---
title: "叶值即坐标：梯度提升集成的精确对比解释方法"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.19127v1"
permalink: "https://feed.astralor.com/items/itm_ec322f253f4419444cbd2554"
published: "2026-08-19T17:20:33.000Z"
score: 75
channel: "ai"
tags: ["可解释AI","可逆建议","对比解释","梯度提升集成","表格数据"]
---

# 叶值即坐标：梯度提升集成的精确对比解释方法

## AI 点评

> 这篇论文的可贵之处在于用最微小的视角转换撬动了一个硬问题，让对比解释从启发式走向精确验证。对于风控建模、反欺诈和贷款审批的从业者，它不仅是学术成果，更是能直接落地审计流程的工具。我尤其欣赏它强调"建议能否被执行"——这戳中了当前可解释AI评价体系的一个盲点。

## 速览

论文提出将梯度提升集成模型中每棵树的叶值视为坐标，而非中间结果，从而使每个实例映射为R^M空间中的点，模型在该空间呈线性操作。这种视角转换使对比解释变为精确计算：两个实例的差异仅在共享叶不一致的坐标上非零，且每个差异坐标可追溯到具体分裂。基于该表示构建的可逆方法经重复交叉验证，在五个表格数据集上以6.2×10^-15精度还原模型决策；在信用数据集上努力-现实性维度Pareto占优；在考虑实际可执行变更（排除年龄、已结清欠款等）后，有效性保留58%，显著优于最强基线41%。

## 关键点

- 将梯度提升集成的叶值视为坐标，实例在R^M空间形成线性表示，对比解释变得精确。
- 两个实例的差异向量仅在未共享叶子的坐标上非零，每个坐标可追溯到真实树的真实分裂。
- 无需拟合、采样或假设特征可加性，加性在坐标空间中天然存在。
- 可逆方法在五个表格数据集上达到6.2×10^-15的决策重构精度，在可执行变更限制下有效性保留58%。

## 全文

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. We build a recourse method on this representation and evaluate it on five tabular datasets under repeated cross-validation. Its recommendation reconstructs the model's own decision to 6.2 x 10^-15, so an auditor can re-check the arithmetic without the model. On the credit datasets it is Pareto-non-dominated on effort against realism. And when recommendations are restricted to changes the subject could actually make -- not their age, not a settled delinquency -- it retains 58% of its validity where the strongest baseline retains 41%, a distinction the standard evaluation cannot see because it never asks whether a recommendation can be carried out.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_ec322f253f4419444cbd2554