Elamur Research⌕
☰
home / papers / 2021 / Sustainable Modular Debiasing of Language Models

Sustainable Modular Debiasing of Language Models

Summary

Pre-trained language models absorb stereotypical associations from their training text, such as linking male terms with certain professions. Earlier debiasing methods retrain every model weight. This is costly, especially for very large models, and can erase useful knowledge gained in pretraining. The authors propose ADELE (Adapter-based DEbiasing of LanguagE Models). It inserts small bottleneck adapter layers into a frozen BERT and trains only those layers. Training uses masked language modeling on a corpus extended with two-sided Counterfactual Data Augmentation (CDA), in which gendered terms are swapped.

Tested on gender bias across three intrinsic and two downstream benchmarks, ADELE reduces bias about as well as full fine-tuning approaches while training far fewer parameters. The paper also identifies "fairness forgetting": debiasing effects fade during large-scale task fine-tuning. Stacking a task adapter on a frozen debiasing adapter largely prevents this. Finally, a debiasing adapter trained on English data on top of multilingual BERT transfers zero-shot to six other languages. Together, these results point toward cheaper, reusable, and composable fairness components.

Key points

  • Architecture: ADELE uses the "Pfeiffer" adapter design, with one adapter per Transformer layer placed after the feed-forward sub-layer. The original BERT weights stay frozen, so the debiasing functionality is isolated and can be switched on when needed.
  • Training data: Debiasing training applies two-sided CDA, keeping both the original and the term-swapped sentences. The adapter sees the original corpus first and the counterfactual portion second.
  • Evaluation: Intrinsic measures are BEC-Pro (2,700 gender–profession sentence pairs), DisCo, and WEAT. Extrinsic measures are Bias-STS-B and Bias-NLI.
  • Fairness forgetting: The variant with an added task adapter (ADELE-TA) stays much fairer after NLI training, with an FN score of 0.557 versus 0.127 for fully fine-tuned ADELE. The cost is about 3 points of NLI accuracy.
  • Multilingual transfer and scope: With mBERT, the English-trained debiasing adapter carries over to six target languages without any target-language debiasing data. The authors acknowledge that the work relies on a binary notion of gender.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Natural language processing

Cited by

85 papers of our library cite it (of the 140884 whose bibliographies we read); 3 of them have a page here, 82 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2021.findings-emnlp.411

@inproceedings{lauscher2021sustainable,
  title = {{Sustainable Modular Debiasing of Language Models}},
  author = {Lauscher, Anne and Lueken, Tobias and Glavaš, Goran},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2021},
  year = {2021},
  doi = {10.18653/v1/2021.findings-emnlp.411},
  url = {https://doi.org/10.18653/v1/2021.findings-emnlp.411}
}

References

56 references, 18 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Soumya Barikeri, Anne Lauscher, Ivan Vulić, and Goran Glavaš. 2021. RedditBias: A real-world resource for bias evaluation and debiasing of conversational language models. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conf… find
  2. Marion Bartl, Malvina Nissim, and Albert Gatt. 2020. Unmasking contextual stereotypes: Measuring and mitigating BERT’s gender bias. In Proceedings of the Second Workshop on Gender Bias in Natural Language Processing, pages 1–16, Barcelona, Spain (Online). Association for Computational Linguistics. find
  3. Su Lin Blodgett, Solon Barocas, Hal Daumé III, and Hanna Wallach. 2020. Language (technology) is power: A critical survey of “bias” in NLP. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5454– 5476, Online. Association for Computational Linguistics. find our card
  4. Tolga Bolukbasi, Kai-Wei Chang, James Zou, Venkatesh Saligrama, and Adam Kalai. 2016. Man is to computer programmer as woman is to homemaker? debiasing word embeddings. In Proceedings of the 30th International Conference on Neural Information Processing Systems, NIPS’16, page 4356–4364, Red Hook, NY… find our card
  5. Kimberlé Crenshaw. 1989. Demarginalizing the intersection of race and sex: A black feminist critique of antidiscrimination doctrine, feminist theory and antiracist politics. u. Chi. Legal f., 1989:139. find
  6. Sunipa Dev, Tao Li, Jeff M. Phillips, and Vivek Srikumar. 2020. On measuring and mitigating biased inferences of word embeddings. In The Thirty-Fourth AAAI Conference on Artificial Intelligence, AAAI 2020, The Thirty-Second Innovative Applications of Artificial Intelligence Conference, IAAI 2020, Th… find
  7. Sunipa Dev and Jeff Phillips. 2019. Attenuating bias in word vectors. In The 22nd International Conference on Artificial Intelligence and Statistics, pages 879– 887. PMLR. find
  8. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,… find
  9. Nikhil Garg, Londa Schiebinger, Dan Jurafsky, and James Zou. 2018. Word embeddings quantify 100 years of gender and ethnic stereotypes. Proceedings of the National Academy of Sciences, 115(16):E3635–E3644. find
  10. Shikha Bordia and Samuel R. Bowman. 2019. Identifying and reducing gender bias in word-level language models. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop, pages 7–15, Minneapolis, Minnesota. Associati… find our card
  11. Goran Glavaš, Ananya Ganesh, and Swapna Somasundaran. 2021. Training and domain adaptation for supervised text segmentation. In Proceedings of the 16th Workshop on Innovative Use of NLP for Building Educational Applications, pages 110–116, Online. Association for Computational Linguistics. find
  12. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, et al. 2020. Language models are few-shot learners. In Advances in Neural Information Processing Systems, volume 33, pages 1877–… find
  13. Rowan Hall Maudslay, Hila Gonen, Ryan Cotterell, and Simone Teufel. 2019. It’s all in the name: Mitigating gender bias with name-based counterfactual data substitution. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conferenc… find
  14. Aylin Caliskan, Joanna J. Bryson, and Arvind Narayanan. 2017. Semantics derived automatically from language corpora contain human-like biases. Science, 356(6334):183–186. find
  15. Daniel Cer, Mona Diab, Eneko Agirre, Iñigo Lopez-Gazpio, and Lucia Specia. 2017. SemEval-2017 task 1: Semantic textual similarity multilingual and crosslingual focused evaluation. In Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017), pages 1–14, Vancouver, Canada.… find
  16. Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin de Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. 2019. Parameter-efficient transfer learning for NLP. In Proceedings of the 36th International Conference on Machine Learning, ICML 2019, volume 97 of Procee… find
  17. Masahiro Kaneko and Danushka Bollegala. 2021. Debiasing pre-trained contextualised embeddings. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pages 1256–1266, Online. Association for Computational Linguistics. find our card
  18. First Workshop on Knowledge Extraction and Integration for Deep Learning Architectures, pages 43– 49, Online. Association for Computational Linguistics. find
  19. Saket Karve, Lyle Ungar, and João Sedoc. 2019. Conceptor debiasing of word representations evaluated on WEAT. In Proceedings of the First Workshop on Gender Bias in Natural Language Processing, pages 40–48, Florence, Italy. Association for Computational Linguistics. find
  20. Dmitry Lepikhin, HyoukJoong Lee, Yuanzhong Xu, Dehao Chen, Orhan Firat, Yanping Huang, Maxim Krikun, Noam Shazeer, and Zhifeng Chen. 2020. Gshard: Scaling giant models with conditional computation and automatic sharding. arXiv preprint arXiv:2006.16668. find
  21. Yunsu Kim, Petre Petrov, Pavel Petrushkov, Shahram Khadivi, and Hermann Ney. 2019. Pivot-based transfer learning for neural machine translation between non-English languages. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Con… find
  22. Diederik P. Kingma and Jimmy Ba. 2015. Adam: A method for stochastic optimization. In 3rd International Conference on Learning Representations, ICLR 2015, Conference Track Proceedings, San Diego, CA, USA. find
  23. James Kirkpatrick, Razvan Pascanu, Neil Rabinowitz, Joel Veness, Guillaume Desjardins, Andrei A Rusu, Kieran Milan, John Quan, Tiago Ramalho, Agnieszka Grabska-Barwinska, et al. 2017. Overcoming catastrophic forgetting in neural networks. Proceedings of the national academy of sciences, 114(13):3521… find
  24. Keita Kurita, Nidhi Vyas, Ayush Pareek, Alan W Black, and Yulia Tsvetkov. 2019. Measuring bias in contextualized word representations. In Proceedings of the First Workshop on Gender Bias in Natural Language Processing, pages 166–172, Florence, Italy. Association for Computational Linguistics. find our card
  25. Anne Lauscher and Goran Glavaš. 2019. Are we consistently biased? multidimensional analysis of biases in distributional word vectors. In Proceedings of the Eighth Joint Conference on Lexical and Computational Semantics (*SEM 2019), pages 85–91, Minneapolis, Minnesota. Association for Computational L… find
  26. Anne Lauscher, Goran Glavaš, Simone Paolo Ponzetto, and Ivan Vulić. 2020a. A general framework for implicit and explicit debiasing of distributional word vector spaces. Proceedings of the AAAI Conference on Artificial Intelligence, 34(05):8131–8138. find
  27. Anne Lauscher, Olga Majewska, Leonardo F. R. Ribeiro, Iryna Gurevych, Nikolai Rozanov, and Goran Glavaš. 2020b. Common sense or world knowledge? investigating adapter-based knowledge injection into pretrained transformers. In Proceedings of Deep Learning Inside Out (DeeLIO): The Sheng Liang, Philipp… find
  28. Zhaojiang Lin, Andrea Madotto, and Pascale Fung. 2020. Exploring versatile generative language model via parameter-efficient transfer learning. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 441–459, Online. Association for Computational Linguistics. find
  29. Kaiji Lu, Piotr Mardziel, Fangjing Wu, Preetam Amancharla, and Anupam Datta. 2020. Gender bias in neural natural language processing. In Logic, Language, and Security, pages 189–202. Springer. find our card
  30. Michael McCloskey and Neal J Cohen. 1989. Catastrophic interference in connectionist networks: The sequential learning problem. In Psychology of learning and motivation, volume 24, pages 109–165. Elsevier. find
  31. Nafise Sadat Moosavi, Angela Fan, Vered Shwartz, Goran Glavaš, Shafiq Joty, Alex Wang, and Thomas Wolf, editors. 2020. Proceedings of SustaiNLP: Workshop on Simple and Efficient Natural Language Processing. Association for Computational Linguistics, Online. find
  32. Nikita Nangia, Clara Vania, Rasika Bhalerao, and Samuel R. Bowman. 2020. CrowS-pairs: A challenge dataset for measuring social biases in masked language models. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1953–1967, Online. Association for… find our card
  33. Matthew Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer. 2018. Deep contextualized word representations. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technolog… find
  34. Jonas Pfeiffer, Aishwarya Kamath, Andreas Rücklé, Kyunghyun Cho, and Iryna Gurevych. 2021. find
  35. AdapterFusion: Non-destructive task composition for transfer learning. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pages 487–503, Online. Association for Computational Linguistics. find our card
  36. Emma Strubell, Ananya Ganesh, and Andrew McCallum. 2019. Energy and policy considerations for deep learning in NLP. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 3645–3650, Florence, Italy. Association for Computational Linguistics. find our card
  37. Jonas Pfeiffer, Andreas Rücklé, Clifton Poth, Aishwarya Kamath, Ivan Vulić, Sebastian Ruder, Kyunghyun Cho, and Iryna Gurevych. 2020a. AdapterHub: A framework for adapting transformers. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations,… find our card
  38. Ahmet Üstün, Arianna Bisazza, Gosse Bouma, and Gertjan van Noord. 2020. UDapter: Language adaptation for truly Universal Dependency parsing. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 2302–2315, Online. Association for Computational Lingu… find
  39. Jonas Pfeiffer, Ivan Vulić, Iryna Gurevych, and Sebastian Ruder. 2020b. MAD-X: An Adapter-Based Framework for Multi-Task Cross-Lingual Transfer. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 7654–7673, Online. Association for Computational… find our card
  40. Minh Quang Pham, Josep Maria Crego, François Yvon, and Jean Senellart. 2020. A study of residual adapters for multi-domain neural machine translation. In Proceedings of the Fifth Conference on Machine Translation, pages 617–628, Online. Association for Computational Linguistics. find
  41. Jerin Philip, Alexandre Berard, Matthias Gallé, and Laurent Besacier. 2020. Monolingual adapters for zero-shot neural machine translation. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 4465–4470, Online. Association for Computational Linguis… find
  42. Yusu Qian, Urwa Muaz, Ben Zhang, and Jae Won Hyun. 2019. Reducing gender bias in word-level language models with a gender-equalizing loss function. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics: Student Research Workshop, pages 223–228, Florence, Italy. A… find
  43. Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. OpenAI Blog, 1(8). find
  44. Sylvestre-Alvise Rebuffi, Andrea Vedaldi, and Hakan Bilen. 2018. Efficient parametrization of multi-domain deep neural networks. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8119–8127. find
  45. Rachel Rudinger, Jason Naradowsky, Brian Leonard, and Benjamin Van Durme. 2018. Gender bias in coreference resolution. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers), pages 8–1… find our card
  46. Marko Vidoni, Ivan Vulić, and Goran Glavaš. 2020. Orthogonal language and task adapters in zero-shot cross-lingual transfer. arXiv preprint arXiv:2012.06460. — [arXiv:2012.06460](https://arxiv.org/abs/2012.06460) find
  47. Ivan Vulić, Simon Baker, Edoardo Maria Ponti, Ulla Petti, Ira Leviant, Kelly Wing, Olga Majewska, Eden Bar, Matt Malone, Thierry Poibeau, Roi Reichart, and Anna Korhonen. 2020. Multi-SimLex: A large-scale evaluation of multilingual and crosslingual lexical semantic similarity. Computational Linguis… find
  48. Tobias Walter, Celina Kirschner, Steffen Eger, Goran Glavaš, Anne Lauscher, and Simone Paolo Ponzetto. 2021. Diachronic analysis of german parliamentary proceedings: Ideological shifts through the lens of political biases. arXiv preprint arXiv:2108.06295. find
  49. Ruize Wang, Duyu Tang, Nan Duan, Zhongyu Wei, Xuanjing Huang, Cuihong Cao, Daxin Jiang, Ming Zhou, et al. 2020. K-adapter: Infusing knowledge into pre-trained models with adapters. arXiv preprint arXiv:2002.01808. find our card
  50. Kellie Webster, Xuezhi Wang, Ian Tenney, Alex Beutel, Emily Pitler, Ellie Pavlick, Jilin Chen, Ed Chi, and Slav Petrov. 2020. Measuring and reducing gendered correlations in pre-trained models. arXiv preprint arXiv:2010.06032. find our card
  51. Adina Williams, Nikita Nangia, and Samuel Bowman. 2018. A broad-coverage challenge corpus for sentence understanding through inference. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Pa… find
  52. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier-ric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lho… find our card
  53. Hao Yang, Minghan Wang, Ning Xie, Ying Qin, and Yao Deng. 2020. Efficient transfer learning for quality estimation with bottleneck adapter layer. In Proceedings of the 22nd Annual Conference of the European Association for Machine Translation, pages 29–34, Lisboa, Portugal. European Association for… find
  54. Jieyu Zhao, Tianlu Wang, Mark Yatskar, Ryan Cotterell, Vicente Ordonez, and Kai-Wei Chang. 2019. Gender bias in contextualized word embeddings. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1… find our card
  55. Jieyu Zhao, Tianlu Wang, Mark Yatskar, Vicente Ordonez, and Kai-Wei Chang. 2018. Gender bias in coreference resolution: Evaluation and debiasing methods. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,… find our card
  56. Yukun Zhu, Ryan Kiros, Rich Zemel, Ruslan Salakhutdinov, Raquel Urtasun, Antonio Torralba, and Sanja Fidler. 2015. Aligning books and movies: Towards story-like visual explanations by watching movies and reading books. In 2015 IEEE International Conference on Computer Vision (ICCV), pages 19–27. find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/lauscher-2021-sustainable-modular-debiasing-language-models/