Statistically Significant Concept-based Explanation of Image Classifiers via Model Knockoffs
This addresses the issue of unreliable explanations in image classification for users needing trustworthy AI, though it is an incremental improvement on existing concept-based methods.
The paper tackles the problem of false positives in concept-based explanations for image classifiers by proposing a method that uses model knockoffs to select statistically significant concepts while controlling the False Discovery Rate (FDR). The results show that the method properly controls FDR and selects highly interpretable concepts to improve model trustworthiness.
A concept-based classifier can explain the decision process of a deep learning model by human-understandable concepts in image classification problems. However, sometimes concept-based explanations may cause false positives, which misregards unrelated concepts as important for the prediction task. Our goal is to find the statistically significant concept for classification to prevent misinterpretation. In this study, we propose a method using a deep learning model to learn the image concept and then using the Knockoff samples to select the important concepts for prediction by controlling the False Discovery Rate (FDR) under a certain value. We evaluate the proposed method in our synthetic and real data experiments. Also, it shows that our method can control the FDR properly while selecting highly interpretable concepts to improve the trustworthiness of the model.