Project: UMAP vs t-SNE 📊
Overview:
This project compares Uniform Manifold Approximation and Projection (UMAP) with t-distributed Stochastic Neighbor Embedding (t-SNE) for dimension reduction tasks. UMAP, a formidable competitor to t-SNE, excels in preserving both local and global structure in high-dimensional data.
How UMAP Works:
UMAP constructs a fuzzy topological representation, optimizing an objective function to balance local and global structure preservation. It achieves this through graph construction and embedding techniques.
Testing and Comparison:
We’ll test UMAP and t-SNE on diverse datasets, evaluating visualization quality, structure preservation, and computational efficiency. The comparison aims to highlight each algorithm’s strengths and weaknesses across different data types.
Impact of Parameters:
Exploring parameters like number of neighbors and distance metrics, we’ll assess their impact on UMAP’s performance.
Contributing 🤝
Contributions are highly encouraged! If you have suggestions, improvements, or feature requests, feel free to reach out to me !
License 📝
This project is licensed under the MIT License - see the LICENSE file for details.
Developed by Pierre LAGUE and Ilian VANDENBERGHE at the University of Lille, France. 🚀📊