학술논문
Explainability Is Not a Game.
Document Type
Article
Author
Source
Subject
*Artificial intelligence
*Decision trees
Human-artificial intelligence interaction
Machine learning
Parameterization
Cooperative game theory
*
Language
ISSN
0001-0782
Abstract
This article presents the concept of explainability in machine learning (ML) decisions, automated approaches that explain the predictions made especially when these decisions impact humans. The article explores both informal and formal approaches to eXplainable artificial intelligence (XIA) with an in depth look at utilizing Shapley values in informal approaches to highlight the limitations to this method.