Fast unsupervised ground metric learning with tree-Wasserstein distance

TL;DR

This paper introduces a fast, scalable method for unsupervised ground metric learning using tree-Wasserstein distance, improving computational efficiency while maintaining approximation quality, demonstrated on genomics data for cell clustering.

Contribution

Proposes a novel tree-embedded Wasserstein singular vector method that reduces computational complexity and retains approximation quality for unsupervised ground metric learning.

Findings

Algorithm converges to better approximation than existing methods.

Achieves $ ext{O}(n^3 + m^3 + mn)$ complexity, faster than previous approaches.

Demonstrates scalability and utility on single-cell RNA sequencing datasets.

Abstract

The performance of unsupervised methods such as clustering depends on the choice of distance metric between features, or ground metric. Commonly, ground metrics are decided with heuristics or learned via supervised algorithms. However, since many interesting datasets are unlabelled, unsupervised ground metric learning approaches have been introduced. One promising option employs Wasserstein singular vectors (WSVs), which emerge when computing optimal transport distances between features and samples simultaneously. WSVs are effective, but can be prohibitively computationally expensive in some applications: $\mathcal{O}(n^2m^2(n \log(n) + m \log(m))$ for $n$ samples and $m$ features. In this work, we propose to augment the WSV method by embedding samples and features on trees, on which we compute the tree-Wasserstein distance (TWD). We demonstrate theoretically and empirically that the algorithm converges to a better approximation of the standard WSV approach than the best known alternatives, and does so with $\mathcal{O}(n^3+m^3+mn)$ complexity. In addition, we prove that the initial tree structure can be chosen flexibly, since tree geometry does not constrain the richness of the approximation up to the number of edge weights. This proof suggests a fast and recursive algorithm for computing the tree parameter basis set, which we find crucial to realising the efficiency gains at scale. Finally, we employ the tree-WSV algorithm to several single-cell RNA sequencing genomics datasets, demonstrating its scalability and utility for unsupervised cell-type clustering problems. These results poise unsupervised ground metric learning with TWD as a low-rank approximation of WSV with the potential for widespread application.