The effect of whitening on explanation performance

TL;DR

This paper investigates how whitening preprocessing affects the accuracy of feature attribution methods in explainable AI, revealing that its impact varies depending on the method and model, and emphasizing the importance of data preprocessing for interpretability.

Contribution

The study empirically evaluates the effect of whitening on explanation quality across multiple attribution methods and models, and provides theoretical insights into its potential to mitigate suppressor variable issues.

Findings

Whitening can improve explanation accuracy for some attribution methods.

The effectiveness of whitening varies significantly across methods and models.

Preprocessing quality plays a crucial role in explanation fidelity.

Abstract

Explainable Artificial Intelligence (XAI) aims to provide transparent insights into machine learning models, yet the reliability of many feature attribution methods remains a critical challenge. Prior research (Haufe et al., 2014; Wilming et al., 2022, 2023) has demonstrated that these methods often erroneously assign significant importance to non-informative variables, such as suppressor variables, leading to fundamental misinterpretations. Since statistical suppression is induced by feature dependencies, this study investigates whether data whitening, a common preprocessing technique for decorrelation, can mitigate such errors. Using the established XAI-TRIS benchmark (Clark et al., 2024b), which offers synthetic ground-truth data and quantitative measures of explanation correctness, we empirically evaluate 16 popular feature attribution methods applied in combination with 5 distinct whitening transforms. Additionally, we analyze a minimal linear two-dimensional classification problem (Wilming et al., 2023) to theoretically assess whether whitening can remove the impact of suppressor features from Bayes-optimal models. Our results indicate that, while specific whitening techniques can improve explanation performance, the degree of improvement varies substantially across XAI methods and model architectures. These findings highlight the complex relationship between data non-linearities, preprocessing quality, and attribution fidelity, underscoring the vital role of pre-processing techniques in enhancing model interpretability.