0.9856 - val_loss: 0.0451\n", "Epoch 5/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m83s\u001b[0m 150ms/step - accuracy: 0.9885 - loss: 0.0365 - val_accuracy: 0.9876 - val_loss: 0.0437\n", "Epoch 6/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m83s\u001b[0m 152ms/step - accuracy: 0.9912 - loss: 0.0300 - val_accuracy: 0.9842 - val_loss: 0.0566\n", "Epoch 7/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m82s\u001b[0m 155ms/step - accuracy: 0.9912 - loss: 0.0306 - val_accuracy: 0.9906 - val_loss: 0.0374\n", "Epoch 8/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m80s\u001b[0m 151ms/step - accuracy: 0.9931 - loss: 0.0218 - val_accuracy: 0.9864 - val_loss: 0.0602\n", "Epoch 9/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m82s\u001b[0m 152ms/step - accuracy: 0.9942 - loss: 0.0187 - val_accuracy: 0.9857 - val_loss: 0.0609\n", "Epoch 10/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m80s\u001b[0m 152ms/step - accuracy: 0.9922 - loss: 0.0243 - val_accuracy: 0.9900 - val_loss: 0.0444\n", "Epoch 11/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m83s\u001b[0m 154ms/step - accuracy: 0.9950 - loss: 0.0162 - val_accuracy: 0.9900 - val_loss: 0.0461\n", "Epoch 12/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m81s\u001b[0m 154ms/step - accuracy: 0.9936 - loss: 0.0199 - val_accuracy: 0.9883 - val_loss: 0.0492\n", "Epoch 13/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m81s\u001b[0m 154ms/step - accuracy: 0.9939 - loss: 0.0199 - val_accuracy: 0.9881 - val_loss: 0.0493\n", "Epoch 14/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m81s\u001b[0m 154ms/step - accuracy: 0.9949 - loss: 0.0150 - val_accuracy: 0.9898 - val_loss: 0.0467\n", "Epoch 15/15\n", "\u001b[1m525/525\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m80s\u001b[0m 152ms/step - accuracy: 0.9966 - loss: 0.0124 - val_accuracy: 0.9905 - val_loss: 0.0577\n" ] } ], "source": [ "history = model.fit(\n", " X_train, y_train,\n", " epochs=15,\n", " batch_size=64,\n", " validation_data=(X_val, y_val),\n", " verbose=1\n", ")\n" ] }, { "cell_type": "markdown", "metadata": { "id": "dyOmo_mTq9bN" }, "source": [ "## **Model Evaluation**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "NNIIKeHYmrz2" }, "outputs": [], "source": [ "def plot_training_history(history):\n", " epochs = range(1, len(history.history[\"accuracy\"]) + 1)\n", "\n", " plt.figure(figsize=(14,5))\n", "\n", " plt.subplot(1,2,1)\n", " plt.plot(epochs, history.history[\"accuracy\"], label=\"Train\")\n", " plt.plot(epochs, history.history[\"val_accuracy\"], label=\"Validation\")\n", " plt.title(\"Accuracy Curve\")\n", " plt.xlabel(\"Epochs\")\n", " plt.ylabel(\"Accuracy\")\n", " plt.legend()\n", " plt.grid(True)\n", "\n", " plt.subplot(1,2,2)\n", " plt.plot(epochs, history.history[\"loss\"], label=\"Train\")\n", " plt.plot(epochs, history.history[\"val_loss\"], label=\"Validation\")\n", " plt.title(\"Loss Curve\")\n", " plt.xlabel(\"Epochs\")\n", " plt.ylabel(\"Loss\")\n", " plt.legend()\n", " plt.grid(True)\n", "\n", " plt.show()\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 427 }, "id": "3gZR7enrmtia", "outputId": "90d50231-dd4b-4d36-a5db-680d5f2e7b44" }, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plot_training_history(history)\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "_J2W394AmwO5", "outputId": "fe4d1e78-8247-40b1-d666-11a271ca39ec" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[1m263/263\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m7s\u001b[0m 24ms/step\n", "Validation Accuracy: 0.9905\n" ] } ], "source": [ "y_val_pred = np.argmax(model.predict(X_val), axis=1)\n", "y_val_true = np.argmax(y_val, axis=1)\n", "\n", "accuracy = accuracy_score(y_val_true, y_val_pred)\n", "print(f\"Validation Accuracy: {accuracy:.4f}\")\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "hGxlF0b4mx76", "outputId": "42235619-c5a8-4fb2-932c-f62a5f9bb160" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ " precision recall f1-score support\n", "\n", " 0 1.00 0.99 0.99 827\n", " 1 0.99 1.00 1.00 937\n", " 2 0.98 1.00 0.99 835\n", " 3 1.00 0.98 0.99 870\n", " 4 0.98 1.00 0.99 814\n", " 5 0.99 0.99 0.99 759\n", " 6 0.99 0.99 0.99 827\n", " 7 0.99 0.99 0.99 880\n", " 8 0.99 0.99 0.99 813\n", " 9 0.99 0.98 0.98 838\n", "\n", " accuracy 0.99 8400\n", " macro avg 0.99 0.99 0.99 8400\n", "weighted avg 0.99 0.99 0.99 8400\n", "\n" ] } ], "source": [ "print(classification_report(y_val_true, y_val_pred))\n" ] }, { "cell_type": "markdown", "metadata": { "id": "ETMSCBGqrGY9" }, "source": [ "## **Prediction**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "jvFqpDUSmzcY", "outputId": "838ac1d7-1726-4abb-8a15-e97e562577a3" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[1m875/875\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 21ms/step\n" ] } ], "source": [ "test_predictions = np.argmax(model.predict(X_test_final), axis=1)\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "AtNEqgOLm144" }, "outputs": [], "source": [ "submission = pd.DataFrame({\n", " \"ImageId\": np.arange(1, len(test_predictions) + 1),\n", " \"Label\": test_predictions\n", "})\n", "\n", "submission.to_csv(\"submission.csv\", index=False)\n" ] }, { "cell_type": "markdown", "metadata": { "id": "U4Dqpj3FrhJh" }, "source": [ "## **Model Saving**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "RaSDXWXZt0S4" }, "outputs": [], "source": [ "model.save(\"cnn_model.keras\")\n" ] }, { "cell_type": "markdown", "metadata": { "id": "yhrsaPpOUNCm" }, "source": [ "## **Conclusion**" ] }, { "cell_type": "markdown", "metadata": { "id": "B5HbR8HCUQzu" }, "source": [ "\n", "- CNN model achieves **99.05% validation accuracy** on the MNIST dataset.\n", "- **High precision, recall, and F1-scores (~0.99)** across all digit classes.\n", "- Performs consistently across all digits (0–9) with minimal misclassification.\n", "- **Robust and reliable** for real-time digit recognition tasks.\n", "- Suitable for applications like **interactive UIs, handwriting recognition, and educational tools**.\n" ] }, { "cell_type": "markdown", "metadata": { "id": "i-BrxksDUMBH" }, "source": [ "###" ] } ], "metadata": { "colab": { "provenance": [] }, "kernelspec": { "display_name": "ml_projects", "language": "python", "name": "python3" }, "language_info": { "name": "python", "version": "3.10.0" } }, "nbformat": 4, "nbformat_minor": 0 }