David Baehrens, Timon Schroeter, Stefan Harmeling, Motoaki Kawanabe, Katja Hansen...
TL;DR本文提出一种基于一组假设的程序，能够解释任何分类方法所做出的决策。
Abstract
After building a classifier with modern tools of machine learning we typically have a black box at hand that is able to predict well for unseen data. Thus, we get an answer to the question what is the most likely