Contextual Importance and Utility: aTheoretical Foundation

TL;DR

This paper establishes a theoretical foundation for the XAI method CIU, based on Multi-Attribute Utility Theory, and demonstrates its advantages over influence-based methods in providing faithful model explanations.

Contribution

It introduces the concept of contextual influence, enabling direct comparison of CIU with AFA methods, and proves CIU's explanation faithfulness through experiments.

Findings

CIU explanations outperform influence-based methods in simple models.

CIU guarantees explanation faithfulness.

Theoretical foundation links CIU to Multi-Attribute Utility Theory.

Abstract

This paper provides new theory to support to the eXplainable AI (XAI) method Contextual Importance and Utility (CIU). CIU arithmetic is based on the concepts of Multi-Attribute Utility Theory, which gives CIU a solid theoretical foundation. The novel concept of contextual influence is also defined, which makes it possible to compare CIU directly with so-called additive feature attribution (AFA) methods for model-agnostic outcome explanation. One key takeaway is that the "influence" concept used by AFA methods is inadequate for outcome explanation purposes even for simple models to explain. Experiments with simple models show that explanations using contextual importance (CI) and contextual utility (CU) produce explanations where influence-based methods fail. It is also shown that CI and CU guarantees explanation faithfulness towards the explained model.