Efficient search for relevance explanations using MAP-independence in Bayesian networks

Enrique Valero-Leal, Concha Bielza, Pedro Larranaga, Silja Renooij*

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

[Formula presented]-independence is a novel concept concerned with explaining the (ir)relevance of intermediate nodes for maximum a posteriori ([Formula presented]) computations in Bayesian networks. Building upon properties of [Formula presented]-independence, we introduce and experiment with methods for finding sets of relevant nodes using both an exhaustive and a heuristic approach. Our experiments show that these properties significantly speed up run time for both approaches. In addition, we link [Formula presented]-independence to defeasible reasoning, a type of reasoning that analyses how new evidence may invalidate an already established conclusion. Ways to present users with an explanation using [Formula presented]-independence are also suggested.

Original languageEnglish
Article number108965
Number of pages22
JournalInternational Journal of Approximate Reasoning
Volume160
DOIs
Publication statusPublished - Sept 2023

Keywords

  • Bayesian network
  • Defeasible reasoning
  • Explainability
  • Map-independence
  • Relevance
  • Robustness

Fingerprint

Dive into the research topics of 'Efficient search for relevance explanations using MAP-independence in Bayesian networks'. Together they form a unique fingerprint.

Cite this