[论文] Leaf Values as Coordinates: Exact Contrastive Explanation for Gradient...

论文概要

研究领域: ML 作者: Emanuele Luzio 发布时间: 2026-08-19 arXiv: 2608.19127

中文摘要

梯度提升集成通过每棵树的一个叶值求和来预测。将这些值视为坐标而非中间结果，每个实例就成为R^M中的一个点，模型在其上线性作用：分数是坐标的和。这一小视角变化使对比解释精确。两个实例之间的差异是一个向量，在它们共享叶子的地方恒为零，因此被拒绝申请者和被接受申请者之间的差距由少数坐标携带，每个都可追溯到真实树中的真实分割。无需拟合、采样或假设特征可加性——可加性已经在正确的空间中存在。我们在此表示上构建一个补救方法，并在重复交叉验证下的五个表格数据集上评估它。其建议重建模型自己的决策到6.2 x 10^-15，因此审计员可以在没有模型的情况下重新检查算术。在信用数据集上，它在努力与真实性之间是Pareto非支配的。当建议被限制为受试者实际可以做出的改变时——不是他们的年龄，不是已解决的拖欠——它保留58%的有效性，而最强基线保留41%，标准评估看不到这一区别，因为它从不问建议是否可以执行。

原文摘要

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-...

--- *自动采集于 2026-08-21*

#论文 #arXiv #ML #小凯