Debiasing Masks: A New Framework for Shortcut Mitigation in NLU

Johannes Mario Meissner, Saku Sugawara, Akiko Aizawa


Abstract
Debiasing language models from unwanted behaviors in Natural Language Understanding (NLU) tasks is a topic with rapidly increasing interest in the NLP community. Spurious statistical correlations in the data allow models to perform shortcuts and avoid uncovering more advanced and desirable linguistic features.A multitude of effective debiasing approaches has been proposed, but flexibility remains a major issue. For the most part, models must be retrained to find a new set of weights with debiased behavior. We propose a new debiasing method in which we identify debiased pruning masks that can be applied to a finetuned model. This enables the selective and conditional application of debiasing behaviors. We assume that bias is caused by a certain subset of weights in the network; our method is, in essence, a mask search to identify and remove biased weights. Our masks show equivalent or superior performance to the standard counterparts, while offering important benefits. Pruning masks can be stored with high efficiency in memory, and it becomes possible to switch among several debiasing behaviors (or revert back to the original biased model) at inference time. Finally, it opens the doors to further research on how biases are acquired by studying the generated masks. For example, we observed that the early layers and attention heads were pruned more aggressively, possibly hinting towards the location in which biases may be encoded.
Anthology ID:
2022.emnlp-main.517
Volume:
Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing
Month:
December
Year:
2022
Address:
Abu Dhabi, United Arab Emirates
Editors:
Yoav Goldberg, Zornitsa Kozareva, Yue Zhang
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
7607–7613
Language:
URL:
https://aclanthology.org/2022.emnlp-main.517
DOI:
10.18653/v1/2022.emnlp-main.517
Bibkey:
Cite (ACL):
Johannes Mario Meissner, Saku Sugawara, and Akiko Aizawa. 2022. Debiasing Masks: A New Framework for Shortcut Mitigation in NLU. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 7607–7613, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
Cite (Informal):
Debiasing Masks: A New Framework for Shortcut Mitigation in NLU (Meissner et al., EMNLP 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.emnlp-main.517.pdf