Skeleton Clustering: Dimension-Free Density-based Clustering

TL;DR

Skeleton clustering is a novel density-based method designed to effectively identify clusters in high-dimensional and irregular data by using surrogate density measures and a hybrid framework combining prototypes, density, and hierarchy.

Contribution

It introduces a dimension-free density measure and a new clustering framework that improves cluster detection in high-dimensional data.

Findings

Reliable clustering in high-dimensional scenarios

Effective detection of irregularly shaped clusters

Theoretical and empirical validation of the method

Abstract

We introduce a density-based clustering method called skeleton clustering that can detect clusters in multivariate and even high-dimensional data with irregular shapes. To bypass the curse of dimensionality, we propose surrogate density measures that are less dependent on the dimension but have intuitive geometric interpretations. The clustering framework constructs a concise representation of the given data as an intermediate step and can be thought of as a combination of prototype methods, density-based clustering, and hierarchical clustering. We show by theoretical analysis and empirical studies that the skeleton clustering leads to reliable clusters in multivariate and high-dimensional scenarios.