Stable and Consistent Density-Based Clustering

Explore a consistent approach to density-based clustering in this 28-minute conference talk by Alexander Rolle. Delve into a 3-parameter hierarchical clustering method for metric probability spaces, examining its stability and how it relates to well-known hierarchical clustering techniques. Learn about the correspondence interleaving distance and robust linkage, and understand the stability theorem that holds without distributional assumptions. Discover how taking 1-parameter slices that are neither horizontal nor vertical leads to a stable and consistent hierarchical clustering algorithm. The talk covers an introduction to density-based clustering, the correspondence interleaving distance, robust linkage, the stability theorem, and concludes with final remarks on the topic.