Reliable Evaluation of Attribution Maps in CNNs: A Perturbation-Based Approach
This provides a more reliable evaluation framework for interpreting CNN predictions, addressing a key bottleneck in explainable AI, though it is incremental as it builds on existing attribution map methods.
The paper tackles the problem of unreliable evaluation of attribution maps in CNNs by showing that existing metrics are susceptible to distribution shifts, and proposes a perturbation-based method that corrects these shifts, demonstrating through Kendall's τ rank correlation that it provides increased consistency across 15 dataset-architecture combinations and identifies SmoothGrad as the best map.
In this paper, we present an approach for evaluating attribution maps, which play a central role in interpreting the predictions of convolutional neural networks (CNNs). We show that the widely used insertion/deletion metrics are susceptible to distribution shifts that affect the reliability of the ranking. Our method proposes to replace pixel modifications with adversarial perturbations, which provides a more robust evaluation framework. By using smoothness and monotonicity measures, we illustrate the effectiveness of our approach in correcting distribution shifts. In addition, we conduct the most comprehensive quantitative and qualitative assessment of attribution maps to date. Introducing baseline attribution maps as sanity checks, we find that our metric is the only contender to pass all checks. Using Kendall's $τ$ rank correlation coefficient, we show the increased consistency of our metric across 15 dataset-architecture combinations. Of the 16 attribution maps tested, our results clearly show SmoothGrad to be the best map currently available. This research makes an important contribution to the development of attribution maps by providing a reliable and consistent evaluation framework. To ensure reproducibility, we will provide the code along with our results.