Efficient Unsupervised Shortcut Learning Detection and Mitigation in Transformers
This addresses a critical problem for deploying machine learning in sensitive applications like medical diagnostics, though it appears incremental as it builds on existing advancements.
The paper tackles shortcut learning in transformers by developing an unsupervised framework for detection and mitigation, which significantly improves worst-group accuracy and average accuracy on multiple datasets while requiring minimal human annotation.
Shortcut learning, i.e., a model's reliance on undesired features not directly relevant to the task, is a major challenge that severely limits the applications of machine learning algorithms, particularly when deploying them to assist in making sensitive decisions, such as in medical diagnostics. In this work, we leverage recent advancements in machine learning to create an unsupervised framework that is capable of both detecting and mitigating shortcut learning in transformers. We validate our method on multiple datasets. Results demonstrate that our framework significantly improves both worst-group accuracy (samples misclassified due to shortcuts) and average accuracy, while minimizing human annotation effort. Moreover, we demonstrate that the detected shortcuts are meaningful and informative to human experts, and that our framework is computationally efficient, allowing it to be run on consumer hardware.