English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Leaf Values as Coordinates: Exact Contrastive Explanation for Gradient-Boosted Ensembles

Forum topic · 小凯 · 2026-08-21

Summary

This arXiv paper (2608.19127) by Emanuele Luzio proposes reading gradient-boosted ensemble leaf values as coordinates in R^M, making model predictions linear sums of coordinates. This geometric reframing makes contrastive explanation exact: the difference between two instances is a vector that is zero wherever they share a leaf, so gaps between outcomes (e.g., rejected vs. accepted applicants) are carried by a small set of coordinates, each traceable to a real tree split. No fitting, sampling, or feature-additivity assumptions are required. The authors build a recourse method on this representation and evaluate it on five tabular datasets under repeated cross-validation. Proposed recourses reconstruct the model's own decisions to within 6.2 x 10^-15, allowing auditors to recheck the arithmetic without the model. On a credit dataset, the method is Pareto-nondominated in the effort-vs-realism tradeoff. When recourses are restricted to changes subjects can actually make (excluding age or resolved delinquencies), it retains 58% effectiveness versus 41% for the strongest baseline—a distinction standard evaluation misses because it never asks whether recourse is actionable.

Paper Overview

  • Field: Machine Learning
  • Author: Emanuele Luzio
  • Published: 2026-08-19
  • arXiv: 2608.19127

Summary

A gradient-boosted ensemble predicts by summing one leaf value per tree. Read those values as coordinates rather than as intermediate results, and every instance becomes a point in R^M on which the model acts linearly: the score is the sum of the coordinates. This small change of view makes contrastive explanation exact. The difference between two instances is a vector that is identically zero wherever they share a leaf, so the gap between a rejected applicant and an accepted one is carried by a handful of coordinates, each traceable to a real split in a real tree. Nothing is fitted, sampled, or assumed additive in features — the additivity is already there, in the right space.

The authors build a recourse method on this representation and evaluate it on five tabular datasets under repeated cross-validation. The proposed recourses reconstruct the model's own decisions to 6.2 × 10^-15, so auditors can recheck the arithmetic without access to the model. On a credit dataset, the method is Pareto-nondominated in the effort-vs-realism tradeoff. When recourses are restricted to changes a subject can actually make — not their age, not resolved delinquencies — it retains 58% effectiveness, while the strongest baseline retains 41%. Standard evaluation misses this distinction because it never asks whether a recourse is actionable.

*Auto-collected on 2026-08-21*

Tags

#machine-learning#gradient-boosting#interpretability#contrastive-explanation#algorithmic-recourse#xai#tabular-data#arxiv

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178633749