Practical Attribution Guidance for Rashomon Sets

TL;DR

This paper introduces a practical sampling method for the Rashomon set in XAI, emphasizing generalizability and sparsity, and demonstrates its effectiveness through mathematical and real-world datasets.

Contribution

It proposes an $psilon$-subgradient sampling approach guided by norms, addressing key axioms often unmet by existing attribution methods.

Findings

The method satisfies generalizability and sparsity axioms.

It outperforms existing sampling methods on practical datasets.

Demonstrates effectiveness on a fundamental mathematical problem.

Abstract

Different prediction models might perform equally well (Rashomon set) in the same task, but offer conflicting interpretations and conclusions about the data. The Rashomon effect in the context of Explainable AI (XAI) has been recognized as a critical factor. Although the Rashomon set has been introduced and studied in various contexts, its practical application is at its infancy stage and lacks adequate guidance and evaluation. We study the problem of the Rashomon set sampling from a practical viewpoint and identify two fundamental axioms - generalizability and implementation sparsity that exploring methods ought to satisfy in practical usage. These two axioms are not satisfied by most known attribution methods, which we consider to be a fundamental weakness. We use the norms to guide the design of an $\epsilon$-subgradient-based sampling method. We apply this method to a fundamental mathematical problem as a proof of concept and to a set of practical datasets to demonstrate its ability compared with existing sampling methods.