可解释机器学习的因果依赖图

Mar, 2023

Causal Dependence Plots for Interpretable Machine Learning

Joshua R. Loftus, Lucius E. J. Bynum, Sakina Hansen

TL;DR通过通过使用已知或假设的输入变量的因果结构，以产生监督学习模型的简单和实用解释，并将这些解释可视化，提出了因果依赖图，即CDP。CDP可以成为可解释的AI或可解释ML工具箱中的有力工具，并有助于科学机器学习和算法公平性的应用，也可以用于模型无关说明。

Abstract

Explaining artificial intelligence or machine learning models is an increasingly important problem. For humans to stay in the loop and control such systems, we must be able to understand how they interact with th