English

Dynamic Similarity Graph Construction with Kernel Density Estimation

Data Structures and Algorithms 2025-07-03 v1 Machine Learning

Abstract

In the kernel density estimation (KDE) problem, we are given a set XX of data points in Rd\mathbb{R}^d, a kernel function k:Rd×RdRk: \mathbb{R}^d \times \mathbb{R}^d \rightarrow \mathbb{R}, and a query point qRd\mathbf{q} \in \mathbb{R}^d, and the objective is to quickly output an estimate of xXk(q,x)\sum_{\mathbf{x} \in X} k(\mathbf{q}, \mathbf{x}). In this paper, we consider KDE\textsf{KDE} in the dynamic setting, and introduce a data structure that efficiently maintains the estimates for a set of query points as data points are added to XX over time. Based on this, we design a dynamic data structure that maintains a sparse approximation of the fully connected similarity graph on XX, and develop a fast dynamic spectral clustering algorithm. We further evaluate the effectiveness of our algorithms on both synthetic and real-world datasets.

Keywords

Cite

@article{arxiv.2507.01696,
  title  = {Dynamic Similarity Graph Construction with Kernel Density Estimation},
  author = {Steinar Laenen and Peter Macgregor and He Sun},
  journal= {arXiv preprint arXiv:2507.01696},
  year   = {2025}
}

Comments

ICML'25