KR2026Proceedings of the 23rd International Conference on Principles of Knowledge Representation and ReasoningProceedings of the 23rd International Conference on Principles of Knowledge Representation and Reasoning

Lisbon, Portugal. July 20-23, 2026.

Edited by
  1. Renata Wassermann
  2. Marie-Laure Mugnier
  3. Franz Baader

ISSN: 2334-1033
ISBN: 978-1-956792-18-8

Sponsored by
Published by

Copyright © 2026 International Joint Conferences on Artificial Intelligence Organization

A Rectification-Based Approach for Distilling Boosted Trees into Decision Trees

  1. Gilles Audemard(Univ. Artois, CNRS, CRIL, Lens, France)
  2. Sylvie Coste-Marquis(Univ. Artois, CNRS, CRIL, Lens, France)
  3. Pierre Marquis(Univ. Artois, CNRS, CRIL, Lens, France, Institut Universitaire de France, France)
  4. Mehdi Sabiri(Univ. Artois, CNRS, CRIL, Lens, France)
  5. Nicolas Szczepanski(Univ. Artois, CNRS, CRIL, Lens, France)
PDF
BibTeX
https://doi.org/10.24963/kr.2026/76

Keywords

  1. null-XAI
  2. null-Knowledge Distillation
  3. null-Model Correction
  4. null-Decision Trees
  5. null-Boosted Trees

Abstract

We present a new approach for distilling boosted trees into decision trees, in the objective of generating an ML model offering an acceptable compromise in terms of predictive performance and interpretability. We explain how the correction approach called rectification

can be used to implement such a distillation process. We show empirically that this approach provides interesting results, in comparison with an approach to distillation achieved by retraining the model.