An Investigation of the (In)effectiveness of Counterfactually Augmented Data

• URL: http://arxiv.org/abs/2107.00753v1
• Date: Thu, 1 Jul 2021 21:46:43 GMT
• Title: An Investigation of the (In)effectiveness of Counterfactually Augmented Data
• Authors: Nitish Joshi, He He
• Abstract summary: We show that while counterfactually-augmented data (CAD) is effective at identifying robust features, it may prevent the model from learning unperturbed robust features. Our results show that the lack of perturbation diversity in current CAD datasets limits its effectiveness on OOD generalization.
• Score: 10.316235366821111
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: While pretrained language models achieve excellent performance on natural language understanding benchmarks, they tend to rely on spurious correlations and generalize poorly to out-of-distribution (OOD) data. Recent work has explored using counterfactually-augmented data (CAD) -- data generated by minimally perturbing examples to flip the ground-truth label -- to identify robust features that are invariant under distribution shift. However, empirical results using CAD for OOD generalization have been mixed. To explain this discrepancy, we draw insights from a linear Gaussian model and demonstrate the pitfalls of CAD. Specifically, we show that (a) while CAD is effective at identifying robust features, it may prevent the model from learning unperturbed robust features, and (b) CAD may exacerbate existing spurious correlations in the data. Our results show that the lack of perturbation diversity in current CAD datasets limits its effectiveness on OOD generalization, calling for innovative crowdsourcing procedures to elicit diverse perturbation of examples.

Related papers

• HGOE: Hybrid External and Internal Graph Outlier Exposure for Graph Out-of-Distribution Detection [78.47008997035158]
  Graph data exhibits greater diversity but lower robustness to perturbations, complicating the integration of outliers. We propose the introduction of textbfHybrid External and Internal textbfGraph textbfOutlier textbfExposure (HGOE) to improve graph OOD detection performance.
  arXiv  Detail & Related papers  (2024-07-31T16:55:18Z)
• Accuracy on the wrong line: On the pitfalls of noisy data for out-of-distribution generalisation [70.36344590967519]
  We show that noisy data and nuisance features can be sufficient to shatter the Accuracy-on-the-line phenomenon. We demonstrate this phenomenon across both synthetic and real datasets with noisy data and nuisance features.
  arXiv  Detail & Related papers  (2024-06-27T09:57:31Z)
• PairCFR: Enhancing Model Training on Paired Counterfactually Augmented Data through Contrastive Learning [49.60634126342945]
  Counterfactually Augmented Data (CAD) involves creating new data samples by applying minimal yet sufficient modifications to flip the label of existing data samples to other classes. Recent research reveals that training with CAD may lead models to overly focus on modified features while ignoring other important contextual information. We employ contrastive learning to promote global feature alignment in addition to learning counterfactual clues.
  arXiv  Detail & Related papers  (2024-06-09T07:29:55Z)
• Unlock the Potential of Counterfactually-Augmented Data in Out-Of-Distribution Generalization [25.36416774024584]
  Counterfactually-Augmented Data (CAD) has the potential to improve the Out-Of-Distribution (OOD) generalization capability of language models. In this study, we attribute the inefficiency to the myopia phenomenon caused by CAD. We introduce two additional constraints based on CAD's structural properties to help language models extract more complete causal features in CAD.
  arXiv  Detail & Related papers  (2023-10-10T14:41:38Z)
• Graph Structure and Feature Extrapolation for Out-of-Distribution Generalization [54.64375566326931]
  Out-of-distribution (OOD) generalization deals with the prevalent learning scenario where test distribution shifts from training distribution. We propose to achieve graph OOD generalization with the novel design of non-Euclidean-space linear extrapolation. Our design tailors OOD samples for specific shifts without corrupting underlying causal mechanisms.
  arXiv  Detail & Related papers  (2023-06-13T18:46:28Z)
• Improving the Out-Of-Distribution Generalization Capability of Language Models: Counterfactually-Augmented Data is not Enough [19.38778317110205]
  Counterfactually-Augmented Data (CAD) has the potential to improve language models' Out-Of-Distribution (OOD) generalization capability. In this paper, we attribute the inefficiency to Myopia Phenomenon caused by CAD. We design two additional constraints to help language models extract more complete causal features contained in CAD.
  arXiv  Detail & Related papers  (2023-02-18T14:39:03Z)
• How robust are pre-trained models to distribution shift? [82.08946007821184]
  We show how spurious correlations affect the performance of popular self-supervised learning (SSL) and auto-encoder based models (AE) We develop a novel evaluation scheme with the linear head trained on out-of-distribution (OOD) data, to isolate the performance of the pre-trained models from a potential bias of the linear head used for evaluation.
  arXiv  Detail & Related papers  (2022-06-17T16:18:28Z)
• Counterfactually Augmented Data and Unintended Bias: The Case of Sexism and Hate Speech Detection [35.29235215101502]
  Over-relying on core features may lead to unintended model bias. We test models for sexism and hate speech detection on challenging data. Using a diverse set of CAD -- construct-driven and construct-agnostic -- reduces such unintended bias.
  arXiv  Detail & Related papers  (2022-05-09T12:39:26Z)
• General Greedy De-bias Learning [163.65789778416172]
  We propose a General Greedy De-bias learning framework (GGD), which greedily trains the biased models and the base model like gradient descent in functional space. GGD can learn a more robust base model under the settings of both task-specific biased models with prior knowledge and self-ensemble biased model without prior knowledge.
  arXiv  Detail & Related papers  (2021-12-20T14:47:32Z)
• How Does Counterfactually Augmented Data Impact Models for Social Computing Constructs? [35.29235215101502]
  We investigate the benefits of counterfactually augmented data (CAD) for social NLP models by focusing on three social computing constructs -- sentiment, sexism, and hate speech. We find that while models trained on CAD show lower in-domain performance, they generalize better out-of-domain.
  arXiv  Detail & Related papers  (2021-09-14T23:46:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.