Multilayer Perceptron for Handwritten Digit Recognition

Inspired by Neural Networks and Deep Learning by Michael Nielsen.(http://neuralnetworksanddeeplearning.com)

Overview

This project implements one of the neuronal networks – Multilayer Perceptron (MLP) for handwritten digit recognition, based on concepts from Neural Networks and Deep Learning by Michael Nielsen. The model is trained on the MNIST dataset, which consists of handwritten digits from 0 to 9.

Requirements

Make sure you have the following dependencies installed:

• Python 3.x
• numpy (for numerical computations)

Installation

Clone the repository:

git clone https://github.com/t3i8m/Multilayer-Perceptron.git
cd Multilayer-Perceptron

Install dependencies:

pip install numpy 

Usage

Run the program using main.py:

python main.py

Dataset

This project uses the MNIST dataset, a collection of 60,000 training images and 10,000 test images of handwritten digits (0-9). Each image is 28x28 pixels in grayscale.

Structure

Multilayer-Perceptron/
│── __pycache__/             # Compiled Python files
│── data/                    # Dataset storage
│   ├── mnist.pkl.gz         # MNIST dataset in pickle format
│── nn/                      # Neural network (scratch implementation)
│   ├── __pycache__/         # Compiled Python files
│   ├── MLP_scratch.py       # MLP implementation from scratch
│── nn_oop/                  # Object-Oriented Neural Network implementation
│   ├── __pycache__/         # Compiled Python files
│   ├── __init__.py          # Package initialization
│   ├── layer.py             # Layer implementation
│   ├── MLP.py               # Main MLP class
│   ├── neuron.py            # Neuron implementation
│── utils/                   # Utility functions
│   ├── __pycache__/         # Compiled Python files
│   ├── __init__.py          # Package initialization
│   ├── activation_loss.py   # Activation functions and loss calculations
│   ├── mnist_loader.py      # MNIST dataset loader
│── main.py                  # Main script to run the model
│── README.md                # Project documentation

License

MIT License