跳转至

文章背景与核心概要

可解释人工智能（XAI）日益强调可操作的反事实追索（actionable counterfactual recourse）的重要性，然而当前的方法经常受到因果无效性、高认知负担以及预测失效等问题的困扰。为了解决这些局限性，本文引入了可操作的基于案例的特征重要性（A-CBFI），这是一个专为表格机器学习定制的诊断与处方一体化框架。

基于结构因果模型（SCMs），A-CBFI 能够隔离协同交互瓶颈并释放抑制性结构锁，将其转化为高度针对性的干预措施。该框架的核心亮点包括：通过将主动用户干预空间（\(L_{\mathrm{active}}\)）与下游效应在数学上进行分离，将超过 98.3% 的干预工作直接集中在诊断出的根本原因上；在金融和医疗领域的实证评估中，将主动人工干预负担减少了 76.9%，同时匹配了详尽因果基线的全局追索成本；在所有因果可行实例上实现了完全的相对收敛，同时保持了因果有效性。

Actionable CBFI: Integrating Structural Decomposition and Causal Counterfactual Recourse for Tabular Machine Learning

可操作的CBFI：将结构分解与因果反事实追索集成用于表格机器学习

Summary

摘要

Explainable artificial intelligence (XAI) increasingly emphasizes the need for actionable counterfactual recourse, yet current methodologies frequently suffer from causal invalidity, high cognitive burdens, and predictive failures. To resolve these limitations, this paper introduces Actionable Case-Based Feature Importance (A-CBFI), a novel diagnosis-prescription integrated framework tailored for tabular machine learning.

Explainable artificial intelligence (XAI) increasingly emphasizes the need for actionable counterfactual recourse, yet current methodologies frequently suffer from causal invalidity, high cognitive burdens, and predictive failures. To resolve these limitations, this paper introduces Actionable Case-Based Feature Importance (A-CBFI), a novel diagnosis-prescription integrated framework tailored for tabular machine learning.

Grounded in Structural Causal Models (SCMs), A-CBFI isolates synergistic interaction bottlenecks and releases suppressive structural locks, translating them into highly targeted interventions. Key highlights of the framework include: * Concentrating over 98.3% of intervention efforts directly on diagnosed root causes by mathematically separating the active user intervention space (\(L_{\mathrm{active}}\)) from downstream effects. * Reducing the active human intervention burden by 76.9% in empirical evaluations across the financial and healthcare domains, while matching the global recourse cost of exhaustive causal baselines. * Achieving full relative convergence across all causally feasible instances while preserving causal validity.

Grounded in Structural Causal Models (SCMs), A-CBFI isolates synergistic interaction bottlenecks and releases suppressive structural locks, translating them into highly targeted interventions. Key highlights of the framework include: * Concentrating over 98.3% of intervention efforts directly on diagnosed root causes by mathematically separating the active user intervention space (\(L_{\mathrm{active}}\)) from downstream effects. * Reducing the active human intervention burden by 76.9% in empirical evaluations across the financial and healthcare domains, while matching the global recourse cost of exhaustive causal baselines. * Achieving full relative convergence across all causally feasible instances while preserving causal validity.

Document Metadata

文档元数据

Metadata Field Details
arXiv ID arXiv:2608.27821 [cs.LG]
Subjects Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Author Sejong Oh
Submitted On August 28, 2026
MSC Classes 62R07
ACM Classes I.2.6
DOI 10.48550/arXiv.2608.27821
Metadata Field Details
arXiv ID arXiv:2608.27821 [cs.LG]
Subjects Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Author Sejong Oh
Submitted On August 28, 2026
MSC Classes 62R07
ACM Classes I.2.6
DOI 10.48550/arXiv.2608.27821

Abstract

摘要正文

Explainable artificial intelligence (XAI) increasingly calls for actionable counterfactual recourse, yet current methodologies face challenges related to causal invalidity, excessive cognitive burden, and predictive failure. Exhaustive causal search algorithms often require modifications to multiple attributes, whereas additive attribution-guided methods, such as SHAP, ignore higher-order feature synergies, leading to suboptimal predictive momentum and diffuse intervention effort in complex nonlinear models, such as XGBoost.

Explainable artificial intelligence (XAI) increasingly calls for actionable counterfactual recourse, yet current methodologies face challenges related to causal invalidity, excessive cognitive burden, and predictive failure. Exhaustive causal search algorithms often require modifications to multiple attributes, whereas additive attribution-guided methods, such as SHAP, ignore higher-order feature synergies, leading to suboptimal predictive momentum and diffuse intervention effort in complex nonlinear models, such as XGBoost.

To bridge this gap, we introduce actionable case-based feature importance (A-CBFI), a diagnosis-prescription integrated framework for tabular machine learning. Grounded in structural causal models (SCMs), A-CBFI isolates synergistic interaction bottlenecks and releases suppressive structural locks, translating them into targeted interventions. By mathematically separating the active user intervention space (\(L_{\mathrm{active}}\)) from downstream effects and concentrating over 98.3% of the intervention effort on diagnosed root causes, A-CBFI enables highly targeted interventions.

To bridge this gap, we introduce actionable case-based feature importance (A-CBFI), a diagnosis-prescription integrated framework for tabular machine learning. Grounded in structural causal models (SCMs), A-CBFI isolates synergistic interaction bottlenecks and releases suppressive structural locks, translating them into targeted interventions. By mathematically separating the active user intervention space (\(L_{\mathrm{active}}\)) from downstream effects and concentrating over 98.3% of the intervention effort on diagnosed root causes, A-CBFI enables highly targeted interventions.

Empirical evaluations across the financial and healthcare domains demonstrate that A-CBFI reduces the active human intervention burden by 76.9% while maintaining comparable global recourse cost to exhaustive causal baselines. By prioritizing the diagnosed causal bottlenecks, A-CBFI provides targeted and actionable recourse while maintaining causal validity and achieving full relative convergence across all causally feasible instances.

Empirical evaluations across the financial and healthcare domains demonstrate that A-CBFI reduces the active human intervention burden by 76.9% while maintaining comparable global recourse cost to exhaustive causal baselines. By prioritizing the diagnosed causal bottlenecks, A-CBFI provides targeted and actionable recourse while maintaining causal validity and achieving full relative convergence across all causally feasible instances.

附加资源与链接