Geometric Remove-and-Retrain (GOAR): Coordinate-Invariant eXplainable AI Assessment
This work addresses a specific challenge in XAI evaluation for researchers and practitioners, offering an incremental improvement over existing methods.
The paper tackles the problem of evaluating feature importance in explainable AI by identifying limitations in pixel-perturbation methods like ROAR, which fail to discriminate between attribution methods, and introduces GOAR as an alternative approach that overcomes these issues, as validated through experiments on synthetic and real datasets.
Identifying the relevant input features that have a critical influence on the output results is indispensable for the development of explainable artificial intelligence (XAI). Remove-and-Retrain (ROAR) is a widely accepted approach for assessing the importance of individual pixels by measuring changes in accuracy following their removal and subsequent retraining of the modified dataset. However, we uncover notable limitations in pixel-perturbation strategies. When viewed from a geometric perspective, we discover that these metrics fail to discriminate between differences among feature attribution methods, thereby compromising the reliability of the evaluation. To address this challenge, we introduce an alternative feature-perturbation approach named Geometric Remove-and-Retrain (GOAR). Through a series of experiments with both synthetic and real datasets, we substantiate that GOAR transcends the limitations of pixel-centric metrics.