# Practical Benefits of Feature Feedback Under Distribution Shift

arXiv: https://arxiv.org/abs/2110.07566 · HTML: https://mlsanity.com/paper/2110.07566/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 31 references that are in the map.
- First posted on arXiv 14 Oct 2021.

## Builds on (most cited first, 19 of 19)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference](https://mlsanity.com/paper/1704.05426/) (2017)
- [ERASER: A Benchmark to Evaluate Rationalized NLP Models](https://mlsanity.com/paper/1911.03429/) (2019)
- [Learning the Difference that Makes a Difference with Counterfactually-Augmented Data](https://mlsanity.com/paper/1909.12434/) (2019)
- [Right for the Right Reasons: Training Differentiable Models by Constraining their Explanations](https://mlsanity.com/paper/1703.03717/) (2017)
- [Rationalizing Neural Predictions](https://mlsanity.com/paper/1606.04155/) (2016)
- [Improving the Adversarial Robustness and Interpretability of Deep Neural Networks by Regularizing their Input Gradients](https://mlsanity.com/paper/1711.09404/) (2017)
- [SemEval-2017 Task 4: Sentiment Analysis in Twitter](https://mlsanity.com/paper/1912.00741/) (2019)
- [Learning to Faithfully Rationalize by Construction](https://mlsanity.com/paper/2005.00115/) (2020)
- [Measuring the Reliability of Hate Speech Annotations: The Case of the European Refugee Crisis](https://mlsanity.com/paper/1701.08118/) (2017)
- [Seeing Things from a Different Angle: Discovering Diverse Perspectives about Claims](https://mlsanity.com/paper/1906.03538/) (2019)
- [Explaining The Efficacy of Counterfactually Augmented Data](https://mlsanity.com/paper/2010.02114/) (2020)
- [Inferring Which Medical Treatments Work from Reports of Clinical Trials](https://mlsanity.com/paper/1904.01606/) (2019)
- [QED: A Framework and Dataset for Explanations in Question Answering](https://mlsanity.com/paper/2009.06354/) (2020)
- [Rationale-Augmented Convolutional Neural Networks for Text Classification](https://mlsanity.com/paper/1605.04469/) (2016)
- [Counterfactually-Augmented SNLI Training Data Does Not Yield Better Generalization Than Unaugmented Data](https://mlsanity.com/paper/2010.04762/) (2020)
- [Weakly- and Semi-supervised Evidence Extraction](https://mlsanity.com/paper/2011.01459/) (2020)
- [Robust Learning from Discriminative Feature Feedback](https://mlsanity.com/paper/2003.03946/) (2020)
- [Model Agnostic Answer Reranking System for Adversarial Question Answering](https://mlsanity.com/paper/2102.03016/) (2021)

## Built on by (most cited first, 5 of 5)

- [Closed-Loop Unsupervised Representation Disentanglement with β-VAE Distillation and Diffusion Probabilistic Feedback](https://mlsanity.com/paper/2402.02346/) (2024)
- [Unsupervised domain adaptation by learning using privileged information](https://mlsanity.com/paper/2303.09350/) (2023)
- [Learning to Ignore Adversarial Attacks](https://mlsanity.com/paper/2205.11551/) (2022)
- [Zero-Shot Cross-Lingual Sentiment Classification under Distribution Shift: an Exploratory Study](https://mlsanity.com/paper/2311.06549/) (2023)
- [Robustifying Sentiment Classification by Maximally Exploiting Few Counterfactuals](https://mlsanity.com/paper/2210.11805/) (2022)

## Abstract (arXiv)

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. Finally, we compare those tasks where feature feedback does (and does not) help.