Harnessing Data Asymmetry: Manifold Learning in the Finsler World

TL;DR

This paper introduces a novel manifold learning approach using Finsler geometry to effectively capture and embed asymmetric dissimilarities in data, improving the preservation of data structure and revealing previously hidden information.

Contribution

It proposes a Finsler manifold learning pipeline that generalizes existing methods to handle asymmetry, broadening applicability beyond traditional symmetric embeddings.

Findings

Finsler-based embeddings outperform Euclidean methods on synthetic and real datasets.

The approach reveals density hierarchies and asymmetric features lost in traditional methods.

Superior embedding quality demonstrated across multiple datasets.

Abstract

Manifold learning is a fundamental task at the core of data analysis and visualisation. It aims to capture the simple underlying structure of complex high-dimensional data by preserving pairwise dissimilarities in low-dimensional embeddings. Traditional methods rely on symmetric Riemannian geometry, thus forcing symmetric dissimilarities and embedding spaces, e.g. Euclidean. However, this discards in practice valuable asymmetric information inherent to the non-uniformity of data samples. We suggest to harness this asymmetry by switching to Finsler geometry, an asymmetric generalisation of Riemannian geometry, and propose a Finsler manifold learning pipeline that constructs asymmetric dissimilarities and embeds in a Finsler space. This greatly broadens the applicability of existing asymmetric embedders beyond traditionally directed data to any data. We also modernise asymmetric embedders by generalising current reference methods to asymmetry, like Finsler t-SNE and Finsler Umap. On controlled synthetic and large real datasets, we show that our asymmetric pipeline reveals valuable information lost in the traditional pipeline, e.g. density hierarchies, and consistently provides superior quality embeddings than their Euclidean counterparts.