On The Reasons Behind Decisions

TL;DR

This paper develops a theoretical framework for understanding the reasons behind decisions made by Boolean classifiers, introducing notions like sufficient and necessary reasons, and providing algorithms for their computation.

Contribution

It introduces a formal theory for explaining Boolean classifier decisions, including new notions and efficient algorithms, with practical case study applications.

Findings

Defined notions of sufficient, necessary, and complete reasons behind decisions

Developed algorithms for computing decision reasons efficiently

Applied the framework to a real-world case study

Abstract

Recent work has shown that some common machine learning classifiers can be compiled into Boolean circuits that have the same input-output behavior. We present a theory for unveiling the reasons behind the decisions made by Boolean classifiers and study some of its theoretical and practical implications. We define notions such as sufficient, necessary and complete reasons behind decisions, in addition to classifier and decision bias. We show how these notions can be used to evaluate counterfactual statements such as "a decision will stick even if ... because ... ." We present efficient algorithms for computing these notions, which are based on new advances on tractable Boolean circuits, and illustrate them using a case study.