Mitigating Gender Bias in Distilled Language Models via Counterfactual Role Reversal. Gupta, U., Dhamala, J., Kumar, V., Verma, A., Pruksachatkun, Y., Krishna, S., Gupta, R., Chang, K., Ver Steeg, G., & Galstyan, A. In Muresan, S., Nakov, P., & Villavicencio, A., editors, Findings of the Association for Computational Linguistics: ACL 2022, pages 658–678, Dublin, Ireland, May, 2022. Association for Computational Linguistics.
Mitigating Gender Bias in Distilled Language Models via Counterfactual Role Reversal [link]Paper  doi  abstract   bibtex   
Language models excel at generating coherent text, and model compression techniques such as knowledge distillation have enabled their use in resource-constrained settings. However, these models can be biased in multiple ways, including the unfounded association of male and female genders with gender-neutral professions. Therefore, knowledge distillation without any fairness constraints may preserve or exaggerate the teacher model's biases onto the distilled model. To this end, we present a novel approach to mitigate gender disparity in text generation by learning a fair model during knowledge distillation. We propose two modifications to the base knowledge distillation based on counterfactual role reversal—modifying teacher probabilities and augmenting the training set. We evaluate gender polarity across professions in open-ended text generated from the resulting distilled and finetuned GPT–2 models and demonstrate a substantial reduction in gender disparity with only a minor compromise in utility. Finally, we observe that language models that reduce gender polarity in language generation do not improve embedding fairness or downstream classification fairness.
@inproceedings{gupta-etal-2022-mitigating,
    title = "Mitigating Gender Bias in Distilled Language Models via Counterfactual Role Reversal",
    author = "Gupta, Umang  and
      Dhamala, Jwala  and
      Kumar, Varun  and
      Verma, Apurv  and
      Pruksachatkun, Yada  and
      Krishna, Satyapriya  and
      Gupta, Rahul  and
      Chang, Kai-Wei  and
      Ver Steeg, Greg  and
      Galstyan, Aram",
    editor = "Muresan, Smaranda  and
      Nakov, Preslav  and
      Villavicencio, Aline",
    booktitle = "Findings of the Association for Computational Linguistics: ACL 2022",
    month = may,
    year = "2022",
    address = "Dublin, Ireland",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.findings-acl.55",
    doi = "10.18653/v1/2022.findings-acl.55",
    pages = "658--678",
    abstract = "Language models excel at generating coherent text, and model compression techniques such as knowledge distillation have enabled their use in resource-constrained settings. However, these models can be biased in multiple ways, including the unfounded association of male and female genders with gender-neutral professions. Therefore, knowledge distillation without any fairness constraints may preserve or exaggerate the teacher model{'}s biases onto the distilled model. To this end, we present a novel approach to mitigate gender disparity in text generation by learning a fair model during knowledge distillation. We propose two modifications to the base knowledge distillation based on counterfactual role reversal{---}modifying teacher probabilities and augmenting the training set. We evaluate gender polarity across professions in open-ended text generated from the resulting distilled and finetuned GPT{--}2 models and demonstrate a substantial reduction in gender disparity with only a minor compromise in utility. Finally, we observe that language models that reduce gender polarity in language generation do not improve embedding fairness or downstream classification fairness.",
}

Downloads: 0