Sensitive Region-based Metamorphic Testing Framework using Explainable AI
This work addresses the challenge of improving testing efficiency for image recognition systems by focusing on sensitive regions, though it is incremental as it builds on existing metamorphic testing methods.
The paper tackled the problem of identifying which image regions are most sensitive to transformations that cause misclassifications in deep learning systems, and proposed a metamorphic testing framework that uses explainable AI to specify these regions, resulting in effective fault detection.
Deep Learning (DL) is one of the most popular research topics in machine learning and DL-driven image recognition systems have developed rapidly. Recent research has employed metamorphic testing (MT) to detect misclassified images. Most of them discuss metamorphic relations (MR), with limited attention given to which regions should be transformed. We focus on the fact that there are sensitive regions where even small transformations can easily change the prediction results and propose an MT framework that efficiently tests for regions prone to misclassification by transforming these sensitive regions. Our evaluation demonstrated that the sensitive regions can be specified by Explainable AI (XAI) and our framework effectively detects faults.