Efficient Unsupervised Shortcut Learning Detection and Mitigation in Transformers

TL;DR

This paper introduces an unsupervised framework for detecting and mitigating shortcut learning in transformer models, significantly improving accuracy and interpretability with minimal human annotation and computational resources.

Contribution

It presents a novel unsupervised method for identifying and reducing shortcut reliance in transformers, enhancing robustness and interpretability without extensive labeling.

Findings

Improves worst-group and average accuracy on multiple datasets

Detects meaningful shortcuts that aid human understanding

Operates efficiently on consumer hardware

Abstract

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.