Hi, welcome to mnist_ML

This repository contains my Jupyter Notebook analysis of several machine learning models for classifying handwritten digits from the MNIST dataset. The analysis explores a variety of models ranging from t-SNE for dimensionality reduction to deep learning with ResNet-9.

Table of Contents

• Overview
• Requirements
• Installation
• Running the Notebook
• Models Analyzed
• Contributing
• License

---

Overview

This project is focused on comparing different machine learning models to classify the MNIST dataset, a well-known benchmark for handwritten digit classification. The models analyzed include:

• t-SNE (t-Distributed Stochastic Neighbor Embedding) for dimensionality reduction
• Traditional Machine Learning models such as Random Forest
• Deep learning models including ResNet-9

The results are visualized through various plots to assess the performance of each model and compare their effectiveness in classifying the digits.

---

Requirements

To run the notebook, you'll need the following libraries installed:

matplotlib
numpy
pandas
seaborn
tensorflow
openTSNE
scipy
scikit-learn

---

Installation

1. Clone the repository:

   git clone https://github.com/kyracho/mnist-classification.git
   cd mnist-classification

2. Set up a virtual environment (optional):

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

3. Install the required packages:

   pip install -r requirements.txt

Alternatively, install the packages manually:

pip install matplotlib numpy pandas seaborn tensorflow openTSNE scipy scikit-learn

---

Running the Notebook

1. Launch Jupyter Notebook:

   jupyter notebook

2. Open the mnist_classification.ipynb file from the Jupyter interface.

3. Run the cells in the notebook to train and evaluate various machine learning models.