English

Efficient Algorithms for Generating Provably Near-Optimal Cluster Descriptors for Explainability

Data Structures and Algorithms 2020-02-10 v1 Artificial Intelligence Discrete Mathematics Optimization and Control

Abstract

Improving the explainability of the results from machine learning methods has become an important research goal. Here, we study the problem of making clusters more interpretable by extending a recent approach of [Davidson et al., NeurIPS 2018] for constructing succinct representations for clusters. Given a set of objects SS, a partition π\pi of SS (into clusters), and a universe TT of tags such that each element in SS is associated with a subset of tags, the goal is to find a representative set of tags for each cluster such that those sets are pairwise-disjoint and the total size of all the representatives is minimized. Since this problem is NP-hard in general, we develop approximation algorithms with provable performance guarantees for the problem. We also show applications to explain clusters from datasets, including clusters of genomic sequences that represent different threat levels.

Keywords

Cite

@article{arxiv.2002.02487,
  title  = {Efficient Algorithms for Generating Provably Near-Optimal Cluster Descriptors for Explainability},
  author = {Prathyush Sambaturu and Aparna Gupta and Ian Davidson and S. S. Ravi and Anil Vullikanti and Andrew Warren},
  journal= {arXiv preprint arXiv:2002.02487},
  year   = {2020}
}
R2 v1 2026-06-23T13:33:33.236Z