Can I read Practical Benefits of Feature Feedback Under Distribution Shift on EtoBox?

Practical Benefits of Feature Feedback Under Distribution Shift by Katakkar, Anurag; Yoo, Clay H.; Wang, Weiqin; Lipton, Zachary C.; Kaushik, Divyansh is a scholarly article available to read on EtoBox.

What is Practical Benefits of Feature Feedback Under Distribution Shift about?

In attempts to develop sample-efficient and interpretable algorithms, researcher have explored myriad mechanisms for collecting and exploiting feature feedback (or rationales) auxiliary annotations provided for training (but not test) instances that highlight salient evidence. Examples include bounding boxes around objects and salient spans in text. Despite its intuitive appeal, feature feedback has not delivered significant gains in practical problems as assessed on iid holdout sets. However, recent works on counterfactually augmented data suggest an alternative benefit of supplemental annotations, beyond interpretability: lessening sensitivity to spurious patterns and consequently delivering gains in out-of-domain evaluations. We speculate that while existing methods for incorporating feature feedback have delivered negligible in-sample performance gains, they may nevertheless provide out-of-domain benefits. Our experiments addressing sentiment analysis, show that feature feedback methods perform significantly better on various natural out-of-domain datasets despite comparable in-domain evaluations. By contrast, performance on natural language inference remains comparable. Finall

Author

Katakkar, Anurag; Yoo, Clay H.; Wang, Weiqin; Lipton, Zachary C.; Kaushik, Divyansh

Published

2021

Language

EN