Visualizing Data using GTSNE

TL;DR

GTSNE is a novel visualization method that improves upon t-SNE and UMAP by better capturing both local and macro structures in high-dimensional data, especially on low-dimensional manifolds.

Contribution

The paper introduces GTSNE, a new variation of t-SNE that enhances visualization quality by preserving macro structures in high-dimensional data.

Findings

GTSNE outperforms t-SNE and UMAP in macro structure preservation across multiple datasets.

GTSNE produces more accurate visualizations of high-dimensional data.

The method effectively captures both local and global data structures.

Abstract

We present a new method GTSNE to visualize high-dimensional data points in the two dimensional map. The technique is a variation of t-SNE that produces better visualizations by capturing both the local neighborhood structure and the macro structure in the data. This is particularly important for high-dimensional data that lie on continuous low-dimensional manifolds. We illustrate the performance of GTSNE on a wide variety of datasets and compare it the state of art methods, including t-SNE and UMAP. The visualizations produced by GTSNE are better than those produced by the other techniques on almost all of the datasets on the macro structure preservation.