Skeleton Clustering: Dimension-Free Density-based Clustering
This addresses the challenge of density-based clustering in high-dimensional spaces for data scientists, though it appears incremental as it combines existing approaches like prototype methods and hierarchical clustering.
The paper tackles the problem of detecting clusters in multivariate and high-dimensional data with irregular shapes by introducing skeleton clustering, a method that uses surrogate density measures to bypass the curse of dimensionality, and shows through theoretical analysis and empirical studies that it leads to reliable clusters.
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.