Relative Wasserstein Spatial Depth for Cluster Number Selection in Distributional Data

Key points

• Contemporary data in many scientific domains, such as images, media streams, biomedical omics, are naturally modeled as probability distributions in Wasserstein space instead of points in Euclidean space.1
• Clustering distribution-valued data in Wasserstein space often requires choosing the number of clusters in advance.1
• We pair this criterion with Wasserstein K-means or a proposed K-medians algorithm.1
• Applications to MNIST images and flow cytometry data illustrate the practical value of RWSD.1

Sources (1)

• [1]Relative Wasserstein Spatial Depth for Cluster Number Selection in Distributional Data
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 09:53 PM

  We propose Relative Wasserstein Spatial Depth (RWSD) as a selection criterion for optimal number of clusters.

  Contemporary data in many scientific domains, such as images, media streams, biomedical omics, are naturally modeled as probability distributions in Wasserstein space instead of points in Euclidean space.

Extractive summary: sentences quoted from the sources.