MNIST-Classification

This repository contains a neural network model implemented in Python for classifying handwritten digits from the MNIST dataset. The model is developed using the PyTorch framework and is presented in a Jupyter Notebook for ease of understanding and experimentation.

Contents

• net.ipynb: Jupyter Notebook detailing the model architecture, training process, and evaluation metrics.
• LICENSE: MIT License under which this project is distributed.

Features

• Data Loading: Utilizes PyTorch's torchvision library to load and preprocess the MNIST dataset.
• Model Architecture: Implements a CNN with layers optimized for digit recognition tasks.
• Training: Includes code for training the model with appropriate loss functions and optimizers.
• Evaluation: Provides methods to assess the model's performance on test data.

Model Training Details

• Loss Function: CrossEntropyLoss
• Optimizer: Adam
• Learning Rate: 0.001
• Batch Size: 32
• Number of Epochs: 10

Network Architecture

• Input Layer: 28x28 grayscale images
• Convolutional Layer 1: 32 filters (3x3)
• MaxPooling Layer 1: 2x2
• Convolutional Layer 2: 64 filters (3x3)
• MaxPooling Layer 2: 2x2
• Fully Connected Layer 1: 1600 → 128
• Fully Connected Layer 2: 128 → 10 (output)

Requirements

• Python 3.x
• PyTorch
• torchvision
• NumPy
• Jupyter Notebook
• MatPlotLib

Installation

1. Clone the repository
   git clone https://github.com/esinocchi/MNIST-Classification.git
2. Install the recquried packages:
   pip install torch torchvision numpy matplotlib jupyter
3. Run the Jupyter Notebook:
   jupyter notebook net.ipynb