XAI-TRIS: Non-linear image benchmarks to quantify false positive post-hoc attribution of feature importance

TL;DR

This paper introduces benchmark datasets and metrics for evaluating the accuracy of post-hoc feature importance explanations in non-linear deep learning models, revealing limitations of current XAI methods.

Contribution

It provides formal ground truth datasets and novel metrics for benchmarking XAI explanations in complex non-linear scenarios, addressing a key gap in the field.

Findings

Popular XAI methods often do not outperform random baselines.

Explanation consistency varies significantly across model architectures.

Many explanations are prone to misinterpretation even under controlled conditions.

Abstract

The field of 'explainable' artificial intelligence (XAI) has produced highly cited methods that seek to make the decisions of complex machine learning (ML) methods 'understandable' to humans, for example by attributing 'importance' scores to input features. Yet, a lack of formal underpinning leaves it unclear as to what conclusions can safely be drawn from the results of a given XAI method and has also so far hindered the theoretical verification and empirical validation of XAI methods. This means that challenging non-linear problems, typically solved by deep neural networks, presently lack appropriate remedies. Here, we craft benchmark datasets for three different non-linear classification scenarios, in which the important class-conditional features are known by design, serving as ground truth explanations. Using novel quantitative metrics, we benchmark the explanation performance of a wide set of XAI methods across three deep learning model architectures. We show that popular XAI methods are often unable to significantly outperform random performance baselines and edge detection methods. Moreover, we demonstrate that explanations derived from different model architectures can be vastly different; thus, prone to misinterpretation even under controlled conditions.