Evaluating Explanations: How Much Do Explanations from the Teacher Aid Students?
Citations Over TimeTop 10% of 2022 papers
Abstract
Abstract While many methods purport to explain predictions by highlighting salient features, what aims these explanations serve and how they ought to be evaluated often go unstated. In this work, we introduce a framework to quantify the value of explanations via the accuracy gains that they confer on a student model trained to simulate a teacher model. Crucially, the explanations are available to the student during training, but are not available at test time. Compared with prior proposals, our approach is less easily gamed, enabling principled, automatic, model-agnostic evaluation of attributions. Using our framework, we compare numerous attribution methods for text classification and question answering, and observe quantitative differences that are consistent (to a moderate to high degree) across different student model architectures and learning strategies.1
Related Papers
- Overview of Question-Answering(2002)
- → NOT ALL FAILURES ARE ALIKE: SELF ATTRIBUTION AND PERCEPTION OF TEACHERS‘ ATTRIBUTIONS FOR FAILING TESTS IN LIKED VERSUS DISLIKED SUBJECTS(1987)3 cited
- → Nurses' Evaluations of Patient Attributions for the Cause and Future of Their Illness1(1990)2 cited
- → The effects of observers' brief task performance and their point of view on their causal attribution for actor's performance.(1984)1 cited
- A Study on Attribution to Learning Success or Failure of English Majors(2013)