ARTFEED — Contemporary Art Intelligence

TNStream: Tightest Neighbors for Multi-Density Data Stream Clustering

other · 2026-05-07

The recently introduced clustering algorithm, TNStream, tackles issues related to data stream clustering by effectively managing high-dimensional, multi-density data with arbitrary shapes and robust outlier resistance. It employs an innovative approach known as Tightest Neighbors alongside a Skeleton Set theory to dynamically establish the clustering radius based on local similarities. This method captures the progression of multi-density data streams through micro-clusters and utilizes a Tightest Neighbors-based technique to create the final clusters. TNStream operates as a fully online algorithm aimed at enhancing efficiency within high-dimensional streaming data contexts. This research was made available on arXiv under ID 2505.00359.

Key facts

  • TNStream is a fully online clustering algorithm.
  • It uses the concept of Tightest Neighbors and Skeleton Set theory.
  • The algorithm adaptively determines clustering radius based on local similarity.
  • It handles arbitrarily shaped, multi-density, high-dimensional data.
  • The algorithm maintains strong outlier resistance.
  • It summarizes data stream evolution in micro-clusters.
  • The paper is available on arXiv with ID 2505.00359.
  • The algorithm aims to improve efficiency in high-dimensional streaming data.

Entities

Institutions

  • arXiv

Sources