Efficient Explanations from Empirical Explainers

TL;DR

This paper introduces Empirical Explainers that efficiently approximate costly neural explanation methods by learning from data, significantly reducing computational costs while maintaining accuracy in language applications.

Contribution

It proposes a novel feature attribution modeling approach that learns to predict explanations, offering a practical solution to reduce the computational burden of neural explainability.

Findings

Empirical Explainers model expensive explainers well in language tasks.

They achieve similar attribution accuracy at a fraction of the computational cost.

The approach is effective in applications tolerant to approximation errors.

Abstract

Amid a discussion about Green AI in which we see explainability neglected, we explore the possibility to efficiently approximate computationally expensive explainers. To this end, we propose feature attribution modelling with Empirical Explainers. Empirical Explainers learn from data to predict the attribution maps of expensive explainers. We train and test Empirical Explainers in the language domain and find that they model their expensive counterparts surprisingly well, at a fraction of the cost. They could thus mitigate the computational burden of neural explanations significantly, in applications that tolerate an approximation error.