Tree-SNE: Hierarchical Clustering and Visualization Using t-SNE

TL;DR

Tree-SNE combines hierarchical clustering with t-SNE visualization using stacked one-dimensional embeddings, and introduces alpha-clustering for optimal, scale-invariant cluster detection, demonstrated on biological and image data.

Contribution

The paper introduces tree-SNE, a novel hierarchical clustering and visualization method based on stacked t-SNE embeddings, along with alpha-clustering for automatic cluster determination.

Findings

Effective hierarchical clustering and visualization demonstrated on biological and image datasets.

Alpha-clustering accurately determines optimal clusters without prior knowledge.

Competitive unsupervised clustering results achieved on multiple datasets.

Abstract

t-SNE and hierarchical clustering are popular methods of exploratory data analysis, particularly in biology. Building on recent advances in speeding up t-SNE and obtaining finer-grained structure, we combine the two to create tree-SNE, a hierarchical clustering and visualization algorithm based on stacked one-dimensional t-SNE embeddings. We also introduce alpha-clustering, which recommends the optimal cluster assignment, without foreknowledge of the number of clusters, based off of the cluster stability across multiple scales. We demonstrate the effectiveness of tree-SNE and alpha-clustering on images of handwritten digits, mass cytometry (CyTOF) data from blood cells, and single-cell RNA-sequencing (scRNA-seq) data from retinal cells. Furthermore, to demonstrate the validity of the visualization, we use alpha-clustering to obtain unsupervised clustering results competitive with the state of the art on several image data sets. Software is available at https://github.com/isaacrob/treesne.