Handwritten Digit Recognition with Neural Networks (MNIST OCR)

Project Overview

This project implements a neural network to recognize handwritten digits from the MNIST dataset. The aim is to create a proof-of-concept model for Optical Character Recognition (OCR) systems using deep learning. This project demonstrates the end-to-end process of dataset preprocessing, building, training, and tuning a neural network to achieve a classification accuracy of at least 90% on the test set.

Project Files

• handwritten_digit_recognition.ipynb: The Jupyter Notebook containing the code for preprocessing, model building, training, evaluation, and saving the model.
• trained_model.pth: The saved trained model, which can be loaded for future inference or analysis.

Project Instructions

The project is divided into several steps:

Step 1: Dataset Loading and Preprocessing

• Dataset: The MNIST handwritten digit dataset is loaded from torchvision.datasets.
• Preprocessing: The dataset is converted to PyTorch tensors and normalized. Input images are flattened into vectors for input to the neural network.
• DataLoader: A PyTorch DataLoader is used to batch the data for efficient training and testing.

Step 2: Data Visualization and Exploration

• Visualize the dataset using a helper function. Explore the size and shape of the data both before and after preprocessing.
• Briefly explain and justify preprocessing choices, such as normalization and flattening.

Step 3: Neural Network Model

• Model: A fully connected neural network is built using PyTorch to classify input digits.
• Optimizer: The model's weights are updated during training using a selected optimizer, such as stochastic gradient descent (SGD) or Adam.
• Training: The network is trained on the training set using the prepared DataLoader.

Step 4: Evaluation and Model Tuning

• Evaluation: The accuracy of the model is evaluated on the test set.
• Tuning: Hyperparameters such as learning rate, batch size, and network architecture are tuned to achieve an accuracy of at least 90% on the test set.

Step 5: Saving the Model

• The trained model is saved using torch.save for reuse in future applications.

Requirements

This project requires the following dependencies:

• Python 3.x
• Jupyter Notebook
• PyTorch
• Torchvision
• Matplotlib (for visualization)

You can install the dependencies using pip:

pip install torch torchvision matplotlib

How to Run

1. Clone the repository:

git clone <repository-url>
cd <repository-folder>

2. Open the Jupyter Notebook:

jupyter notebook handwritten_digit_recognition.ipynb

3. Run the notebook cells step-by-step to load the data, train the model, and evaluate its performance.